Merge branch 'master' into gg/flash-attn
ggml-ci
This commit is contained in:
Georgi Gerganov
commit
8937ec5307
57 changed files with 1576 additions and 8245 deletions
2
.github/workflows/bench.yml
vendored
2
.github/workflows/bench.yml
vendored
|
|
@ -32,7 +32,7 @@ on:
|
|||
- cron: '04 2 * * *'
|
||||
|
||||
concurrency:
|
||||
group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}-${{ github.event.inputs.sha }}
|
||||
group: ${{ github.workflow }}-${{ github.ref || github.run_id }}-${{ github.event.inputs.sha }}
|
||||
cancel-in-progress: true
|
||||
|
||||
jobs:
|
||||
|
|
|
|||
33
.github/workflows/build.yml
vendored
33
.github/workflows/build.yml
vendored
|
|
@ -32,6 +32,8 @@ jobs:
|
|||
- name: Clone
|
||||
id: checkout
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
|
||||
- name: Dependencies
|
||||
id: depends
|
||||
|
|
@ -88,6 +90,8 @@ jobs:
|
|||
- name: Clone
|
||||
id: checkout
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
|
||||
- name: Dependencies
|
||||
id: depends
|
||||
|
|
@ -206,6 +210,8 @@ jobs:
|
|||
- name: Clone
|
||||
id: checkout
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
fetch-depth: 0
|
||||
|
||||
- name: Dependencies
|
||||
id: depends
|
||||
|
|
@ -238,6 +244,33 @@ jobs:
|
|||
./bin/convert-llama2c-to-ggml --copy-vocab-from-model ./tok512.bin --llama2c-model stories260K.bin --llama2c-output-model stories260K.gguf
|
||||
./bin/main -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256
|
||||
|
||||
- name: Determine tag name
|
||||
id: tag
|
||||
shell: bash
|
||||
run: |
|
||||
BUILD_NUMBER="$(git rev-list --count HEAD)"
|
||||
SHORT_HASH="$(git rev-parse --short=7 HEAD)"
|
||||
if [[ "${{ env.BRANCH_NAME }}" == "master" ]]; then
|
||||
echo "name=b${BUILD_NUMBER}" >> $GITHUB_OUTPUT
|
||||
else
|
||||
SAFE_NAME=$(echo "${{ env.BRANCH_NAME }}" | tr '/' '-')
|
||||
echo "name=${SAFE_NAME}-b${BUILD_NUMBER}-${SHORT_HASH}" >> $GITHUB_OUTPUT
|
||||
fi
|
||||
|
||||
- name: Pack artifacts
|
||||
id: pack_artifacts
|
||||
if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
|
||||
run: |
|
||||
cp LICENSE ./build/bin/
|
||||
zip -r llama-${{ steps.tag.outputs.name }}-bin-ubuntu-x64.zip ./build/bin/*
|
||||
|
||||
- name: Upload artifacts
|
||||
if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
|
||||
uses: actions/upload-artifact@v4
|
||||
with:
|
||||
path: llama-${{ steps.tag.outputs.name }}-bin-ubuntu-x64.zip
|
||||
name: llama-bin-ubuntu-x64.zip
|
||||
|
||||
# ubuntu-latest-cmake-sanitizer:
|
||||
# runs-on: ubuntu-latest
|
||||
#
|
||||
|
|
|
|||
2
.github/workflows/server.yml
vendored
2
.github/workflows/server.yml
vendored
|
|
@ -23,7 +23,7 @@ on:
|
|||
- cron: '2 4 * * *'
|
||||
|
||||
concurrency:
|
||||
group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
|
||||
group: ${{ github.workflow }}-${{ github.ref || github.run_id }}
|
||||
cancel-in-progress: true
|
||||
|
||||
jobs:
|
||||
|
|
|
|||
4
.gitignore
vendored
4
.gitignore
vendored
|
|
@ -34,6 +34,7 @@ lcov-report/
|
|||
gcovr-report/
|
||||
|
||||
build*
|
||||
!build.zig
|
||||
cmake-build-*
|
||||
out/
|
||||
tmp/
|
||||
|
|
@ -100,6 +101,9 @@ qnt-*.txt
|
|||
perf-*.txt
|
||||
|
||||
examples/jeopardy/results.txt
|
||||
examples/server/*.html.hpp
|
||||
examples/server/*.js.hpp
|
||||
examples/server/*.mjs.hpp
|
||||
|
||||
poetry.lock
|
||||
poetry.toml
|
||||
|
|
|
|||
|
|
@ -43,17 +43,11 @@ else()
|
|||
set(LLAMA_METAL_DEFAULT OFF)
|
||||
endif()
|
||||
|
||||
# TODO: fix this for Android CI
|
||||
# https://github.com/ggerganov/llama.cpp/pull/6716#issuecomment-2061509191
|
||||
#if (CMAKE_SYSTEM_NAME MATCHES "ANDROID")
|
||||
# set(LLAMA_LLAMAFILE_DEFAULT OFF)
|
||||
#else()
|
||||
# set(LLAMA_LLAMAFILE_DEFAULT ON)
|
||||
#endif()
|
||||
|
||||
# TODO: temporary disable until MoE is fixed
|
||||
# https://github.com/ggerganov/llama.cpp/pull/6716
|
||||
if (CMAKE_SYSTEM_NAME MATCHES "ANDROID")
|
||||
set(LLAMA_LLAMAFILE_DEFAULT OFF)
|
||||
else()
|
||||
set(LLAMA_LLAMAFILE_DEFAULT ON)
|
||||
endif()
|
||||
|
||||
# general
|
||||
option(BUILD_SHARED_LIBS "build shared libraries" OFF)
|
||||
|
|
|
|||
17
Makefile
17
Makefile
|
|
@ -384,10 +384,6 @@ ifdef LLAMA_OPENBLAS
|
|||
MK_LDFLAGS += $(shell pkg-config --libs openblas)
|
||||
endif # LLAMA_OPENBLAS
|
||||
|
||||
# TODO: temporary disable until MoE is fixed
|
||||
# https://github.com/ggerganov/llama.cpp/pull/6716
|
||||
LLAMA_NO_LLAMAFILE := 1
|
||||
|
||||
ifndef LLAMA_NO_LLAMAFILE
|
||||
MK_CPPFLAGS += -DGGML_USE_LLAMAFILE
|
||||
OBJS += sgemm.o
|
||||
|
|
@ -699,7 +695,7 @@ OBJS += ggml-alloc.o ggml-backend.o ggml-quants.o unicode.o unicode-data.o
|
|||
llama.o: llama.cpp unicode.h ggml.h ggml-alloc.h ggml-backend.h ggml-cuda.h ggml-metal.h llama.h
|
||||
$(CXX) $(CXXFLAGS) -c $< -o $@
|
||||
|
||||
COMMON_H_DEPS = common/common.h common/sampling.h common/log.h
|
||||
COMMON_H_DEPS = common/common.h common/sampling.h common/log.h llama.h
|
||||
COMMON_DEPS = common.o sampling.o grammar-parser.o build-info.o json-schema-to-grammar.o
|
||||
|
||||
common.o: common/common.cpp $(COMMON_H_DEPS)
|
||||
|
|
@ -800,10 +796,19 @@ save-load-state: examples/save-load-state/save-load-state.cpp ggml.o llama.o $(C
|
|||
$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
|
||||
$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
|
||||
|
||||
server: examples/server/server.cpp examples/server/utils.hpp examples/server/httplib.h common/json.hpp examples/server/index.html.hpp examples/server/index.js.hpp examples/server/completion.js.hpp common/stb_image.h ggml.o llama.o $(COMMON_DEPS) grammar-parser.o $(OBJS)
|
||||
server: examples/server/server.cpp examples/server/utils.hpp examples/server/httplib.h common/json.hpp examples/server/index.html.hpp examples/server/index.js.hpp examples/server/completion.js.hpp examples/server/json-schema-to-grammar.mjs.hpp common/stb_image.h ggml.o llama.o $(COMMON_DEPS) grammar-parser.o $(OBJS)
|
||||
$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
|
||||
$(CXX) $(CXXFLAGS) $(filter-out %.h %.hpp $<,$^) -Iexamples/server $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS) $(LWINSOCK2)
|
||||
|
||||
# Portable equivalent of `cd examples/server/public && xxd -i $(notdir $<) ../$(notdir $<).hpp`:
|
||||
examples/server/%.hpp: examples/server/public/% Makefile
|
||||
@( export NAME=$(subst .,_,$(subst -,_,$(notdir $<))) && \
|
||||
echo "unsigned char $${NAME}[] = {" && \
|
||||
cat $< | od -v -t x1 -An | sed -E 's/([0-9a-fA-F]+)/0x\1, /g' && \
|
||||
echo "};" && \
|
||||
echo "unsigned int $${NAME}_len = $(shell cat $< | wc -c );" \
|
||||
) > $@
|
||||
|
||||
gguf: examples/gguf/gguf.cpp ggml.o $(OBJS)
|
||||
$(CXX) $(CXXFLAGS) -c $< -o $(call GET_OBJ_FILE, $<)
|
||||
$(CXX) $(CXXFLAGS) $(filter-out %.h $<,$^) $(call GET_OBJ_FILE, $<) -o $@ $(LDFLAGS)
|
||||
|
|
|
|||
|
|
@ -229,12 +229,12 @@ source /opt/intel/oneapi/setvars.sh
|
|||
# Build LLAMA with MKL BLAS acceleration for intel GPU
|
||||
mkdir -p build && cd build
|
||||
|
||||
# Option 1: Use FP16 for better performance in long-prompt inference
|
||||
#cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
|
||||
|
||||
# Option 2: Use FP32 by default
|
||||
# Option 1: Use FP32 (recommended for better performance in most cases)
|
||||
cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
|
||||
|
||||
# Option 2: Use FP16
|
||||
cmake .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
|
||||
|
||||
#build all binary
|
||||
cmake --build . --config Release -j -v
|
||||
```
|
||||
|
|
@ -250,12 +250,12 @@ export CPLUS_INCLUDE_DIR=/path/to/oneMKL/include:$CPLUS_INCLUDE_DIR
|
|||
# Build LLAMA with Nvidia BLAS acceleration through SYCL
|
||||
mkdir -p build && cd build
|
||||
|
||||
# Option 1: Use FP16 for better performance in long-prompt inference
|
||||
cmake .. -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
|
||||
|
||||
# Option 2: Use FP32 by default
|
||||
# Option 1: Use FP32 (recommended for better performance in most cases)
|
||||
cmake .. -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
|
||||
|
||||
# Option 2: Use FP16
|
||||
cmake .. -DLLAMA_SYCL=ON -DLLAMA_SYCL_TARGET=NVIDIA -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
|
||||
|
||||
#build all binary
|
||||
cmake --build . --config Release -j -v
|
||||
|
||||
|
|
@ -416,6 +416,10 @@ mkdir -p build
|
|||
cd build
|
||||
@call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat" intel64 --force
|
||||
|
||||
# Option 1: Use FP32 (recommended for better performance in most cases)
|
||||
cmake -G "MinGW Makefiles" .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icx -DCMAKE_BUILD_TYPE=Release
|
||||
|
||||
# Option 2: Or FP16
|
||||
cmake -G "MinGW Makefiles" .. -DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icx -DCMAKE_BUILD_TYPE=Release -DLLAMA_SYCL_F16=ON
|
||||
|
||||
make -j
|
||||
|
|
|
|||
11
README.md
11
README.md
|
|
@ -10,6 +10,7 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others)
|
|||
|
||||
### Recent API changes
|
||||
|
||||
- [2024 Apr 21] `llama_token_to_piece` can now optionally render special tokens https://github.com/ggerganov/llama.cpp/pull/6807
|
||||
- [2024 Apr 4] State and session file functions reorganized under `llama_state_*` https://github.com/ggerganov/llama.cpp/pull/6341
|
||||
- [2024 Mar 26] Logits and embeddings API updated for compactness https://github.com/ggerganov/llama.cpp/pull/6122
|
||||
- [2024 Mar 13] Add `llama_synchronize()` + `llama_context_params.n_ubatch` https://github.com/ggerganov/llama.cpp/pull/6017
|
||||
|
|
@ -95,7 +96,7 @@ Typically finetunes of the base models below are supported as well.
|
|||
- [X] [Mistral 7B](https://huggingface.co/mistralai/Mistral-7B-v0.1)
|
||||
- [x] [Mixtral MoE](https://huggingface.co/models?search=mistral-ai/Mixtral)
|
||||
- [x] [DBRX](https://huggingface.co/databricks/dbrx-instruct)
|
||||
- [X] Falcon
|
||||
- [X] [Falcon](https://huggingface.co/models?search=tiiuae/falcon)
|
||||
- [X] [Chinese LLaMA / Alpaca](https://github.com/ymcui/Chinese-LLaMA-Alpaca) and [Chinese LLaMA-2 / Alpaca-2](https://github.com/ymcui/Chinese-LLaMA-Alpaca-2)
|
||||
- [X] [Vigogne (French)](https://github.com/bofenghuang/vigogne)
|
||||
- [X] [Koala](https://bair.berkeley.edu/blog/2023/04/03/koala/)
|
||||
|
|
@ -549,7 +550,7 @@ Building the program with BLAS support may lead to some performance improvements
|
|||
OpenCL acceleration is provided by the matrix multiplication kernels from the [CLBlast](https://github.com/CNugteren/CLBlast) project and custom kernels for ggml that can generate tokens on the GPU.
|
||||
|
||||
You will need the [OpenCL SDK](https://github.com/KhronosGroup/OpenCL-SDK).
|
||||
- For Ubuntu or Debian, the packages `opencl-headers`, `ocl-icd` may be needed.
|
||||
- For Ubuntu, Debian, and Fedora the packages `opencl-headers`, `ocl-icd` may be needed.
|
||||
|
||||
- For Windows, a pre-built SDK is available on the [OpenCL Releases](https://github.com/KhronosGroup/OpenCL-SDK/releases) page.
|
||||
|
||||
|
|
@ -574,6 +575,12 @@ Building the program with BLAS support may lead to some performance improvements
|
|||
|
||||
Pre-built CLBlast binaries may be found on the [CLBlast Releases](https://github.com/CNugteren/CLBlast/releases) page. For Unix variants, it may also be found in your operating system's packages.
|
||||
|
||||
Linux packaging:
|
||||
Fedora Linux:
|
||||
```bash
|
||||
sudo dnf install clblast
|
||||
```
|
||||
|
||||
Alternatively, they may be built from source.
|
||||
|
||||
- <details>
|
||||
|
|
|
|||
29
build.zig
29
build.zig
|
|
@ -140,4 +140,33 @@ pub fn build(b: *std.build.Builder) !void {
|
|||
if (server.target.isWindows()) {
|
||||
server.linkSystemLibrary("ws2_32");
|
||||
}
|
||||
|
||||
const server_assets = [_][]const u8{ "index.html", "index.js", "completion.js", "json-schema-to-grammar.mjs" };
|
||||
for (server_assets) |asset| {
|
||||
const input_path = b.fmt("examples/server/public/{s}", .{asset});
|
||||
const output_path = b.fmt("examples/server/{s}.hpp", .{asset});
|
||||
|
||||
// Portable equivalent of `b.addSystemCommand(&.{ "xxd", "-n", asset, "-i", input_path, output_path }) })`:
|
||||
|
||||
const input = try std.fs.cwd().readFileAlloc(b.allocator, input_path, std.math.maxInt(usize));
|
||||
defer b.allocator.free(input);
|
||||
|
||||
var buf = std.ArrayList(u8).init(b.allocator);
|
||||
defer buf.deinit();
|
||||
|
||||
for (input) |byte| {
|
||||
try std.fmt.format(buf.writer(), "0x{X:0>2}, ", .{byte});
|
||||
}
|
||||
|
||||
var name = try std.mem.replaceOwned(u8, b.allocator, asset, "-", "_");
|
||||
defer b.allocator.free(name);
|
||||
std.mem.replaceScalar(u8, name, '.', '_');
|
||||
|
||||
try std.fs.cwd().writeFile(output_path, b.fmt(
|
||||
"unsigned char {s}[] = {{{s}}};\nunsigned int {s}_len = {d};\n",
|
||||
.{ name, buf.items, name, input.len },
|
||||
));
|
||||
|
||||
std.debug.print("Dumped hex of \"{s}\" ({s}) to {s}\n", .{ input_path, name, output_path });
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -108,7 +108,7 @@ int32_t get_num_physical_cores() {
|
|||
return n_threads > 0 ? (n_threads <= 4 ? n_threads : n_threads / 2) : 4;
|
||||
}
|
||||
|
||||
#if defined(__x86_64__) && defined(__linux__)
|
||||
#if defined(__x86_64__) && defined(__linux__) && !defined(__ANDROID__)
|
||||
#include <pthread.h>
|
||||
|
||||
static void cpuid(unsigned leaf, unsigned subleaf,
|
||||
|
|
@ -162,7 +162,7 @@ static int count_math_cpus(int cpu_count) {
|
|||
* Returns number of CPUs on system that are useful for math.
|
||||
*/
|
||||
int get_math_cpu_count() {
|
||||
#if defined(__x86_64__) && defined(__linux__)
|
||||
#if defined(__x86_64__) && defined(__linux__) && !defined(__ANDROID__)
|
||||
int cpu_count = sysconf(_SC_NPROCESSORS_ONLN);
|
||||
if (cpu_count < 1) {
|
||||
return get_num_physical_cores();
|
||||
|
|
@ -242,7 +242,9 @@ bool gpt_params_find_arg(int argc, char ** argv, const std::string & arg, gpt_pa
|
|||
invalid_param = true;
|
||||
return true;
|
||||
}
|
||||
// This is temporary, in the future the samplign state will be moved fully to llama_sampling_context.
|
||||
params.seed = std::stoul(argv[i]);
|
||||
sparams.seed = std::stoul(argv[i]);
|
||||
return true;
|
||||
}
|
||||
if (arg == "-t" || arg == "--threads") {
|
||||
|
|
@ -2332,12 +2334,12 @@ std::vector<llama_token> llama_tokenize(
|
|||
return result;
|
||||
}
|
||||
|
||||
std::string llama_token_to_piece(const struct llama_context * ctx, llama_token token) {
|
||||
std::string llama_token_to_piece(const struct llama_context * ctx, llama_token token, bool special) {
|
||||
std::vector<char> result(8, 0);
|
||||
const int n_tokens = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size());
|
||||
const int n_tokens = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size(), special);
|
||||
if (n_tokens < 0) {
|
||||
result.resize(-n_tokens);
|
||||
int check = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size());
|
||||
int check = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size(), special);
|
||||
GGML_ASSERT(check == -n_tokens);
|
||||
} else {
|
||||
result.resize(n_tokens);
|
||||
|
|
|
|||
|
|
@ -238,11 +238,12 @@ std::vector<llama_token> llama_tokenize(
|
|||
bool add_special,
|
||||
bool parse_special = false);
|
||||
|
||||
// tokenizes a token into a piece
|
||||
// tokenizes a token into a piece, optionally renders special/control tokens
|
||||
// should work similar to Python's `tokenizer.id_to_piece`
|
||||
std::string llama_token_to_piece(
|
||||
const struct llama_context * ctx,
|
||||
llama_token token);
|
||||
llama_token token,
|
||||
bool special = true);
|
||||
|
||||
// TODO: these should be moved in llama.h C-style API under single `llama_detokenize` function
|
||||
// that takes into account the tokenizer type and decides how to handle the leading space
|
||||
|
|
|
|||
|
|
@ -1,4 +1,6 @@
|
|||
#define LLAMA_API_INTERNAL
|
||||
#include "sampling.h"
|
||||
#include <random>
|
||||
|
||||
struct llama_sampling_context * llama_sampling_init(const struct llama_sampling_params & params) {
|
||||
struct llama_sampling_context * result = new llama_sampling_context();
|
||||
|
|
@ -33,6 +35,8 @@ struct llama_sampling_context * llama_sampling_init(const struct llama_sampling_
|
|||
|
||||
result->prev.resize(params.n_prev);
|
||||
|
||||
llama_sampling_set_rng_seed(result, params.seed);
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
|
|
@ -62,6 +66,13 @@ void llama_sampling_reset(llama_sampling_context * ctx) {
|
|||
ctx->cur.clear();
|
||||
}
|
||||
|
||||
void llama_sampling_set_rng_seed(struct llama_sampling_context * ctx, uint32_t seed) {
|
||||
if (seed == LLAMA_DEFAULT_SEED) {
|
||||
seed = time(NULL);
|
||||
}
|
||||
ctx->rng.seed(seed);
|
||||
}
|
||||
|
||||
void llama_sampling_cp(llama_sampling_context * src, llama_sampling_context * dst) {
|
||||
if (dst->grammar) {
|
||||
llama_grammar_free(dst->grammar);
|
||||
|
|
@ -203,7 +214,7 @@ static llama_token llama_sampling_sample_impl(
|
|||
|
||||
sampler_queue(ctx_main, params, cur_p, min_keep);
|
||||
|
||||
id = llama_sample_token(ctx_main, &cur_p);
|
||||
id = llama_sample_token_with_rng(ctx_main, &cur_p, ctx_sampling->rng);
|
||||
|
||||
//{
|
||||
// const int n_top = 10;
|
||||
|
|
|
|||
|
|
@ -4,9 +4,10 @@
|
|||
|
||||
#include "grammar-parser.h"
|
||||
|
||||
#include <random>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
#include <unordered_map>
|
||||
#include <vector>
|
||||
|
||||
// sampler types
|
||||
enum class llama_sampler_type : char {
|
||||
|
|
@ -39,6 +40,7 @@ typedef struct llama_sampling_params {
|
|||
float mirostat_tau = 5.00f; // target entropy
|
||||
float mirostat_eta = 0.10f; // learning rate
|
||||
bool penalize_nl = false; // consider newlines as a repeatable token
|
||||
uint32_t seed = LLAMA_DEFAULT_SEED; // the seed used to initialize llama_sampling_context
|
||||
|
||||
std::vector<llama_sampler_type> samplers_sequence = {
|
||||
llama_sampler_type::TOP_K,
|
||||
|
|
@ -79,6 +81,8 @@ struct llama_sampling_context {
|
|||
// TODO: replace with ring-buffer
|
||||
std::vector<llama_token> prev;
|
||||
std::vector<llama_token_data> cur;
|
||||
|
||||
std::mt19937 rng;
|
||||
};
|
||||
|
||||
#include "common.h"
|
||||
|
|
@ -93,6 +97,9 @@ void llama_sampling_free(struct llama_sampling_context * ctx);
|
|||
// - reset grammar
|
||||
void llama_sampling_reset(llama_sampling_context * ctx);
|
||||
|
||||
// Set the sampler seed
|
||||
void llama_sampling_set_rng_seed(struct llama_sampling_context * ctx, uint32_t seed);
|
||||
|
||||
// Copy the sampler context
|
||||
void llama_sampling_cp(llama_sampling_context * src, llama_sampling_context * dst);
|
||||
|
||||
|
|
|
|||
|
|
@ -363,6 +363,16 @@ class Model(ABC):
|
|||
scores.append(-1000.0)
|
||||
toktypes.append(SentencePieceTokenTypes.USER_DEFINED)
|
||||
|
||||
if vocab_size > len(tokens):
|
||||
pad_count = vocab_size - len(tokens)
|
||||
print(
|
||||
f"Padding vocab with {pad_count} token(s) - [PAD1] through [PAD{pad_count}]"
|
||||
)
|
||||
for i in range(1, pad_count + 1):
|
||||
tokens.append(f"[PAD{i}]")
|
||||
scores.append(-1000.0)
|
||||
toktypes.append(SentencePieceTokenTypes.UNUSED)
|
||||
|
||||
assert len(tokens) == vocab_size
|
||||
|
||||
self.gguf_writer.add_tokenizer_model("llama")
|
||||
|
|
@ -1301,10 +1311,18 @@ class LlamaModel(Model):
|
|||
try:
|
||||
self. _set_vocab_sentencepiece()
|
||||
except FileNotFoundError:
|
||||
try:
|
||||
self._set_vocab_llama_hf()
|
||||
except (FileNotFoundError, TypeError):
|
||||
# Llama 3
|
||||
self._set_vocab_gpt2()
|
||||
|
||||
special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=False,
|
||||
special_token_types = ['prefix', 'suffix', 'middle', 'eot'])
|
||||
# Apply to CodeLlama only (and ignore for Llama 3 with a vocab size of 128256)
|
||||
if self.hparams.get("vocab_size", 32000) == 32016:
|
||||
special_vocab = gguf.SpecialVocab(
|
||||
self.dir_model, load_merges=False,
|
||||
special_token_types = ['prefix', 'suffix', 'middle', 'eot']
|
||||
)
|
||||
special_vocab._set_special_token("prefix", 32007)
|
||||
special_vocab._set_special_token("suffix", 32008)
|
||||
special_vocab._set_special_token("middle", 32009)
|
||||
|
|
@ -1781,6 +1799,12 @@ class QwenModel(Model):
|
|||
class Qwen2Model(Model):
|
||||
model_arch = gguf.MODEL_ARCH.QWEN2
|
||||
|
||||
def set_vocab(self):
|
||||
try:
|
||||
self._set_vocab_sentencepiece()
|
||||
except FileNotFoundError:
|
||||
self._set_vocab_gpt2()
|
||||
|
||||
|
||||
@Model.register("Qwen2MoeForCausalLM")
|
||||
class Qwen2MoeModel(Model):
|
||||
|
|
@ -1971,6 +1995,91 @@ class Phi2Model(Model):
|
|||
self.gguf_writer.add_add_bos_token(False)
|
||||
|
||||
|
||||
@Model.register("Phi3ForCausalLM")
|
||||
class Phi3MiniModel(Model):
|
||||
model_arch = gguf.MODEL_ARCH.PHI3
|
||||
|
||||
def set_vocab(self):
|
||||
from sentencepiece import SentencePieceProcessor
|
||||
|
||||
tokenizer_path = self.dir_model / 'tokenizer.model'
|
||||
|
||||
if not tokenizer_path.is_file():
|
||||
print(f'Error: Missing {tokenizer_path}', file=sys.stderr)
|
||||
sys.exit(1)
|
||||
|
||||
tokenizer = SentencePieceProcessor(str(tokenizer_path))
|
||||
|
||||
vocab_size = self.hparams.get('vocab_size', tokenizer.vocab_size())
|
||||
|
||||
tokens: list[bytes] = [f"[PAD{i}]".encode("utf-8") for i in range(vocab_size)]
|
||||
scores: list[float] = [-10000.0] * vocab_size
|
||||
toktypes: list[int] = [SentencePieceTokenTypes.UNKNOWN] * vocab_size
|
||||
|
||||
for token_id in range(tokenizer.vocab_size()):
|
||||
|
||||
piece = tokenizer.id_to_piece(token_id)
|
||||
text = piece.encode("utf-8")
|
||||
score = tokenizer.get_score(token_id)
|
||||
|
||||
toktype = SentencePieceTokenTypes.NORMAL
|
||||
if tokenizer.is_unknown(token_id):
|
||||
toktype = SentencePieceTokenTypes.UNKNOWN
|
||||
elif tokenizer.is_control(token_id):
|
||||
toktype = SentencePieceTokenTypes.CONTROL
|
||||
elif tokenizer.is_unused(token_id):
|
||||
toktype = SentencePieceTokenTypes.UNUSED
|
||||
elif tokenizer.is_byte(token_id):
|
||||
toktype = SentencePieceTokenTypes.BYTE
|
||||
|
||||
tokens[token_id] = text
|
||||
scores[token_id] = score
|
||||
toktypes[token_id] = toktype
|
||||
|
||||
added_tokens_file = self.dir_model / 'added_tokens.json'
|
||||
if added_tokens_file.is_file():
|
||||
with open(added_tokens_file, "r", encoding="utf-8") as f:
|
||||
added_tokens_json = json.load(f)
|
||||
|
||||
for key in added_tokens_json:
|
||||
token_id = added_tokens_json[key]
|
||||
if (token_id >= vocab_size):
|
||||
print(f'ignore token {token_id}: id is out of range, max={vocab_size - 1}')
|
||||
continue
|
||||
|
||||
tokens[token_id] = key.encode("utf-8")
|
||||
scores[token_id] = -1000.0
|
||||
toktypes[token_id] = SentencePieceTokenTypes.USER_DEFINED
|
||||
|
||||
self.gguf_writer.add_tokenizer_model("llama")
|
||||
self.gguf_writer.add_token_list(tokens)
|
||||
self.gguf_writer.add_token_scores(scores)
|
||||
self.gguf_writer.add_token_types(toktypes)
|
||||
|
||||
special_vocab = gguf.SpecialVocab(self.dir_model, n_vocab=len(tokens))
|
||||
special_vocab.add_to_gguf(self.gguf_writer)
|
||||
|
||||
def set_gguf_parameters(self):
|
||||
block_count = self.find_hparam(["num_hidden_layers", "n_layer"])
|
||||
|
||||
rot_pct = 1.0
|
||||
n_embd = self.find_hparam(["hidden_size", "n_embd"])
|
||||
n_head = self.find_hparam(["num_attention_heads", "n_head"])
|
||||
rms_eps = self.find_hparam(["rms_norm_eps"])
|
||||
|
||||
self.gguf_writer.add_name("Phi3")
|
||||
self.gguf_writer.add_context_length(self.find_hparam(["n_positions", "max_position_embeddings"]))
|
||||
|
||||
self.gguf_writer.add_embedding_length(n_embd)
|
||||
self.gguf_writer.add_feed_forward_length(8192)
|
||||
self.gguf_writer.add_block_count(block_count)
|
||||
self.gguf_writer.add_head_count(n_head)
|
||||
self.gguf_writer.add_head_count_kv(n_head)
|
||||
self.gguf_writer.add_layer_norm_rms_eps(rms_eps)
|
||||
self.gguf_writer.add_rope_dimension_count(int(rot_pct * n_embd) // n_head)
|
||||
self.gguf_writer.add_file_type(self.ftype)
|
||||
|
||||
|
||||
@Model.register("PlamoForCausalLM")
|
||||
class PlamoModel(Model):
|
||||
model_arch = gguf.MODEL_ARCH.PLAMO
|
||||
|
|
@ -2194,6 +2303,8 @@ class InternLM2Model(Model):
|
|||
old_eos = special_vocab.special_token_ids["eos"]
|
||||
if "chat" in os.path.basename(self.dir_model.absolute()):
|
||||
# For the chat model, we replace the eos with '<|im_end|>'.
|
||||
# TODO: this is a hack, should be fixed
|
||||
# https://github.com/ggerganov/llama.cpp/pull/6745#issuecomment-2067687048
|
||||
special_vocab.special_token_ids["eos"] = self._try_get_sft_eos(tokenizer)
|
||||
print(f"Replace eos:{old_eos} with a special token:{special_vocab.special_token_ids['eos']} \
|
||||
in chat mode so that the conversation can end normally.")
|
||||
|
|
@ -2429,12 +2540,15 @@ class GemmaModel(Model):
|
|||
|
||||
def set_vocab(self):
|
||||
self._set_vocab_sentencepiece()
|
||||
|
||||
# TODO: these special tokens should be exported only for the CodeGemma family
|
||||
special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=False,
|
||||
special_token_types = ['prefix', 'suffix', 'middle', 'eot'])
|
||||
special_token_types = ['prefix', 'suffix', 'middle', 'fsep', 'eot'])
|
||||
special_vocab._set_special_token("prefix", 67)
|
||||
special_vocab._set_special_token("suffix", 69)
|
||||
special_vocab._set_special_token("middle", 68)
|
||||
special_vocab._set_special_token("eot", 70)
|
||||
special_vocab._set_special_token("fsep", 70)
|
||||
special_vocab._set_special_token("eot", 107)
|
||||
special_vocab.add_to_gguf(self.gguf_writer)
|
||||
|
||||
def set_gguf_parameters(self):
|
||||
|
|
@ -2523,16 +2637,22 @@ class MambaModel(Model):
|
|||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.MODEL)
|
||||
self.gguf_writer.add_tokenizer_model(bytes(field.parts[-1]))
|
||||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.LIST)
|
||||
self.gguf_writer.add_token_list([bytes(field.parts[i]) for i in field.data][:vocab_size])
|
||||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.TOKEN_TYPE)
|
||||
self.gguf_writer.add_token_types([field.parts[i].tolist()[0] for i in field.data][:vocab_size])
|
||||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.MERGES)
|
||||
self.gguf_writer.add_token_merges([bytes(field.parts[i]) for i in field.data])
|
||||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.BOS_ID)
|
||||
self.gguf_writer.add_bos_token_id(field.parts[-1].tolist()[0])
|
||||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.EOS_ID)
|
||||
self.gguf_writer.add_eos_token_id(field.parts[-1].tolist()[0])
|
||||
|
||||
field = neox_reader.get_field(gguf.Keys.Tokenizer.UNK_ID)
|
||||
self.gguf_writer.add_unk_token_id(field.parts[-1].tolist()[0])
|
||||
|
||||
|
|
|
|||
|
|
@ -525,7 +525,14 @@ class LlamaHfVocab(Vocab):
|
|||
|
||||
# pre-check so we know if we need transformers
|
||||
tokenizer_model: dict[str, Any] = tokenizer_json['model']
|
||||
if (
|
||||
is_llama3 = (
|
||||
tokenizer_model['type'] == 'BPE' and tokenizer_model.get('ignore_merges', False)
|
||||
and not tokenizer_model.get('byte_fallback', True)
|
||||
)
|
||||
if is_llama3:
|
||||
raise TypeError('Llama 3 must be converted with BpeVocab')
|
||||
|
||||
if not is_llama3 and (
|
||||
tokenizer_model['type'] != 'BPE' or not tokenizer_model.get('byte_fallback', False)
|
||||
or tokenizer_json['decoder']['type'] != 'Sequence'
|
||||
):
|
||||
|
|
|
|||
|
|
@ -153,7 +153,7 @@ while n_cur <= n_len {
|
|||
// const llama_token new_token_id = llama_sample_token_greedy(ctx, &candidates_p);
|
||||
|
||||
// is it an end of stream? -> mark the stream as finished
|
||||
if new_token_id == llama_token_eos(model) || n_cur == n_len {
|
||||
if llama_token_is_eog(model, new_token_id) || n_cur == n_len {
|
||||
i_batch[i] = -1
|
||||
// print("")
|
||||
if n_parallel > 1 {
|
||||
|
|
@ -229,7 +229,7 @@ private func tokenize(text: String, add_bos: Bool) -> [llama_token] {
|
|||
|
||||
private func token_to_piece(token: llama_token, buffer: inout [CChar]) -> String? {
|
||||
var result = [CChar](repeating: 0, count: 8)
|
||||
let nTokens = llama_token_to_piece(model, token, &result, Int32(result.count))
|
||||
let nTokens = llama_token_to_piece(model, token, &result, Int32(result.count), false)
|
||||
if nTokens < 0 {
|
||||
let actualTokensCount = -Int(nTokens)
|
||||
result = .init(repeating: 0, count: actualTokensCount)
|
||||
|
|
@ -237,7 +237,8 @@ private func token_to_piece(token: llama_token, buffer: inout [CChar]) -> String
|
|||
model,
|
||||
token,
|
||||
&result,
|
||||
Int32(result.count)
|
||||
Int32(result.count),
|
||||
false
|
||||
)
|
||||
assert(check == actualTokensCount)
|
||||
} else {
|
||||
|
|
|
|||
|
|
@ -191,8 +191,8 @@ int main(int argc, char ** argv) {
|
|||
|
||||
//const llama_token new_token_id = llama_sample_token_greedy(ctx, &candidates_p);
|
||||
|
||||
// is it an end of stream? -> mark the stream as finished
|
||||
if (new_token_id == llama_token_eos(model) || n_cur == n_len) {
|
||||
// is it an end of generation? -> mark the stream as finished
|
||||
if (llama_token_is_eog(model, new_token_id) || n_cur == n_len) {
|
||||
i_batch[i] = -1;
|
||||
LOG_TEE("\n");
|
||||
if (n_parallel > 1) {
|
||||
|
|
|
|||
|
|
@ -47,7 +47,7 @@ struct beam_search_callback_data {
|
|||
// In this case, end-of-beam (eob) is equivalent to end-of-sentence (eos) but this need not always be the same.
|
||||
// For example, eob can be flagged due to maximum token length, stop words, etc.
|
||||
static bool is_at_eob(const beam_search_callback_data & callback_data, const llama_token * tokens, size_t n_tokens) {
|
||||
return n_tokens && tokens[n_tokens-1] == llama_token_eos(llama_get_model(callback_data.ctx));
|
||||
return n_tokens && llama_token_is_eog(llama_get_model(callback_data.ctx), tokens[n_tokens-1]);
|
||||
}
|
||||
|
||||
// Function matching type llama_beam_search_callback_fn_t.
|
||||
|
|
|
|||
|
|
@ -651,8 +651,8 @@ int main(int argc, char ** argv) {
|
|||
// LOG_TEE("took new input\n");
|
||||
is_interacting = false;
|
||||
}
|
||||
// deal with end of text token in interactive mode
|
||||
else if (llama_sampling_last(ctx_sampling) == llama_token_eos(model)) {
|
||||
// deal with end of generation tokens in interactive mode
|
||||
else if (llama_token_is_eog(model, llama_sampling_last(ctx_sampling))) {
|
||||
LOG("found EOS token\n");
|
||||
|
||||
if (params.interactive) {
|
||||
|
|
@ -731,8 +731,8 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
}
|
||||
|
||||
// end of text token
|
||||
if (!embd.empty() && embd.back() == llama_token_eos(model) && !params.interactive) {
|
||||
// end of generation
|
||||
if (!embd.empty() && llama_token_is_eog(model, embd.back()) && !params.interactive) {
|
||||
break;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -408,7 +408,7 @@ Java_com_example_llama_Llm_completion_1loop(
|
|||
const auto new_token_id = llama_sample_token_greedy(context, &candidates_p);
|
||||
|
||||
const auto n_cur = env->CallIntMethod(intvar_ncur, la_int_var_value);
|
||||
if (new_token_id == llama_token_eos(model) || n_cur == n_len) {
|
||||
if (llama_token_is_eog(model, new_token_id) || n_cur == n_len) {
|
||||
return env->NewStringUTF("");
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -158,7 +158,7 @@ actor LlamaContext {
|
|||
new_token_id = llama_sample_token_greedy(context, &candidates_p)
|
||||
}
|
||||
|
||||
if new_token_id == llama_token_eos(model) || n_cur == n_len {
|
||||
if llama_token_is_eog(model, new_token_id) || n_cur == n_len {
|
||||
print("\n")
|
||||
let new_token_str = String(cString: temporary_invalid_cchars + [0])
|
||||
temporary_invalid_cchars.removeAll()
|
||||
|
|
@ -322,7 +322,7 @@ actor LlamaContext {
|
|||
defer {
|
||||
result.deallocate()
|
||||
}
|
||||
let nTokens = llama_token_to_piece(model, token, result, 8)
|
||||
let nTokens = llama_token_to_piece(model, token, result, 8, false)
|
||||
|
||||
if nTokens < 0 {
|
||||
let newResult = UnsafeMutablePointer<Int8>.allocate(capacity: Int(-nTokens))
|
||||
|
|
@ -330,7 +330,7 @@ actor LlamaContext {
|
|||
defer {
|
||||
newResult.deallocate()
|
||||
}
|
||||
let nNewTokens = llama_token_to_piece(model, token, newResult, -nTokens)
|
||||
let nNewTokens = llama_token_to_piece(model, token, newResult, -nTokens, false)
|
||||
let bufferPointer = UnsafeBufferPointer(start: newResult, count: Int(nNewTokens))
|
||||
return Array(bufferPointer)
|
||||
} else {
|
||||
|
|
|
|||
|
|
@ -3,6 +3,7 @@
|
|||
// I'll gradually clean and extend it
|
||||
// Note: Even when using identical normalized image inputs (see normalize_image_u8_to_f32()) we have a significant difference in resulting embeddings compared to pytorch
|
||||
#include "clip.h"
|
||||
#include "log.h"
|
||||
#include "ggml.h"
|
||||
#include "ggml-alloc.h"
|
||||
#include "ggml-backend.h"
|
||||
|
|
@ -23,7 +24,6 @@
|
|||
#include <cstdlib>
|
||||
#include <cstring>
|
||||
#include <fstream>
|
||||
#include <iostream>
|
||||
#include <map>
|
||||
#include <regex>
|
||||
#include <stdexcept>
|
||||
|
|
@ -145,7 +145,7 @@ static std::map<projector_type, std::string> PROJECTOR_TYPE_NAMES = {
|
|||
static int get_key_idx(const gguf_context * ctx, const char * key) {
|
||||
int i = gguf_find_key(ctx, key);
|
||||
if (i == -1) {
|
||||
fprintf(stderr, "key %s not found in file\n", key);
|
||||
LOG_TEE("key %s not found in file\n", key);
|
||||
throw std::runtime_error(format("Missing required key: %s", key));
|
||||
}
|
||||
|
||||
|
|
@ -247,7 +247,7 @@ static std::string gguf_kv_to_str(const struct gguf_context * ctx_gguf, int i) {
|
|||
|
||||
static void print_tensor_info(const ggml_tensor * tensor, const char * prefix = "") {
|
||||
size_t tensor_size = ggml_nbytes(tensor);
|
||||
printf("%s: n_dims = %d, name = %s, tensor_size=%zu, shape:[%" PRId64 ", %" PRId64 ", %" PRId64 ", %" PRId64 "], type = %s\n",
|
||||
LOG_TEE("%s: n_dims = %d, name = %s, tensor_size=%zu, shape:[%" PRId64 ", %" PRId64 ", %" PRId64 ", %" PRId64 "], type = %s\n",
|
||||
prefix, ggml_n_dims(tensor), tensor->name, tensor_size,
|
||||
tensor->ne[0], tensor->ne[1], tensor->ne[2], tensor->ne[3], ggml_type_name(tensor->type));
|
||||
}
|
||||
|
|
@ -265,7 +265,7 @@ static projector_type clip_projector_type_from_string(const std::string & name)
|
|||
static void clip_image_write_image_to_ppm(const clip_image_u8& img, const std::string& filename) {
|
||||
std::ofstream file(filename, std::ios::binary);
|
||||
if (!file.is_open()) {
|
||||
std::cerr << "Failed to open file for writing: " << filename << std::endl;
|
||||
LOG_TEE("Failed to open file for writing: %s\n", filename.c_str());
|
||||
return;
|
||||
}
|
||||
|
||||
|
|
@ -284,7 +284,7 @@ static void clip_image_write_image_to_ppm(const clip_image_u8& img, const std::s
|
|||
static void clip_image_save_to_bmp(const clip_image_u8& img, const std::string& filename) {
|
||||
std::ofstream file(filename, std::ios::binary);
|
||||
if (!file.is_open()) {
|
||||
std::cerr << "Failed to open file for writing: " << filename << std::endl;
|
||||
LOG_TEE("Failed to open file for writing: %s\n", filename.c_str());
|
||||
return;
|
||||
}
|
||||
|
||||
|
|
@ -515,7 +515,7 @@ struct clip_ctx {
|
|||
|
||||
static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32_batch * imgs) {
|
||||
if (!ctx->has_vision_encoder) {
|
||||
printf("This gguf file seems to have no vision encoder\n");
|
||||
LOG_TEE("This gguf file seems to have no vision encoder\n");
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
|
|
@ -879,21 +879,21 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
const int idx_name = gguf_find_key(ctx, KEY_NAME);
|
||||
if (idx_name != -1) { // make name optional temporarily as some of the uploaded models missing it due to a bug
|
||||
const std::string name = gguf_get_val_str(ctx, idx_name);
|
||||
printf("%s: model name: %s\n", __func__, name.c_str());
|
||||
LOG_TEE("%s: model name: %s\n", __func__, name.c_str());
|
||||
}
|
||||
printf("%s: description: %s\n", __func__, description.c_str());
|
||||
printf("%s: GGUF version: %d\n", __func__, gguf_get_version(ctx));
|
||||
printf("%s: alignment: %zu\n", __func__, gguf_get_alignment(ctx));
|
||||
printf("%s: n_tensors: %d\n", __func__, n_tensors);
|
||||
printf("%s: n_kv: %d\n", __func__, n_kv);
|
||||
printf("%s: ftype: %s\n", __func__, ftype_str.c_str());
|
||||
printf("\n");
|
||||
LOG_TEE("%s: description: %s\n", __func__, description.c_str());
|
||||
LOG_TEE("%s: GGUF version: %d\n", __func__, gguf_get_version(ctx));
|
||||
LOG_TEE("%s: alignment: %zu\n", __func__, gguf_get_alignment(ctx));
|
||||
LOG_TEE("%s: n_tensors: %d\n", __func__, n_tensors);
|
||||
LOG_TEE("%s: n_kv: %d\n", __func__, n_kv);
|
||||
LOG_TEE("%s: ftype: %s\n", __func__, ftype_str.c_str());
|
||||
LOG_TEE("\n");
|
||||
}
|
||||
const int n_tensors = gguf_get_n_tensors(ctx);
|
||||
|
||||
// kv
|
||||
const int n_kv = gguf_get_n_kv(ctx);
|
||||
printf("%s: loaded meta data with %d key-value pairs and %d tensors from %s\n",
|
||||
LOG_TEE("%s: loaded meta data with %d key-value pairs and %d tensors from %s\n",
|
||||
__func__, n_kv, n_tensors, fname);
|
||||
{
|
||||
std::map<enum ggml_type, uint32_t> n_type;
|
||||
|
|
@ -904,7 +904,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
n_type[type]++;
|
||||
}
|
||||
|
||||
printf("%s: Dumping metadata keys/values. Note: KV overrides do not apply in this output.\n", __func__);
|
||||
LOG_TEE("%s: Dumping metadata keys/values. Note: KV overrides do not apply in this output.\n", __func__);
|
||||
for (int i = 0; i < n_kv; i++) {
|
||||
const char * name = gguf_get_key(ctx, i);
|
||||
const enum gguf_type type = gguf_get_kv_type(ctx, i);
|
||||
|
|
@ -920,7 +920,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
}
|
||||
replace_all(value, "\n", "\\n");
|
||||
|
||||
printf("%s: - kv %3d: %42s %-16s = %s\n", __func__, i, name, type_name.c_str(), value.c_str());
|
||||
LOG_TEE("%s: - kv %3d: %42s %-16s = %s\n", __func__, i, name, type_name.c_str(), value.c_str());
|
||||
}
|
||||
|
||||
// print type counts
|
||||
|
|
@ -929,7 +929,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
continue;
|
||||
}
|
||||
|
||||
printf("%s: - type %4s: %4d tensors\n", __func__, ggml_type_name(kv.first), kv.second);
|
||||
LOG_TEE("%s: - type %4s: %4d tensors\n", __func__, ggml_type_name(kv.first), kv.second);
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -944,7 +944,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
size_t tensor_size = ggml_nbytes(cur);
|
||||
model_size += tensor_size;
|
||||
if (verbosity >= 3) {
|
||||
printf("%s: tensor[%d]: n_dims = %d, name = %s, tensor_size=%zu, offset=%zu, shape:[%" PRIu64 ", %" PRIu64 ", %" PRIu64 ", %" PRIu64 "], type = %s\n",
|
||||
LOG_TEE("%s: tensor[%d]: n_dims = %d, name = %s, tensor_size=%zu, offset=%zu, shape:[%" PRIu64 ", %" PRIu64 ", %" PRIu64 ", %" PRIu64 "], type = %s\n",
|
||||
__func__, i, ggml_n_dims(cur), cur->name, tensor_size, offset, cur->ne[0], cur->ne[1], cur->ne[2], cur->ne[3], ggml_type_name(type));
|
||||
}
|
||||
}
|
||||
|
|
@ -971,18 +971,18 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
|
||||
#ifdef GGML_USE_CUDA
|
||||
new_clip->backend = ggml_backend_cuda_init(0);
|
||||
printf("%s: CLIP using CUDA backend\n", __func__);
|
||||
LOG_TEE("%s: CLIP using CUDA backend\n", __func__);
|
||||
#endif
|
||||
|
||||
#ifdef GGML_USE_METAL
|
||||
new_clip->backend = ggml_backend_metal_init();
|
||||
printf("%s: CLIP using Metal backend\n", __func__);
|
||||
LOG_TEE("%s: CLIP using Metal backend\n", __func__);
|
||||
#endif
|
||||
|
||||
|
||||
if (!new_clip->backend) {
|
||||
new_clip->backend = ggml_backend_cpu_init();
|
||||
printf("%s: CLIP using CPU backend\n", __func__);
|
||||
LOG_TEE("%s: CLIP using CPU backend\n", __func__);
|
||||
}
|
||||
|
||||
// model size and capabilities
|
||||
|
|
@ -1006,15 +1006,15 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
new_clip->use_gelu = gguf_get_val_bool(ctx, idx);
|
||||
|
||||
if (verbosity >= 1) {
|
||||
printf("%s: text_encoder: %d\n", __func__, new_clip->has_text_encoder);
|
||||
printf("%s: vision_encoder: %d\n", __func__, new_clip->has_vision_encoder);
|
||||
printf("%s: llava_projector: %d\n", __func__, new_clip->has_llava_projector);
|
||||
printf("%s: model size: %.2f MB\n", __func__, model_size / 1024.0 / 1024.0);
|
||||
printf("%s: metadata size: %.2f MB\n", __func__, ggml_get_mem_size(meta) / 1024.0 / 1024.0);
|
||||
LOG_TEE("%s: text_encoder: %d\n", __func__, new_clip->has_text_encoder);
|
||||
LOG_TEE("%s: vision_encoder: %d\n", __func__, new_clip->has_vision_encoder);
|
||||
LOG_TEE("%s: llava_projector: %d\n", __func__, new_clip->has_llava_projector);
|
||||
LOG_TEE("%s: model size: %.2f MB\n", __func__, model_size / 1024.0 / 1024.0);
|
||||
LOG_TEE("%s: metadata size: %.2f MB\n", __func__, ggml_get_mem_size(meta) / 1024.0 / 1024.0);
|
||||
}
|
||||
}
|
||||
|
||||
printf("%s: params backend buffer size = % 6.2f MB (%i tensors)\n", __func__, model_size / (1024.0 * 1024.0), n_tensors);
|
||||
LOG_TEE("%s: params backend buffer size = % 6.2f MB (%i tensors)\n", __func__, model_size / (1024.0 * 1024.0), n_tensors);
|
||||
|
||||
// load tensors
|
||||
{
|
||||
|
|
@ -1027,7 +1027,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
|
||||
new_clip->ctx_data = ggml_init(params);
|
||||
if (!new_clip->ctx_data) {
|
||||
fprintf(stderr, "%s: ggml_init() failed\n", __func__);
|
||||
LOG_TEE("%s: ggml_init() failed\n", __func__);
|
||||
clip_free(new_clip);
|
||||
gguf_free(ctx);
|
||||
return nullptr;
|
||||
|
|
@ -1035,7 +1035,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
|
||||
auto fin = std::ifstream(fname, std::ios::binary);
|
||||
if (!fin) {
|
||||
printf("cannot open model file for loading tensors\n");
|
||||
LOG_TEE("cannot open model file for loading tensors\n");
|
||||
clip_free(new_clip);
|
||||
gguf_free(ctx);
|
||||
return nullptr;
|
||||
|
|
@ -1057,7 +1057,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
const size_t offset = gguf_get_data_offset(ctx) + gguf_get_tensor_offset(ctx, i);
|
||||
fin.seekg(offset, std::ios::beg);
|
||||
if (!fin) {
|
||||
printf("%s: failed to seek for tensor %s\n", __func__, name);
|
||||
LOG_TEE("%s: failed to seek for tensor %s\n", __func__, name);
|
||||
clip_free(new_clip);
|
||||
gguf_free(ctx);
|
||||
return nullptr;
|
||||
|
|
@ -1128,23 +1128,23 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
}
|
||||
|
||||
if (verbosity >= 2) {
|
||||
printf("\n%s: vision model hparams\n", __func__);
|
||||
printf("image_size %d\n", hparams.image_size);
|
||||
printf("patch_size %d\n", hparams.patch_size);
|
||||
printf("v_hidden_size %d\n", hparams.hidden_size);
|
||||
printf("v_n_intermediate %d\n", hparams.n_intermediate);
|
||||
printf("v_projection_dim %d\n", hparams.projection_dim);
|
||||
printf("v_n_head %d\n", hparams.n_head);
|
||||
printf("v_n_layer %d\n", hparams.n_layer);
|
||||
printf("v_eps %f\n", hparams.eps);
|
||||
printf("v_image_mean %f %f %f\n", new_clip->image_mean[0], new_clip->image_mean[1], new_clip->image_mean[2]);
|
||||
printf("v_image_std %f %f %f\n", new_clip->image_std[0], new_clip->image_std[1], new_clip->image_std[2]);
|
||||
printf("v_image_grid_pinpoints: ");
|
||||
LOG_TEE("\n%s: vision model hparams\n", __func__);
|
||||
LOG_TEE("image_size %d\n", hparams.image_size);
|
||||
LOG_TEE("patch_size %d\n", hparams.patch_size);
|
||||
LOG_TEE("v_hidden_size %d\n", hparams.hidden_size);
|
||||
LOG_TEE("v_n_intermediate %d\n", hparams.n_intermediate);
|
||||
LOG_TEE("v_projection_dim %d\n", hparams.projection_dim);
|
||||
LOG_TEE("v_n_head %d\n", hparams.n_head);
|
||||
LOG_TEE("v_n_layer %d\n", hparams.n_layer);
|
||||
LOG_TEE("v_eps %f\n", hparams.eps);
|
||||
LOG_TEE("v_image_mean %f %f %f\n", new_clip->image_mean[0], new_clip->image_mean[1], new_clip->image_mean[2]);
|
||||
LOG_TEE("v_image_std %f %f %f\n", new_clip->image_std[0], new_clip->image_std[1], new_clip->image_std[2]);
|
||||
LOG_TEE("v_image_grid_pinpoints: ");
|
||||
for (int i = 0; i < 32 && (hparams.image_grid_pinpoints[i] != 0); ++i) {
|
||||
printf("%d ", hparams.image_grid_pinpoints[i]);
|
||||
LOG_TEE("%d ", hparams.image_grid_pinpoints[i]);
|
||||
}
|
||||
printf("\n");
|
||||
printf("v_mm_patch_merge_type: %s\n", hparams.mm_patch_merge_type);
|
||||
LOG_TEE("\n");
|
||||
LOG_TEE("v_mm_patch_merge_type: %s\n", hparams.mm_patch_merge_type);
|
||||
|
||||
}
|
||||
|
||||
|
|
@ -1155,7 +1155,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
vision_model.pre_ln_w = get_tensor(new_clip->ctx_data, format(TN_LN_PRE, "v", "weight"));
|
||||
vision_model.pre_ln_b = get_tensor(new_clip->ctx_data, format(TN_LN_PRE, "v", "bias"));
|
||||
} catch(const std::exception& e) {
|
||||
fprintf(stderr, "%s: failed to load vision model tensors\n", __func__);
|
||||
LOG_TEE("%s: failed to load vision model tensors\n", __func__);
|
||||
}
|
||||
|
||||
// LLaVA projection
|
||||
|
|
@ -1184,7 +1184,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
} catch (std::runtime_error & e) { }
|
||||
try {
|
||||
vision_model.image_newline = get_tensor(new_clip->ctx_data, TN_IMAGE_NEWLINE);
|
||||
// fprintf(stderr, "%s: image_newline tensor (llava-1.6) found\n", __func__);
|
||||
// LOG_TEE("%s: image_newline tensor (llava-1.6) found\n", __func__);
|
||||
} catch (std::runtime_error & e) { }
|
||||
} else if (new_clip->proj_type == PROJECTOR_TYPE_LDP) {
|
||||
// MobileVLM projection
|
||||
|
|
@ -1264,7 +1264,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
|
|||
ggml_cgraph * gf = clip_image_build_graph(new_clip, &batch);
|
||||
ggml_gallocr_reserve(new_clip->compute_alloc, gf);
|
||||
size_t compute_memory_buffer_size = ggml_gallocr_get_buffer_size(new_clip->compute_alloc, 0);
|
||||
printf("%s: compute allocated memory: %.2f MB\n", __func__, compute_memory_buffer_size /1024.0/1024.0);
|
||||
LOG_TEE("%s: compute allocated memory: %.2f MB\n", __func__, compute_memory_buffer_size /1024.0/1024.0);
|
||||
}
|
||||
|
||||
return new_clip;
|
||||
|
|
@ -1304,7 +1304,7 @@ bool clip_image_load_from_file(const char * fname, clip_image_u8 * img) {
|
|||
int nx, ny, nc;
|
||||
auto * data = stbi_load(fname, &nx, &ny, &nc, 3);
|
||||
if (!data) {
|
||||
fprintf(stderr, "%s: failed to load image '%s'\n", __func__, fname);
|
||||
LOG_TEE("%s: failed to load image '%s'\n", __func__, fname);
|
||||
return false;
|
||||
}
|
||||
build_clip_img_from_data(data, nx, ny, img);
|
||||
|
|
@ -1316,7 +1316,7 @@ bool clip_image_load_from_bytes(const unsigned char * bytes, size_t bytes_length
|
|||
int nx, ny, nc;
|
||||
auto * data = stbi_load_from_memory(bytes, bytes_length, &nx, &ny, &nc, 3);
|
||||
if (!data) {
|
||||
fprintf(stderr, "%s: failed to decode image bytes\n", __func__);
|
||||
LOG_TEE("%s: failed to decode image bytes\n", __func__);
|
||||
return false;
|
||||
}
|
||||
build_clip_img_from_data(data, nx, ny, img);
|
||||
|
|
@ -1506,7 +1506,7 @@ static std::pair<int, int> select_best_resolution(const std::pair<int, int> & or
|
|||
int downscaled_height = static_cast<int>(original_height * scale);
|
||||
int effective_resolution = std::min(downscaled_width * downscaled_height, original_width * original_height);
|
||||
int wasted_resolution = (width * height) - effective_resolution;
|
||||
// fprintf(stderr, "resolution: %d %d, scale: %f, downscaled: %d %d, effective: %d, wasted: %d\n", width, height, scale, downscaled_width, downscaled_height, effective_resolution, wasted_resolution);
|
||||
// LOG_TEE("resolution: %d %d, scale: %f, downscaled: %d %d, effective: %d, wasted: %d\n", width, height, scale, downscaled_width, downscaled_height, effective_resolution, wasted_resolution);
|
||||
if (effective_resolution > max_effective_resolution || (effective_resolution == max_effective_resolution && wasted_resolution < min_wasted_resolution)) {
|
||||
max_effective_resolution = effective_resolution;
|
||||
min_wasted_resolution = wasted_resolution;
|
||||
|
|
@ -1545,7 +1545,7 @@ static std::vector<clip_image_u8*> divide_to_patches_u8(const clip_image_u8 & im
|
|||
bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, clip_image_f32_batch * res_imgs) {
|
||||
bool pad_to_square = true;
|
||||
if (!ctx->has_vision_encoder) {
|
||||
printf("This gguf file seems to have no vision encoder\n");
|
||||
LOG_TEE("This gguf file seems to have no vision encoder\n");
|
||||
return false;
|
||||
}
|
||||
auto & params = ctx->vision_model.hparams;
|
||||
|
|
@ -1622,7 +1622,7 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, cli
|
|||
}
|
||||
|
||||
for (size_t i = 0; i < patches.size(); i++) {
|
||||
// printf("patch %d: %d %d\n", i, patches[i]->nx, patches[i]->ny);
|
||||
// LOG_TEE("patch %d: %d %d\n", i, patches[i]->nx, patches[i]->ny);
|
||||
clip_image_u8_free(patches[i]);
|
||||
}
|
||||
|
||||
|
|
@ -1765,7 +1765,7 @@ int clip_n_patches(const struct clip_ctx * ctx) {
|
|||
|
||||
bool clip_image_encode(struct clip_ctx * ctx, const int n_threads, clip_image_f32 * img, float * vec) {
|
||||
if (!ctx->has_vision_encoder) {
|
||||
printf("This gguf file seems to have no vision encoder\n");
|
||||
LOG_TEE("This gguf file seems to have no vision encoder\n");
|
||||
return false;
|
||||
}
|
||||
|
||||
|
|
@ -1777,7 +1777,7 @@ bool clip_image_encode(struct clip_ctx * ctx, const int n_threads, clip_image_f3
|
|||
|
||||
bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_image_f32_batch * imgs, float * vec) {
|
||||
if (!ctx->has_vision_encoder) {
|
||||
printf("This gguf file seems to have no vision encoder\n");
|
||||
LOG_TEE("This gguf file seems to have no vision encoder\n");
|
||||
return false;
|
||||
}
|
||||
|
||||
|
|
@ -1939,7 +1939,7 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i
|
|||
new_type = type;
|
||||
if (new_type >= GGML_TYPE_Q2_K && name.find("embd") != std::string::npos) {
|
||||
new_type = GGML_TYPE_Q8_0; // ggml_get_rows needs non K type
|
||||
// fprintf(stderr, "%s: quantizing %s to %s\n", __func__, name.c_str(), ggml_type_name(new_type));
|
||||
// LOG_TEE("%s: quantizing %s to %s\n", __func__, name.c_str(), ggml_type_name(new_type));
|
||||
}
|
||||
const size_t n_elms = ggml_nelements(cur);
|
||||
float * f32_data;
|
||||
|
|
@ -1958,7 +1958,7 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i
|
|||
f32_data = (float *)conv_buf.data();
|
||||
break;
|
||||
default:
|
||||
printf("Please use an input file in f32 or f16\n");
|
||||
LOG_TEE("Please use an input file in f32 or f16\n");
|
||||
gguf_free(ctx_out);
|
||||
return false;
|
||||
}
|
||||
|
|
@ -1985,7 +1985,7 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i
|
|||
fout.put(0);
|
||||
}
|
||||
|
||||
printf("%s: n_dims = %d | quantize=%d | size = %f MB -> %f MB\n", name.c_str(), ggml_n_dims(cur), quantize,
|
||||
LOG_TEE("%s: n_dims = %d | quantize=%d | size = %f MB -> %f MB\n", name.c_str(), ggml_n_dims(cur), quantize,
|
||||
orig_size / 1024.0 / 1024.0, new_size / 1024.0 / 1024.0);
|
||||
}
|
||||
|
||||
|
|
@ -2001,8 +2001,8 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i
|
|||
gguf_free(ctx_out);
|
||||
|
||||
{
|
||||
printf("%s: original size = %8.2f MB\n", __func__, total_size_org / 1024.0 / 1024.0);
|
||||
printf("%s: quantized size = %8.2f MB\n", __func__, total_size_new / 1024.0 / 1024.0);
|
||||
LOG_TEE("%s: original size = %8.2f MB\n", __func__, total_size_org / 1024.0 / 1024.0);
|
||||
LOG_TEE("%s: quantized size = %8.2f MB\n", __func__, total_size_new / 1024.0 / 1024.0);
|
||||
}
|
||||
|
||||
return true;
|
||||
|
|
|
|||
|
|
@ -1,4 +1,5 @@
|
|||
#include "ggml.h"
|
||||
#include "log.h"
|
||||
#include "common.h"
|
||||
#include "clip.h"
|
||||
#include "llava.h"
|
||||
|
|
@ -18,7 +19,7 @@ static bool eval_tokens(struct llama_context * ctx_llama, std::vector<llama_toke
|
|||
n_eval = n_batch;
|
||||
}
|
||||
if (llama_decode(ctx_llama, llama_batch_get_one(&tokens[i], n_eval, *n_past, 0))) {
|
||||
fprintf(stderr, "%s : failed to eval. token %d/%d (batch size %d, n_past %d)\n", __func__, i, N, n_batch, *n_past);
|
||||
LOG_TEE("%s : failed to eval. token %d/%d (batch size %d, n_past %d)\n", __func__, i, N, n_batch, *n_past);
|
||||
return false;
|
||||
}
|
||||
*n_past += n_eval;
|
||||
|
|
@ -45,7 +46,7 @@ static const char * sample(struct llama_sampling_context * ctx_sampling,
|
|||
const llama_token id = llama_sampling_sample(ctx_sampling, ctx_llama, NULL);
|
||||
llama_sampling_accept(ctx_sampling, ctx_llama, id, true);
|
||||
static std::string ret;
|
||||
if (id == llama_token_eos(llama_get_model(ctx_llama))) {
|
||||
if (llama_token_is_eog(llama_get_model(ctx_llama), id)) {
|
||||
ret = "</s>";
|
||||
} else {
|
||||
ret = llama_token_to_piece(ctx_llama, id);
|
||||
|
|
@ -73,7 +74,7 @@ static llava_image_embed * llava_image_embed_make_with_prompt_base64(struct clip
|
|||
size_t img_base64_str_start, img_base64_str_end;
|
||||
find_image_tag_in_prompt(prompt, img_base64_str_start, img_base64_str_end);
|
||||
if (img_base64_str_start == std::string::npos || img_base64_str_end == std::string::npos) {
|
||||
fprintf(stderr, "%s: invalid base64 image tag. must be %s<base64 byte string>%s\n", __func__, IMG_BASE64_TAG_BEGIN, IMG_BASE64_TAG_END);
|
||||
LOG_TEE("%s: invalid base64 image tag. must be %s<base64 byte string>%s\n", __func__, IMG_BASE64_TAG_BEGIN, IMG_BASE64_TAG_END);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
@ -87,7 +88,7 @@ static llava_image_embed * llava_image_embed_make_with_prompt_base64(struct clip
|
|||
|
||||
auto embed = llava_image_embed_make_with_bytes(ctx_clip, n_threads, img_bytes.data(), img_bytes.size());
|
||||
if (!embed) {
|
||||
fprintf(stderr, "%s: could not load image from base64 string.\n", __func__);
|
||||
LOG_TEE("%s: could not load image from base64 string.\n", __func__);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
@ -112,8 +113,8 @@ struct llava_context {
|
|||
};
|
||||
|
||||
static void show_additional_info(int /*argc*/, char ** argv) {
|
||||
fprintf(stderr, "\n example usage: %s -m <llava-v1.5-7b/ggml-model-q5_k.gguf> --mmproj <llava-v1.5-7b/mmproj-model-f16.gguf> --image <path/to/an/image.jpg> [--temp 0.1] [-p \"describe the image in detail.\"]\n", argv[0]);
|
||||
fprintf(stderr, " note: a lower temperature value like 0.1 is recommended for better quality.\n");
|
||||
LOG_TEE("\n example usage: %s -m <llava-v1.5-7b/ggml-model-q5_k.gguf> --mmproj <llava-v1.5-7b/mmproj-model-f16.gguf> --image <path/to/an/image.jpg> [--temp 0.1] [-p \"describe the image in detail.\"]\n", argv[0]);
|
||||
LOG_TEE(" note: a lower temperature value like 0.1 is recommended for better quality.\n");
|
||||
}
|
||||
|
||||
static struct llava_image_embed * load_image(llava_context * ctx_llava, gpt_params * params) {
|
||||
|
|
@ -123,18 +124,18 @@ static struct llava_image_embed * load_image(llava_context * ctx_llava, gpt_para
|
|||
auto prompt = params->prompt;
|
||||
if (prompt_contains_image(prompt)) {
|
||||
if (!params->image.empty()) {
|
||||
fprintf(stderr, "using base64 encoded image instead of command line image path\n");
|
||||
LOG_TEE("using base64 encoded image instead of command line image path\n");
|
||||
}
|
||||
embed = llava_image_embed_make_with_prompt_base64(ctx_llava->ctx_clip, params->n_threads, prompt);
|
||||
if (!embed) {
|
||||
fprintf(stderr, "%s: can't load image from prompt\n", __func__);
|
||||
LOG_TEE("%s: can't load image from prompt\n", __func__);
|
||||
return NULL;
|
||||
}
|
||||
params->prompt = remove_image_from_prompt(prompt);
|
||||
} else {
|
||||
embed = llava_image_embed_make_with_filename(ctx_llava->ctx_clip, params->n_threads, params->image.c_str());
|
||||
if (!embed) {
|
||||
fprintf(stderr, "%s: is %s really an image file?\n", __func__, params->image.c_str());
|
||||
LOG_TEE("%s: is %s really an image file?\n", __func__, params->image.c_str());
|
||||
return NULL;
|
||||
}
|
||||
}
|
||||
|
|
@ -153,18 +154,18 @@ static void process_prompt(struct llava_context * ctx_llava, struct llava_image_
|
|||
// new templating mode: Provide the full prompt including system message and use <image> as a placeholder for the image
|
||||
system_prompt = prompt.substr(0, image_pos);
|
||||
user_prompt = prompt.substr(image_pos + std::string("<image>").length());
|
||||
printf("system_prompt: %s\n", system_prompt.c_str());
|
||||
LOG_TEE("system_prompt: %s\n", system_prompt.c_str());
|
||||
if (params->verbose_prompt) {
|
||||
auto tmp = ::llama_tokenize(ctx_llava->ctx_llama, system_prompt, true, true);
|
||||
for (int i = 0; i < (int) tmp.size(); i++) {
|
||||
printf("%6d -> '%s'\n", tmp[i], llama_token_to_piece(ctx_llava->ctx_llama, tmp[i]).c_str());
|
||||
LOG_TEE("%6d -> '%s'\n", tmp[i], llama_token_to_piece(ctx_llava->ctx_llama, tmp[i]).c_str());
|
||||
}
|
||||
}
|
||||
printf("user_prompt: %s\n", user_prompt.c_str());
|
||||
LOG_TEE("user_prompt: %s\n", user_prompt.c_str());
|
||||
if (params->verbose_prompt) {
|
||||
auto tmp = ::llama_tokenize(ctx_llava->ctx_llama, user_prompt, true, true);
|
||||
for (int i = 0; i < (int) tmp.size(); i++) {
|
||||
printf("%6d -> '%s'\n", tmp[i], llama_token_to_piece(ctx_llava->ctx_llama, tmp[i]).c_str());
|
||||
LOG_TEE("%6d -> '%s'\n", tmp[i], llama_token_to_piece(ctx_llava->ctx_llama, tmp[i]).c_str());
|
||||
}
|
||||
}
|
||||
} else {
|
||||
|
|
@ -174,7 +175,7 @@ static void process_prompt(struct llava_context * ctx_llava, struct llava_image_
|
|||
if (params->verbose_prompt) {
|
||||
auto tmp = ::llama_tokenize(ctx_llava->ctx_llama, user_prompt, true, true);
|
||||
for (int i = 0; i < (int) tmp.size(); i++) {
|
||||
printf("%6d -> '%s'\n", tmp[i], llama_token_to_piece(ctx_llava->ctx_llama, tmp[i]).c_str());
|
||||
LOG_TEE("%6d -> '%s'\n", tmp[i], llama_token_to_piece(ctx_llava->ctx_llama, tmp[i]).c_str());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -185,7 +186,7 @@ static void process_prompt(struct llava_context * ctx_llava, struct llava_image_
|
|||
|
||||
// generate the response
|
||||
|
||||
fprintf(stderr, "\n");
|
||||
LOG_TEE("\n");
|
||||
|
||||
struct llama_sampling_context * ctx_sampling = llama_sampling_init(params->sparams);
|
||||
std::string response = "";
|
||||
|
|
@ -224,7 +225,7 @@ static struct llava_context * llava_init(gpt_params * params) {
|
|||
|
||||
llama_model * model = llama_load_model_from_file(params->model.c_str(), model_params);
|
||||
if (model == NULL) {
|
||||
fprintf(stderr , "%s: error: unable to load model\n" , __func__);
|
||||
LOG_TEE("%s: error: unable to load model\n" , __func__);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
@ -234,7 +235,7 @@ static struct llava_context * llava_init(gpt_params * params) {
|
|||
llama_context * ctx_llama = llama_new_context_with_model(model, ctx_params);
|
||||
|
||||
if (ctx_llama == NULL) {
|
||||
fprintf(stderr , "%s: error: failed to create the llama_context\n" , __func__);
|
||||
LOG_TEE("%s: error: failed to create the llama_context\n" , __func__);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
@ -257,6 +258,12 @@ static void llava_free(struct llava_context * ctx_llava) {
|
|||
llama_backend_free();
|
||||
}
|
||||
|
||||
static void llama_log_callback_logTee(ggml_log_level level, const char * text, void * user_data) {
|
||||
(void) level;
|
||||
(void) user_data;
|
||||
LOG_TEE("%s", text);
|
||||
}
|
||||
|
||||
int main(int argc, char ** argv) {
|
||||
ggml_time_init();
|
||||
|
||||
|
|
@ -266,6 +273,14 @@ int main(int argc, char ** argv) {
|
|||
show_additional_info(argc, argv);
|
||||
return 1;
|
||||
}
|
||||
|
||||
#ifndef LOG_DISABLE_LOGS
|
||||
log_set_target(log_filename_generator("llava", "log"));
|
||||
LOG_TEE("Log start\n");
|
||||
log_dump_cmdline(argc, argv);
|
||||
llama_log_set(llama_log_callback_logTee, nullptr);
|
||||
#endif // LOG_DISABLE_LOGS
|
||||
|
||||
if (params.mmproj.empty() || (params.image.empty() && !prompt_contains_image(params.prompt))) {
|
||||
gpt_print_usage(argc, argv, params);
|
||||
show_additional_info(argc, argv);
|
||||
|
|
@ -274,7 +289,7 @@ int main(int argc, char ** argv) {
|
|||
|
||||
auto ctx_llava = llava_init(¶ms);
|
||||
if (ctx_llava == NULL) {
|
||||
fprintf(stderr, "%s: error: failed to init llava\n", __func__);
|
||||
LOG_TEE("%s: error: failed to init llava\n", __func__);
|
||||
return 1;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -54,7 +54,7 @@ static std::pair<int, int> select_best_resolution(const std::pair<int, int>& ori
|
|||
int downscaled_height = static_cast<int>(original_height * scale);
|
||||
int effective_resolution = std::min(downscaled_width * downscaled_height, original_width * original_height);
|
||||
int wasted_resolution = (width * height) - effective_resolution;
|
||||
// fprintf(stderr, "resolution: %d %d, scale: %f, downscaled: %d %d, effective: %d, wasted: %d\n", width, height, scale, downscaled_width, downscaled_height, effective_resolution, wasted_resolution);
|
||||
// LOG_TEE("resolution: %d %d, scale: %f, downscaled: %d %d, effective: %d, wasted: %d\n", width, height, scale, downscaled_width, downscaled_height, effective_resolution, wasted_resolution);
|
||||
if (effective_resolution > max_effective_resolution || (effective_resolution == max_effective_resolution && wasted_resolution < min_wasted_resolution)) {
|
||||
max_effective_resolution = effective_resolution;
|
||||
min_wasted_resolution = wasted_resolution;
|
||||
|
|
@ -154,13 +154,13 @@ static bool clip_llava_handle_patches(clip_ctx * ctx_clip, std::vector<float *>
|
|||
model.newline = ggml_new_tensor_1d(model.ctx, GGML_TYPE_F32, newline_tmp->ne[0]);
|
||||
if (newline_tmp->backend != GGML_BACKEND_TYPE_CPU) {
|
||||
if (newline_tmp->buffer == NULL) {
|
||||
printf("newline_tmp tensor buffer is NULL\n");
|
||||
LOG_TEE("newline_tmp tensor buffer is NULL\n");
|
||||
}
|
||||
ggml_backend_tensor_get(newline_tmp, model.newline->data, 0, ggml_nbytes(newline_tmp));
|
||||
} else {
|
||||
model.newline->data = newline_tmp->data;
|
||||
if (model.newline->data == NULL) {
|
||||
printf("newline_tmp tensor data is NULL\n");
|
||||
LOG_TEE("newline_tmp tensor data is NULL\n");
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -224,7 +224,7 @@ static bool encode_image_with_clip(clip_ctx * ctx_clip, int n_threads, const cli
|
|||
img_res_v.size = 0;
|
||||
img_res_v.data = nullptr;
|
||||
if (!clip_image_preprocess(ctx_clip, img, &img_res_v)) {
|
||||
fprintf(stderr, "%s: unable to preprocess image\n", __func__);
|
||||
LOG_TEE("%s: unable to preprocess image\n", __func__);
|
||||
delete[] img_res_v.data;
|
||||
return false;
|
||||
}
|
||||
|
|
@ -239,7 +239,7 @@ static bool encode_image_with_clip(clip_ctx * ctx_clip, int n_threads, const cli
|
|||
bool encoded = clip_image_encode(ctx_clip, n_threads, &img_res_v.data[0], image_embd); // image_embd shape is 576 x 4096
|
||||
delete[] img_res_v.data;
|
||||
if (!encoded) {
|
||||
fprintf(stderr, "Unable to encode image\n");
|
||||
LOG_TEE("Unable to encode image\n");
|
||||
|
||||
return false;
|
||||
}
|
||||
|
|
@ -252,12 +252,12 @@ static bool encode_image_with_clip(clip_ctx * ctx_clip, int n_threads, const cli
|
|||
image_embd_v[i] = (float *)malloc(clip_embd_nbytes(ctx_clip)); // 576 patches * 4096 embeddings * 4 bytes = 9437184
|
||||
const bool encoded = clip_image_encode(ctx_clip, n_threads, &img_res_v.data[i], image_embd_v[i]); // image data is in 3x336x336 format and will be converted to 336x336x3 inside
|
||||
if (!encoded) {
|
||||
fprintf(stderr, "Unable to encode image - spatial_unpad - subimage %d of %d\n", (int) i+1, (int) img_res_v.size);
|
||||
LOG_TEE("Unable to encode image - spatial_unpad - subimage %d of %d\n", (int) i+1, (int) img_res_v.size);
|
||||
return false;
|
||||
}
|
||||
}
|
||||
const int64_t t_img_enc_batch_us = ggml_time_us();
|
||||
printf("%s: %d segments encoded in %8.2f ms\n", __func__, (int)img_res_v.size, (t_img_enc_batch_us - t_img_enc_start_us) / 1000.0);
|
||||
LOG_TEE("%s: %d segments encoded in %8.2f ms\n", __func__, (int)img_res_v.size, (t_img_enc_batch_us - t_img_enc_start_us) / 1000.0);
|
||||
|
||||
const int32_t * image_grid = clip_image_grid(ctx_clip);
|
||||
|
||||
|
|
@ -290,12 +290,12 @@ static bool encode_image_with_clip(clip_ctx * ctx_clip, int n_threads, const cli
|
|||
// clip_image_save_to_bmp(*tmp, "image_feature.bmp");
|
||||
}
|
||||
|
||||
printf("%s: image embedding created: %d tokens\n", __func__, *n_img_pos);
|
||||
LOG_TEE("%s: image embedding created: %d tokens\n", __func__, *n_img_pos);
|
||||
|
||||
const int64_t t_img_enc_end_us = ggml_time_us();
|
||||
float t_img_enc_ms = (t_img_enc_end_us - t_img_enc_start_us) / 1000.0;
|
||||
|
||||
printf("\n%s: image encoded in %8.2f ms by CLIP (%8.2f ms per image patch)\n", __func__, t_img_enc_ms, t_img_enc_ms / *n_img_pos);
|
||||
LOG_TEE("\n%s: image encoded in %8.2f ms by CLIP (%8.2f ms per image patch)\n", __func__, t_img_enc_ms, t_img_enc_ms / *n_img_pos);
|
||||
|
||||
return true;
|
||||
}
|
||||
|
|
@ -305,7 +305,7 @@ bool llava_validate_embed_size(const llama_context * ctx_llama, const clip_ctx *
|
|||
int n_llama_embd = llama_n_embd(llama_get_model(ctx_llama));
|
||||
auto n_image_embd = clip_n_mmproj_embd(ctx_clip);
|
||||
if (n_image_embd != n_llama_embd) {
|
||||
printf("%s: embedding dim of the multimodal projector (%d) is not equal to that of LLaMA (%d). Make sure that you use the correct mmproj file.\n", __func__, n_image_embd, n_llama_embd);
|
||||
LOG_TEE("%s: embedding dim of the multimodal projector (%d) is not equal to that of LLaMA (%d). Make sure that you use the correct mmproj file.\n", __func__, n_image_embd, n_llama_embd);
|
||||
return false;
|
||||
}
|
||||
return true;
|
||||
|
|
@ -314,13 +314,13 @@ bool llava_validate_embed_size(const llama_context * ctx_llama, const clip_ctx *
|
|||
bool llava_image_embed_make_with_clip_img(clip_ctx * ctx_clip, int n_threads, const clip_image_u8 * img, float ** image_embd_out, int * n_img_pos_out) {
|
||||
float * image_embd = (float *)malloc(clip_embd_nbytes(ctx_clip)*6); // TODO: base on gridsize/llava model
|
||||
if (!image_embd) {
|
||||
fprintf(stderr, "Unable to allocate memory for image embeddings\n");
|
||||
LOG_TEE("Unable to allocate memory for image embeddings\n");
|
||||
return false;
|
||||
}
|
||||
|
||||
int n_img_pos;
|
||||
if (!encode_image_with_clip(ctx_clip, n_threads, img, image_embd, &n_img_pos)) {
|
||||
fprintf(stderr, "%s: cannot encode image, aborting\n", __func__);
|
||||
LOG_TEE("%s: cannot encode image, aborting\n", __func__);
|
||||
free(image_embd);
|
||||
return false;
|
||||
}
|
||||
|
|
@ -340,7 +340,7 @@ bool llava_eval_image_embed(llama_context * ctx_llama, const struct llava_image_
|
|||
}
|
||||
llama_batch batch = {int32_t(n_eval), nullptr, (image_embed->embed+i*n_embd), nullptr, nullptr, nullptr, nullptr, *n_past, 1, 0, };
|
||||
if (llama_decode(ctx_llama, batch)) {
|
||||
fprintf(stderr, "%s : failed to eval\n", __func__);
|
||||
LOG_TEE("%s : failed to eval\n", __func__);
|
||||
return false;
|
||||
}
|
||||
*n_past += n_eval;
|
||||
|
|
@ -352,7 +352,7 @@ struct llava_image_embed * llava_image_embed_make_with_bytes(struct clip_ctx * c
|
|||
clip_image_u8 * img = clip_image_u8_init();
|
||||
if (!clip_image_load_from_bytes(image_bytes, image_bytes_length, img)) {
|
||||
clip_image_u8_free(img);
|
||||
fprintf(stderr, "%s: can't load image from bytes, is it a valid image?", __func__);
|
||||
LOG_TEE("%s: can't load image from bytes, is it a valid image?", __func__);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
@ -361,7 +361,7 @@ struct llava_image_embed * llava_image_embed_make_with_bytes(struct clip_ctx * c
|
|||
bool image_embed_result = llava_image_embed_make_with_clip_img(ctx_clip, n_threads, img, &image_embed, &n_image_pos);
|
||||
if (!image_embed_result) {
|
||||
clip_image_u8_free(img);
|
||||
fprintf(stderr, "%s: coulnd't embed the image\n", __func__);
|
||||
LOG_TEE("%s: coulnd't embed the image\n", __func__);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
@ -375,7 +375,7 @@ struct llava_image_embed * llava_image_embed_make_with_bytes(struct clip_ctx * c
|
|||
static bool load_file_to_bytes(const char* path, unsigned char** bytesOut, long *sizeOut) {
|
||||
auto file = fopen(path, "rb");
|
||||
if (file == NULL) {
|
||||
fprintf(stderr, "%s: can't read file %s\n", __func__, path);
|
||||
LOG_TEE("%s: can't read file %s\n", __func__, path);
|
||||
return false;
|
||||
}
|
||||
|
||||
|
|
@ -385,7 +385,7 @@ static bool load_file_to_bytes(const char* path, unsigned char** bytesOut, long
|
|||
|
||||
auto buffer = (unsigned char *)malloc(fileSize); // Allocate memory to hold the file data
|
||||
if (buffer == NULL) {
|
||||
fprintf(stderr, "%s: failed to alloc %ld bytes for file %s\n", __func__, fileSize, path);
|
||||
LOG_TEE("%s: failed to alloc %ld bytes for file %s\n", __func__, fileSize, path);
|
||||
perror("Memory allocation error");
|
||||
fclose(file);
|
||||
return false;
|
||||
|
|
@ -410,7 +410,7 @@ struct llava_image_embed * llava_image_embed_make_with_filename(struct clip_ctx
|
|||
long image_bytes_length;
|
||||
auto loaded = load_file_to_bytes(image_path, &image_bytes, &image_bytes_length);
|
||||
if (!loaded) {
|
||||
fprintf(stderr, "%s: failed to load %s\n", __func__, image_path);
|
||||
LOG_TEE("%s: failed to load %s\n", __func__, image_path);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -299,7 +299,7 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
fflush(stdout);
|
||||
|
||||
if (id == llama_token_eos(model)) {
|
||||
if (llama_token_is_eog(model, id)) {
|
||||
has_eos = true;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -30,7 +30,6 @@ int main(int argc, char ** argv){
|
|||
|
||||
// load the model
|
||||
std::tie(model, ctx) = llama_init_from_gpt_params(params);
|
||||
llama_set_rng_seed(ctx, params.seed);
|
||||
GGML_ASSERT(llama_n_vocab(model) < (1 << 16));
|
||||
|
||||
// tokenize the prompt
|
||||
|
|
|
|||
|
|
@ -38,7 +38,6 @@ int main(int argc, char ** argv){
|
|||
|
||||
// load the model
|
||||
std::tie(model, ctx) = llama_init_from_gpt_params(params);
|
||||
llama_set_rng_seed(ctx, params.seed);
|
||||
GGML_ASSERT(llama_n_vocab(model) < (1 << 16));
|
||||
|
||||
// tokenize the prompt
|
||||
|
|
@ -141,7 +140,7 @@ int main(int argc, char ** argv){
|
|||
printf("%s", token_str.c_str());
|
||||
}
|
||||
|
||||
if (id == llama_token_eos(model)) {
|
||||
if (llama_token_is_eog(model, id)) {
|
||||
has_eos = true;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -240,7 +240,6 @@ int main(int argc, char ** argv) {
|
|||
return 1;
|
||||
}
|
||||
session_tokens.resize(n_token_count_out);
|
||||
llama_set_rng_seed(ctx, params.seed);
|
||||
LOG_TEE("%s: loaded a session with prompt size of %d tokens\n", __func__, (int)session_tokens.size());
|
||||
}
|
||||
}
|
||||
|
|
@ -795,8 +794,8 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
}
|
||||
|
||||
// deal with end of text token in interactive mode
|
||||
if (llama_sampling_last(ctx_sampling) == llama_token_eos(model)) {
|
||||
// deal with end of generation tokens in interactive mode
|
||||
if (llama_token_is_eog(model, llama_sampling_last(ctx_sampling))) {
|
||||
LOG("found EOS token\n");
|
||||
|
||||
if (params.interactive) {
|
||||
|
|
@ -920,8 +919,8 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
}
|
||||
|
||||
// end of text token
|
||||
if (!embd.empty() && embd.back() == llama_token_eos(model) && !(params.instruct || params.interactive || params.chatml)) {
|
||||
// end of generation
|
||||
if (!embd.empty() && llama_token_is_eog(model, embd.back()) && !(params.instruct || params.interactive || params.chatml)) {
|
||||
LOG_TEE(" [end of text]\n");
|
||||
break;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -359,7 +359,7 @@ int main(int argc, char ** argv) {
|
|||
// client.id, client.seq_id, id, client.n_decoded, client.i_batch, token_str.c_str());
|
||||
|
||||
if (client.n_decoded > 2 &&
|
||||
(id == llama_token_eos(model) ||
|
||||
(llama_token_is_eog(model, id) ||
|
||||
(params.n_predict > 0 && client.n_decoded + client.n_prompt >= params.n_predict) ||
|
||||
client.response.find("User:") != std::string::npos ||
|
||||
client.response.find('\n') != std::string::npos)) {
|
||||
|
|
|
|||
|
|
@ -252,8 +252,8 @@ int main(int argc, char ** argv) {
|
|||
// sample the most likely token
|
||||
const llama_token new_token_id = llama_sample_token_greedy(ctx, &candidates_p);
|
||||
|
||||
// is it an end of stream?
|
||||
if (new_token_id == llama_token_eos(model) || n_cur == n_len) {
|
||||
// is it an end of generation?
|
||||
if (llama_token_is_eog(model, new_token_id) || n_cur == n_len) {
|
||||
LOG_TEE("\n");
|
||||
|
||||
break;
|
||||
|
|
|
|||
|
|
@ -1,12 +1,29 @@
|
|||
set(TARGET server)
|
||||
option(LLAMA_SERVER_VERBOSE "Build verbose logging option for Server" ON)
|
||||
option(LLAMA_SERVER_SSL "Build SSL support for the server" OFF)
|
||||
include_directories(${CMAKE_CURRENT_SOURCE_DIR})
|
||||
add_executable(${TARGET}
|
||||
include_directories(${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_BINARY_DIR})
|
||||
set(TARGET_SRCS
|
||||
server.cpp
|
||||
utils.hpp
|
||||
httplib.h
|
||||
)
|
||||
set(PUBLIC_ASSETS
|
||||
index.html
|
||||
index.js
|
||||
completion.js
|
||||
json-schema-to-grammar.mjs
|
||||
)
|
||||
foreach(asset ${PUBLIC_ASSETS})
|
||||
set(input "${CMAKE_CURRENT_SOURCE_DIR}/public/${asset}")
|
||||
set(output "${CMAKE_CURRENT_BINARY_DIR}/${asset}.hpp")
|
||||
list(APPEND TARGET_SRCS ${output})
|
||||
add_custom_command(
|
||||
DEPENDS "${input}"
|
||||
OUTPUT "${output}"
|
||||
COMMAND "${CMAKE_COMMAND}" "-DINPUT=${input}" "-DOUTPUT=${output}" -P "${PROJECT_SOURCE_DIR}/scripts/xxd.cmake"
|
||||
)
|
||||
endforeach()
|
||||
add_executable(${TARGET} ${TARGET_SRCS})
|
||||
install(TARGETS ${TARGET} RUNTIME)
|
||||
target_compile_definitions(${TARGET} PRIVATE
|
||||
SERVER_VERBOSE=$<BOOL:${LLAMA_SERVER_VERBOSE}>
|
||||
|
|
|
|||
|
|
@ -1,496 +0,0 @@
|
|||
unsigned char completion_js[] = {
|
||||
0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x44,
|
||||
0x65, 0x66, 0x61, 0x75, 0x6c, 0x74, 0x73, 0x20, 0x3d, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x73, 0x74, 0x72, 0x65, 0x61, 0x6d, 0x3a, 0x20, 0x74, 0x72,
|
||||
0x75, 0x65, 0x2c, 0x0a, 0x20, 0x20, 0x6e, 0x5f, 0x70, 0x72, 0x65, 0x64,
|
||||
0x69, 0x63, 0x74, 0x3a, 0x20, 0x35, 0x30, 0x30, 0x2c, 0x0a, 0x20, 0x20,
|
||||
0x74, 0x65, 0x6d, 0x70, 0x65, 0x72, 0x61, 0x74, 0x75, 0x72, 0x65, 0x3a,
|
||||
0x20, 0x30, 0x2e, 0x32, 0x2c, 0x0a, 0x20, 0x20, 0x73, 0x74, 0x6f, 0x70,
|
||||
0x3a, 0x20, 0x5b, 0x22, 0x3c, 0x2f, 0x73, 0x3e, 0x22, 0x5d, 0x0a, 0x7d,
|
||||
0x3b, 0x0a, 0x0a, 0x6c, 0x65, 0x74, 0x20, 0x67, 0x65, 0x6e, 0x65, 0x72,
|
||||
0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73, 0x65, 0x74, 0x74, 0x69, 0x6e,
|
||||
0x67, 0x73, 0x20, 0x3d, 0x20, 0x6e, 0x75, 0x6c, 0x6c, 0x3b, 0x0a, 0x0a,
|
||||
0x0a, 0x2f, 0x2f, 0x20, 0x43, 0x6f, 0x6d, 0x70, 0x6c, 0x65, 0x74, 0x65,
|
||||
0x73, 0x20, 0x74, 0x68, 0x65, 0x20, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74,
|
||||
0x20, 0x61, 0x73, 0x20, 0x61, 0x20, 0x67, 0x65, 0x6e, 0x65, 0x72, 0x61,
|
||||
0x74, 0x6f, 0x72, 0x2e, 0x20, 0x52, 0x65, 0x63, 0x6f, 0x6d, 0x6d, 0x65,
|
||||
0x6e, 0x64, 0x65, 0x64, 0x20, 0x66, 0x6f, 0x72, 0x20, 0x6d, 0x6f, 0x73,
|
||||
0x74, 0x20, 0x75, 0x73, 0x65, 0x20, 0x63, 0x61, 0x73, 0x65, 0x73, 0x2e,
|
||||
0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x45, 0x78, 0x61, 0x6d, 0x70,
|
||||
0x6c, 0x65, 0x3a, 0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20,
|
||||
0x20, 0x69, 0x6d, 0x70, 0x6f, 0x72, 0x74, 0x20, 0x7b, 0x20, 0x6c, 0x6c,
|
||||
0x61, 0x6d, 0x61, 0x20, 0x7d, 0x20, 0x66, 0x72, 0x6f, 0x6d, 0x20, 0x27,
|
||||
0x2f, 0x63, 0x6f, 0x6d, 0x70, 0x6c, 0x65, 0x74, 0x69, 0x6f, 0x6e, 0x2e,
|
||||
0x6a, 0x73, 0x27, 0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x72, 0x65, 0x71, 0x75, 0x65,
|
||||
0x73, 0x74, 0x20, 0x3d, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x28, 0x22,
|
||||
0x54, 0x65, 0x6c, 0x6c, 0x20, 0x6d, 0x65, 0x20, 0x61, 0x20, 0x6a, 0x6f,
|
||||
0x6b, 0x65, 0x22, 0x2c, 0x20, 0x7b, 0x6e, 0x5f, 0x70, 0x72, 0x65, 0x64,
|
||||
0x69, 0x63, 0x74, 0x3a, 0x20, 0x38, 0x30, 0x30, 0x7d, 0x29, 0x0a, 0x2f,
|
||||
0x2f, 0x20, 0x20, 0x20, 0x20, 0x66, 0x6f, 0x72, 0x20, 0x61, 0x77, 0x61,
|
||||
0x69, 0x74, 0x20, 0x28, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x63, 0x68,
|
||||
0x75, 0x6e, 0x6b, 0x20, 0x6f, 0x66, 0x20, 0x72, 0x65, 0x71, 0x75, 0x65,
|
||||
0x73, 0x74, 0x29, 0x20, 0x7b, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x64, 0x6f, 0x63, 0x75, 0x6d, 0x65, 0x6e, 0x74, 0x2e, 0x77,
|
||||
0x72, 0x69, 0x74, 0x65, 0x28, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64,
|
||||
0x61, 0x74, 0x61, 0x2e, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x29,
|
||||
0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x2f, 0x2f, 0x0a,
|
||||
0x65, 0x78, 0x70, 0x6f, 0x72, 0x74, 0x20, 0x61, 0x73, 0x79, 0x6e, 0x63,
|
||||
0x20, 0x66, 0x75, 0x6e, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x2a, 0x20, 0x6c,
|
||||
0x6c, 0x61, 0x6d, 0x61, 0x28, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x2c,
|
||||
0x20, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x73, 0x20, 0x3d, 0x20, 0x7b, 0x7d,
|
||||
0x2c, 0x20, 0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x20, 0x3d, 0x20, 0x7b,
|
||||
0x7d, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x6c, 0x65, 0x74, 0x20, 0x63,
|
||||
0x6f, 0x6e, 0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x20, 0x3d, 0x20,
|
||||
0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x2e, 0x63, 0x6f, 0x6e, 0x74, 0x72,
|
||||
0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x3b, 0x0a, 0x20, 0x20, 0x63, 0x6f, 0x6e,
|
||||
0x73, 0x74, 0x20, 0x61, 0x70, 0x69, 0x5f, 0x75, 0x72, 0x6c, 0x20, 0x3d,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x2e, 0x61, 0x70, 0x69, 0x5f,
|
||||
0x75, 0x72, 0x6c, 0x20, 0x7c, 0x7c, 0x20, 0x22, 0x22, 0x3b, 0x0a, 0x0a,
|
||||
0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x21, 0x63, 0x6f, 0x6e, 0x74, 0x72,
|
||||
0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x20,
|
||||
0x3d, 0x20, 0x6e, 0x65, 0x77, 0x20, 0x41, 0x62, 0x6f, 0x72, 0x74, 0x43,
|
||||
0x6f, 0x6e, 0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x28, 0x29, 0x3b,
|
||||
0x0a, 0x20, 0x20, 0x7d, 0x0a, 0x0a, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73,
|
||||
0x74, 0x20, 0x63, 0x6f, 0x6d, 0x70, 0x6c, 0x65, 0x74, 0x69, 0x6f, 0x6e,
|
||||
0x50, 0x61, 0x72, 0x61, 0x6d, 0x73, 0x20, 0x3d, 0x20, 0x7b, 0x20, 0x2e,
|
||||
0x2e, 0x2e, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x44, 0x65, 0x66, 0x61, 0x75,
|
||||
0x6c, 0x74, 0x73, 0x2c, 0x20, 0x2e, 0x2e, 0x2e, 0x70, 0x61, 0x72, 0x61,
|
||||
0x6d, 0x73, 0x2c, 0x20, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x20, 0x7d,
|
||||
0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x72,
|
||||
0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x20, 0x3d, 0x20, 0x61, 0x77,
|
||||
0x61, 0x69, 0x74, 0x20, 0x66, 0x65, 0x74, 0x63, 0x68, 0x28, 0x60, 0x24,
|
||||
0x7b, 0x61, 0x70, 0x69, 0x5f, 0x75, 0x72, 0x6c, 0x7d, 0x2f, 0x63, 0x6f,
|
||||
0x6d, 0x70, 0x6c, 0x65, 0x74, 0x69, 0x6f, 0x6e, 0x60, 0x2c, 0x20, 0x7b,
|
||||
0x0a, 0x20, 0x20, 0x20, 0x20, 0x6d, 0x65, 0x74, 0x68, 0x6f, 0x64, 0x3a,
|
||||
0x20, 0x27, 0x50, 0x4f, 0x53, 0x54, 0x27, 0x2c, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x62, 0x6f, 0x64, 0x79, 0x3a, 0x20, 0x4a, 0x53, 0x4f, 0x4e, 0x2e,
|
||||
0x73, 0x74, 0x72, 0x69, 0x6e, 0x67, 0x69, 0x66, 0x79, 0x28, 0x63, 0x6f,
|
||||
0x6d, 0x70, 0x6c, 0x65, 0x74, 0x69, 0x6f, 0x6e, 0x50, 0x61, 0x72, 0x61,
|
||||
0x6d, 0x73, 0x29, 0x2c, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x68, 0x65, 0x61,
|
||||
0x64, 0x65, 0x72, 0x73, 0x3a, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x27, 0x43, 0x6f, 0x6e, 0x6e, 0x65, 0x63, 0x74, 0x69, 0x6f,
|
||||
0x6e, 0x27, 0x3a, 0x20, 0x27, 0x6b, 0x65, 0x65, 0x70, 0x2d, 0x61, 0x6c,
|
||||
0x69, 0x76, 0x65, 0x27, 0x2c, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x27, 0x43, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x2d, 0x54, 0x79, 0x70,
|
||||
0x65, 0x27, 0x3a, 0x20, 0x27, 0x61, 0x70, 0x70, 0x6c, 0x69, 0x63, 0x61,
|
||||
0x74, 0x69, 0x6f, 0x6e, 0x2f, 0x6a, 0x73, 0x6f, 0x6e, 0x27, 0x2c, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x27, 0x41, 0x63, 0x63, 0x65, 0x70,
|
||||
0x74, 0x27, 0x3a, 0x20, 0x27, 0x74, 0x65, 0x78, 0x74, 0x2f, 0x65, 0x76,
|
||||
0x65, 0x6e, 0x74, 0x2d, 0x73, 0x74, 0x72, 0x65, 0x61, 0x6d, 0x27, 0x2c,
|
||||
0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2e, 0x2e, 0x2e, 0x28, 0x70,
|
||||
0x61, 0x72, 0x61, 0x6d, 0x73, 0x2e, 0x61, 0x70, 0x69, 0x5f, 0x6b, 0x65,
|
||||
0x79, 0x20, 0x3f, 0x20, 0x7b, 0x27, 0x41, 0x75, 0x74, 0x68, 0x6f, 0x72,
|
||||
0x69, 0x7a, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x27, 0x3a, 0x20, 0x60, 0x42,
|
||||
0x65, 0x61, 0x72, 0x65, 0x72, 0x20, 0x24, 0x7b, 0x70, 0x61, 0x72, 0x61,
|
||||
0x6d, 0x73, 0x2e, 0x61, 0x70, 0x69, 0x5f, 0x6b, 0x65, 0x79, 0x7d, 0x60,
|
||||
0x7d, 0x20, 0x3a, 0x20, 0x7b, 0x7d, 0x29, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x7d, 0x2c, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x73, 0x69, 0x67, 0x6e, 0x61,
|
||||
0x6c, 0x3a, 0x20, 0x63, 0x6f, 0x6e, 0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65,
|
||||
0x72, 0x2e, 0x73, 0x69, 0x67, 0x6e, 0x61, 0x6c, 0x2c, 0x0a, 0x20, 0x20,
|
||||
0x7d, 0x29, 0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74,
|
||||
0x20, 0x72, 0x65, 0x61, 0x64, 0x65, 0x72, 0x20, 0x3d, 0x20, 0x72, 0x65,
|
||||
0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x2e, 0x62, 0x6f, 0x64, 0x79, 0x2e,
|
||||
0x67, 0x65, 0x74, 0x52, 0x65, 0x61, 0x64, 0x65, 0x72, 0x28, 0x29, 0x3b,
|
||||
0x0a, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x64, 0x65, 0x63,
|
||||
0x6f, 0x64, 0x65, 0x72, 0x20, 0x3d, 0x20, 0x6e, 0x65, 0x77, 0x20, 0x54,
|
||||
0x65, 0x78, 0x74, 0x44, 0x65, 0x63, 0x6f, 0x64, 0x65, 0x72, 0x28, 0x29,
|
||||
0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x6c, 0x65, 0x74, 0x20, 0x63, 0x6f, 0x6e,
|
||||
0x74, 0x65, 0x6e, 0x74, 0x20, 0x3d, 0x20, 0x22, 0x22, 0x3b, 0x0a, 0x20,
|
||||
0x20, 0x6c, 0x65, 0x74, 0x20, 0x6c, 0x65, 0x66, 0x74, 0x6f, 0x76, 0x65,
|
||||
0x72, 0x20, 0x3d, 0x20, 0x22, 0x22, 0x3b, 0x20, 0x2f, 0x2f, 0x20, 0x42,
|
||||
0x75, 0x66, 0x66, 0x65, 0x72, 0x20, 0x66, 0x6f, 0x72, 0x20, 0x70, 0x61,
|
||||
0x72, 0x74, 0x69, 0x61, 0x6c, 0x6c, 0x79, 0x20, 0x72, 0x65, 0x61, 0x64,
|
||||
0x20, 0x6c, 0x69, 0x6e, 0x65, 0x73, 0x0a, 0x0a, 0x20, 0x20, 0x74, 0x72,
|
||||
0x79, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x6c, 0x65, 0x74, 0x20,
|
||||
0x63, 0x6f, 0x6e, 0x74, 0x20, 0x3d, 0x20, 0x74, 0x72, 0x75, 0x65, 0x3b,
|
||||
0x0a, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x77, 0x68, 0x69, 0x6c, 0x65, 0x20,
|
||||
0x28, 0x63, 0x6f, 0x6e, 0x74, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x72, 0x65, 0x73,
|
||||
0x75, 0x6c, 0x74, 0x20, 0x3d, 0x20, 0x61, 0x77, 0x61, 0x69, 0x74, 0x20,
|
||||
0x72, 0x65, 0x61, 0x64, 0x65, 0x72, 0x2e, 0x72, 0x65, 0x61, 0x64, 0x28,
|
||||
0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20,
|
||||
0x28, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x64, 0x6f, 0x6e, 0x65,
|
||||
0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x62, 0x72, 0x65, 0x61, 0x6b, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x7d, 0x0a, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f,
|
||||
0x20, 0x41, 0x64, 0x64, 0x20, 0x61, 0x6e, 0x79, 0x20, 0x6c, 0x65, 0x66,
|
||||
0x74, 0x6f, 0x76, 0x65, 0x72, 0x20, 0x64, 0x61, 0x74, 0x61, 0x20, 0x74,
|
||||
0x6f, 0x20, 0x74, 0x68, 0x65, 0x20, 0x63, 0x75, 0x72, 0x72, 0x65, 0x6e,
|
||||
0x74, 0x20, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x20, 0x6f, 0x66, 0x20, 0x64,
|
||||
0x61, 0x74, 0x61, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f,
|
||||
0x6e, 0x73, 0x74, 0x20, 0x74, 0x65, 0x78, 0x74, 0x20, 0x3d, 0x20, 0x6c,
|
||||
0x65, 0x66, 0x74, 0x6f, 0x76, 0x65, 0x72, 0x20, 0x2b, 0x20, 0x64, 0x65,
|
||||
0x63, 0x6f, 0x64, 0x65, 0x72, 0x2e, 0x64, 0x65, 0x63, 0x6f, 0x64, 0x65,
|
||||
0x28, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x76, 0x61, 0x6c, 0x75,
|
||||
0x65, 0x29, 0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f,
|
||||
0x2f, 0x20, 0x43, 0x68, 0x65, 0x63, 0x6b, 0x20, 0x69, 0x66, 0x20, 0x74,
|
||||
0x68, 0x65, 0x20, 0x6c, 0x61, 0x73, 0x74, 0x20, 0x63, 0x68, 0x61, 0x72,
|
||||
0x61, 0x63, 0x74, 0x65, 0x72, 0x20, 0x69, 0x73, 0x20, 0x61, 0x20, 0x6c,
|
||||
0x69, 0x6e, 0x65, 0x20, 0x62, 0x72, 0x65, 0x61, 0x6b, 0x0a, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x65, 0x6e,
|
||||
0x64, 0x73, 0x57, 0x69, 0x74, 0x68, 0x4c, 0x69, 0x6e, 0x65, 0x42, 0x72,
|
||||
0x65, 0x61, 0x6b, 0x20, 0x3d, 0x20, 0x74, 0x65, 0x78, 0x74, 0x2e, 0x65,
|
||||
0x6e, 0x64, 0x73, 0x57, 0x69, 0x74, 0x68, 0x28, 0x27, 0x5c, 0x6e, 0x27,
|
||||
0x29, 0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f,
|
||||
0x20, 0x53, 0x70, 0x6c, 0x69, 0x74, 0x20, 0x74, 0x68, 0x65, 0x20, 0x74,
|
||||
0x65, 0x78, 0x74, 0x20, 0x69, 0x6e, 0x74, 0x6f, 0x20, 0x6c, 0x69, 0x6e,
|
||||
0x65, 0x73, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x6c, 0x65, 0x74,
|
||||
0x20, 0x6c, 0x69, 0x6e, 0x65, 0x73, 0x20, 0x3d, 0x20, 0x74, 0x65, 0x78,
|
||||
0x74, 0x2e, 0x73, 0x70, 0x6c, 0x69, 0x74, 0x28, 0x27, 0x5c, 0x6e, 0x27,
|
||||
0x29, 0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f,
|
||||
0x20, 0x49, 0x66, 0x20, 0x74, 0x68, 0x65, 0x20, 0x74, 0x65, 0x78, 0x74,
|
||||
0x20, 0x64, 0x6f, 0x65, 0x73, 0x6e, 0x27, 0x74, 0x20, 0x65, 0x6e, 0x64,
|
||||
0x20, 0x77, 0x69, 0x74, 0x68, 0x20, 0x61, 0x20, 0x6c, 0x69, 0x6e, 0x65,
|
||||
0x20, 0x62, 0x72, 0x65, 0x61, 0x6b, 0x2c, 0x20, 0x74, 0x68, 0x65, 0x6e,
|
||||
0x20, 0x74, 0x68, 0x65, 0x20, 0x6c, 0x61, 0x73, 0x74, 0x20, 0x6c, 0x69,
|
||||
0x6e, 0x65, 0x20, 0x69, 0x73, 0x20, 0x69, 0x6e, 0x63, 0x6f, 0x6d, 0x70,
|
||||
0x6c, 0x65, 0x74, 0x65, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f,
|
||||
0x2f, 0x20, 0x53, 0x74, 0x6f, 0x72, 0x65, 0x20, 0x69, 0x74, 0x20, 0x69,
|
||||
0x6e, 0x20, 0x6c, 0x65, 0x66, 0x74, 0x6f, 0x76, 0x65, 0x72, 0x20, 0x74,
|
||||
0x6f, 0x20, 0x62, 0x65, 0x20, 0x61, 0x64, 0x64, 0x65, 0x64, 0x20, 0x74,
|
||||
0x6f, 0x20, 0x74, 0x68, 0x65, 0x20, 0x6e, 0x65, 0x78, 0x74, 0x20, 0x63,
|
||||
0x68, 0x75, 0x6e, 0x6b, 0x20, 0x6f, 0x66, 0x20, 0x64, 0x61, 0x74, 0x61,
|
||||
0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x21,
|
||||
0x65, 0x6e, 0x64, 0x73, 0x57, 0x69, 0x74, 0x68, 0x4c, 0x69, 0x6e, 0x65,
|
||||
0x42, 0x72, 0x65, 0x61, 0x6b, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x6c, 0x65, 0x66, 0x74, 0x6f, 0x76, 0x65,
|
||||
0x72, 0x20, 0x3d, 0x20, 0x6c, 0x69, 0x6e, 0x65, 0x73, 0x2e, 0x70, 0x6f,
|
||||
0x70, 0x28, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d,
|
||||
0x20, 0x65, 0x6c, 0x73, 0x65, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x6c, 0x65, 0x66, 0x74, 0x6f, 0x76, 0x65, 0x72,
|
||||
0x20, 0x3d, 0x20, 0x22, 0x22, 0x3b, 0x20, 0x2f, 0x2f, 0x20, 0x52, 0x65,
|
||||
0x73, 0x65, 0x74, 0x20, 0x6c, 0x65, 0x66, 0x74, 0x6f, 0x76, 0x65, 0x72,
|
||||
0x20, 0x69, 0x66, 0x20, 0x77, 0x65, 0x20, 0x68, 0x61, 0x76, 0x65, 0x20,
|
||||
0x61, 0x20, 0x6c, 0x69, 0x6e, 0x65, 0x20, 0x62, 0x72, 0x65, 0x61, 0x6b,
|
||||
0x20, 0x61, 0x74, 0x20, 0x74, 0x68, 0x65, 0x20, 0x65, 0x6e, 0x64, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x2f, 0x2f, 0x20, 0x50, 0x61, 0x72, 0x73, 0x65, 0x20,
|
||||
0x61, 0x6c, 0x6c, 0x20, 0x73, 0x73, 0x65, 0x20, 0x65, 0x76, 0x65, 0x6e,
|
||||
0x74, 0x73, 0x20, 0x61, 0x6e, 0x64, 0x20, 0x61, 0x64, 0x64, 0x20, 0x74,
|
||||
0x68, 0x65, 0x6d, 0x20, 0x74, 0x6f, 0x20, 0x72, 0x65, 0x73, 0x75, 0x6c,
|
||||
0x74, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73,
|
||||
0x74, 0x20, 0x72, 0x65, 0x67, 0x65, 0x78, 0x20, 0x3d, 0x20, 0x2f, 0x5e,
|
||||
0x28, 0x5c, 0x53, 0x2b, 0x29, 0x3a, 0x5c, 0x73, 0x28, 0x2e, 0x2a, 0x29,
|
||||
0x24, 0x2f, 0x67, 0x6d, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x66, 0x6f, 0x72, 0x20, 0x28, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x6c,
|
||||
0x69, 0x6e, 0x65, 0x20, 0x6f, 0x66, 0x20, 0x6c, 0x69, 0x6e, 0x65, 0x73,
|
||||
0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x6d, 0x61, 0x74, 0x63, 0x68, 0x20,
|
||||
0x3d, 0x20, 0x72, 0x65, 0x67, 0x65, 0x78, 0x2e, 0x65, 0x78, 0x65, 0x63,
|
||||
0x28, 0x6c, 0x69, 0x6e, 0x65, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x6d, 0x61, 0x74, 0x63,
|
||||
0x68, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x5b, 0x6d, 0x61,
|
||||
0x74, 0x63, 0x68, 0x5b, 0x31, 0x5d, 0x5d, 0x20, 0x3d, 0x20, 0x6d, 0x61,
|
||||
0x74, 0x63, 0x68, 0x5b, 0x32, 0x5d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f, 0x20, 0x73, 0x69, 0x6e, 0x63,
|
||||
0x65, 0x20, 0x77, 0x65, 0x20, 0x6b, 0x6e, 0x6f, 0x77, 0x20, 0x74, 0x68,
|
||||
0x69, 0x73, 0x20, 0x69, 0x73, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x2e,
|
||||
0x63, 0x70, 0x70, 0x2c, 0x20, 0x6c, 0x65, 0x74, 0x27, 0x73, 0x20, 0x6a,
|
||||
0x75, 0x73, 0x74, 0x20, 0x64, 0x65, 0x63, 0x6f, 0x64, 0x65, 0x20, 0x74,
|
||||
0x68, 0x65, 0x20, 0x6a, 0x73, 0x6f, 0x6e, 0x20, 0x69, 0x6e, 0x20, 0x64,
|
||||
0x61, 0x74, 0x61, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74,
|
||||
0x2e, 0x64, 0x61, 0x74, 0x61, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x72, 0x65, 0x73,
|
||||
0x75, 0x6c, 0x74, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x20, 0x3d, 0x20, 0x4a,
|
||||
0x53, 0x4f, 0x4e, 0x2e, 0x70, 0x61, 0x72, 0x73, 0x65, 0x28, 0x72, 0x65,
|
||||
0x73, 0x75, 0x6c, 0x74, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x29, 0x3b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x20, 0x2b, 0x3d, 0x20, 0x72,
|
||||
0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x63,
|
||||
0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f, 0x20,
|
||||
0x79, 0x69, 0x65, 0x6c, 0x64, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x79, 0x69, 0x65, 0x6c, 0x64, 0x20,
|
||||
0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x3b, 0x0a, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f, 0x20,
|
||||
0x69, 0x66, 0x20, 0x77, 0x65, 0x20, 0x67, 0x6f, 0x74, 0x20, 0x61, 0x20,
|
||||
0x73, 0x74, 0x6f, 0x70, 0x20, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x20, 0x66,
|
||||
0x72, 0x6f, 0x6d, 0x20, 0x73, 0x65, 0x72, 0x76, 0x65, 0x72, 0x2c, 0x20,
|
||||
0x77, 0x65, 0x20, 0x77, 0x69, 0x6c, 0x6c, 0x20, 0x62, 0x72, 0x65, 0x61,
|
||||
0x6b, 0x20, 0x68, 0x65, 0x72, 0x65, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x72,
|
||||
0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x73,
|
||||
0x74, 0x6f, 0x70, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20,
|
||||
0x28, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x64, 0x61, 0x74, 0x61,
|
||||
0x2e, 0x67, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f,
|
||||
0x73, 0x65, 0x74, 0x74, 0x69, 0x6e, 0x67, 0x73, 0x29, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x67, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x69,
|
||||
0x6f, 0x6e, 0x5f, 0x73, 0x65, 0x74, 0x74, 0x69, 0x6e, 0x67, 0x73, 0x20,
|
||||
0x3d, 0x20, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x64, 0x61, 0x74,
|
||||
0x61, 0x2e, 0x67, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e,
|
||||
0x5f, 0x73, 0x65, 0x74, 0x74, 0x69, 0x6e, 0x67, 0x73, 0x3b, 0x0a, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x74, 0x20, 0x3d, 0x20,
|
||||
0x66, 0x61, 0x6c, 0x73, 0x65, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x62, 0x72, 0x65,
|
||||
0x61, 0x6b, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x72, 0x65, 0x73, 0x75,
|
||||
0x6c, 0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x29, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x74, 0x72, 0x79, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x72, 0x65, 0x73, 0x75,
|
||||
0x6c, 0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x20, 0x3d, 0x20, 0x4a,
|
||||
0x53, 0x4f, 0x4e, 0x2e, 0x70, 0x61, 0x72, 0x73, 0x65, 0x28, 0x72, 0x65,
|
||||
0x73, 0x75, 0x6c, 0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x29, 0x3b,
|
||||
0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x72, 0x65, 0x73, 0x75, 0x6c,
|
||||
0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x2e, 0x6d, 0x65, 0x73, 0x73,
|
||||
0x61, 0x67, 0x65, 0x2e, 0x69, 0x6e, 0x63, 0x6c, 0x75, 0x64, 0x65, 0x73,
|
||||
0x28, 0x27, 0x73, 0x6c, 0x6f, 0x74, 0x20, 0x75, 0x6e, 0x61, 0x76, 0x61,
|
||||
0x69, 0x6c, 0x61, 0x62, 0x6c, 0x65, 0x27, 0x29, 0x29, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x2f, 0x2f, 0x20, 0x54, 0x68, 0x72, 0x6f, 0x77,
|
||||
0x20, 0x61, 0x6e, 0x20, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x20, 0x74, 0x6f,
|
||||
0x20, 0x62, 0x65, 0x20, 0x63, 0x61, 0x75, 0x67, 0x68, 0x74, 0x20, 0x62,
|
||||
0x79, 0x20, 0x75, 0x70, 0x73, 0x74, 0x72, 0x65, 0x61, 0x6d, 0x20, 0x63,
|
||||
0x61, 0x6c, 0x6c, 0x65, 0x72, 0x73, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x74,
|
||||
0x68, 0x72, 0x6f, 0x77, 0x20, 0x6e, 0x65, 0x77, 0x20, 0x45, 0x72, 0x72,
|
||||
0x6f, 0x72, 0x28, 0x27, 0x73, 0x6c, 0x6f, 0x74, 0x20, 0x75, 0x6e, 0x61,
|
||||
0x76, 0x61, 0x69, 0x6c, 0x61, 0x62, 0x6c, 0x65, 0x27, 0x29, 0x3b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x7d, 0x20, 0x65, 0x6c, 0x73, 0x65, 0x20, 0x7b, 0x0a, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x6f, 0x6c, 0x65, 0x2e, 0x65,
|
||||
0x72, 0x72, 0x6f, 0x72, 0x28, 0x60, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x2e,
|
||||
0x63, 0x70, 0x70, 0x20, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x20, 0x5b, 0x24,
|
||||
0x7b, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f,
|
||||
0x72, 0x2e, 0x63, 0x6f, 0x64, 0x65, 0x7d, 0x20, 0x2d, 0x20, 0x24, 0x7b,
|
||||
0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72,
|
||||
0x2e, 0x74, 0x79, 0x70, 0x65, 0x7d, 0x5d, 0x3a, 0x20, 0x24, 0x7b, 0x72,
|
||||
0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x2e,
|
||||
0x6d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x7d, 0x60, 0x29, 0x3b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x7d, 0x20, 0x63, 0x61, 0x74, 0x63, 0x68, 0x28,
|
||||
0x65, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x6f,
|
||||
0x6c, 0x65, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x28, 0x60, 0x6c, 0x6c,
|
||||
0x61, 0x6d, 0x61, 0x2e, 0x63, 0x70, 0x70, 0x20, 0x65, 0x72, 0x72, 0x6f,
|
||||
0x72, 0x20, 0x24, 0x7b, 0x72, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x2e, 0x65,
|
||||
0x72, 0x72, 0x6f, 0x72, 0x7d, 0x60, 0x29, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20,
|
||||
0x7d, 0x20, 0x63, 0x61, 0x74, 0x63, 0x68, 0x20, 0x28, 0x65, 0x29, 0x20,
|
||||
0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x65, 0x2e,
|
||||
0x6e, 0x61, 0x6d, 0x65, 0x20, 0x21, 0x3d, 0x3d, 0x20, 0x27, 0x41, 0x62,
|
||||
0x6f, 0x72, 0x74, 0x45, 0x72, 0x72, 0x6f, 0x72, 0x27, 0x29, 0x20, 0x7b,
|
||||
0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x6f,
|
||||
0x6c, 0x65, 0x2e, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x28, 0x22, 0x6c, 0x6c,
|
||||
0x61, 0x6d, 0x61, 0x20, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x3a, 0x20, 0x22,
|
||||
0x2c, 0x20, 0x65, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x74, 0x68, 0x72, 0x6f, 0x77, 0x20, 0x65, 0x3b,
|
||||
0x0a, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x66, 0x69, 0x6e, 0x61, 0x6c,
|
||||
0x6c, 0x79, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e,
|
||||
0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x2e, 0x61, 0x62, 0x6f, 0x72,
|
||||
0x74, 0x28, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x7d, 0x0a, 0x0a, 0x20, 0x20,
|
||||
0x72, 0x65, 0x74, 0x75, 0x72, 0x6e, 0x20, 0x63, 0x6f, 0x6e, 0x74, 0x65,
|
||||
0x6e, 0x74, 0x3b, 0x0a, 0x7d, 0x0a, 0x0a, 0x2f, 0x2f, 0x20, 0x43, 0x61,
|
||||
0x6c, 0x6c, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x2c, 0x20, 0x72, 0x65,
|
||||
0x74, 0x75, 0x72, 0x6e, 0x20, 0x61, 0x6e, 0x20, 0x65, 0x76, 0x65, 0x6e,
|
||||
0x74, 0x20, 0x74, 0x61, 0x72, 0x67, 0x65, 0x74, 0x20, 0x74, 0x68, 0x61,
|
||||
0x74, 0x20, 0x79, 0x6f, 0x75, 0x20, 0x63, 0x61, 0x6e, 0x20, 0x73, 0x75,
|
||||
0x62, 0x73, 0x63, 0x72, 0x69, 0x62, 0x65, 0x20, 0x74, 0x6f, 0x0a, 0x2f,
|
||||
0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x45, 0x78, 0x61, 0x6d, 0x70, 0x6c, 0x65,
|
||||
0x3a, 0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x69,
|
||||
0x6d, 0x70, 0x6f, 0x72, 0x74, 0x20, 0x7b, 0x20, 0x6c, 0x6c, 0x61, 0x6d,
|
||||
0x61, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x54, 0x61, 0x72, 0x67, 0x65, 0x74,
|
||||
0x20, 0x7d, 0x20, 0x66, 0x72, 0x6f, 0x6d, 0x20, 0x27, 0x2f, 0x63, 0x6f,
|
||||
0x6d, 0x70, 0x6c, 0x65, 0x74, 0x69, 0x6f, 0x6e, 0x2e, 0x6a, 0x73, 0x27,
|
||||
0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f,
|
||||
0x6e, 0x73, 0x74, 0x20, 0x63, 0x6f, 0x6e, 0x6e, 0x20, 0x3d, 0x20, 0x6c,
|
||||
0x6c, 0x61, 0x6d, 0x61, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x54, 0x61, 0x72,
|
||||
0x67, 0x65, 0x74, 0x28, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x29, 0x0a,
|
||||
0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x6e, 0x2e, 0x61,
|
||||
0x64, 0x64, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x4c, 0x69, 0x73, 0x74, 0x65,
|
||||
0x6e, 0x65, 0x72, 0x28, 0x22, 0x6d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65,
|
||||
0x22, 0x2c, 0x20, 0x28, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x29, 0x20, 0x3d,
|
||||
0x3e, 0x20, 0x7b, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x64, 0x6f, 0x63, 0x75, 0x6d, 0x65, 0x6e, 0x74, 0x2e, 0x77, 0x72, 0x69,
|
||||
0x74, 0x65, 0x28, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x65, 0x74,
|
||||
0x61, 0x69, 0x6c, 0x2e, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x29,
|
||||
0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x29, 0x0a, 0x2f, 0x2f,
|
||||
0x0a, 0x65, 0x78, 0x70, 0x6f, 0x72, 0x74, 0x20, 0x63, 0x6f, 0x6e, 0x73,
|
||||
0x74, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x45, 0x76, 0x65, 0x6e, 0x74,
|
||||
0x54, 0x61, 0x72, 0x67, 0x65, 0x74, 0x20, 0x3d, 0x20, 0x28, 0x70, 0x72,
|
||||
0x6f, 0x6d, 0x70, 0x74, 0x2c, 0x20, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x73,
|
||||
0x20, 0x3d, 0x20, 0x7b, 0x7d, 0x2c, 0x20, 0x63, 0x6f, 0x6e, 0x66, 0x69,
|
||||
0x67, 0x20, 0x3d, 0x20, 0x7b, 0x7d, 0x29, 0x20, 0x3d, 0x3e, 0x20, 0x7b,
|
||||
0x0a, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x65, 0x76, 0x65,
|
||||
0x6e, 0x74, 0x54, 0x61, 0x72, 0x67, 0x65, 0x74, 0x20, 0x3d, 0x20, 0x6e,
|
||||
0x65, 0x77, 0x20, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x54, 0x61, 0x72, 0x67,
|
||||
0x65, 0x74, 0x28, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x28, 0x61, 0x73, 0x79,
|
||||
0x6e, 0x63, 0x20, 0x28, 0x29, 0x20, 0x3d, 0x3e, 0x20, 0x7b, 0x0a, 0x20,
|
||||
0x20, 0x20, 0x20, 0x6c, 0x65, 0x74, 0x20, 0x63, 0x6f, 0x6e, 0x74, 0x65,
|
||||
0x6e, 0x74, 0x20, 0x3d, 0x20, 0x22, 0x22, 0x3b, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x66, 0x6f, 0x72, 0x20, 0x61, 0x77, 0x61, 0x69, 0x74, 0x20, 0x28,
|
||||
0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x20,
|
||||
0x6f, 0x66, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x28, 0x70, 0x72, 0x6f,
|
||||
0x6d, 0x70, 0x74, 0x2c, 0x20, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x73, 0x2c,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x29, 0x29, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x63, 0x68,
|
||||
0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x29, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x74,
|
||||
0x65, 0x6e, 0x74, 0x20, 0x2b, 0x3d, 0x20, 0x63, 0x68, 0x75, 0x6e, 0x6b,
|
||||
0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e,
|
||||
0x74, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x65,
|
||||
0x76, 0x65, 0x6e, 0x74, 0x54, 0x61, 0x72, 0x67, 0x65, 0x74, 0x2e, 0x64,
|
||||
0x69, 0x73, 0x70, 0x61, 0x74, 0x63, 0x68, 0x45, 0x76, 0x65, 0x6e, 0x74,
|
||||
0x28, 0x6e, 0x65, 0x77, 0x20, 0x43, 0x75, 0x73, 0x74, 0x6f, 0x6d, 0x45,
|
||||
0x76, 0x65, 0x6e, 0x74, 0x28, 0x22, 0x6d, 0x65, 0x73, 0x73, 0x61, 0x67,
|
||||
0x65, 0x22, 0x2c, 0x20, 0x7b, 0x20, 0x64, 0x65, 0x74, 0x61, 0x69, 0x6c,
|
||||
0x3a, 0x20, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74, 0x61,
|
||||
0x20, 0x7d, 0x29, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28,
|
||||
0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x67,
|
||||
0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73, 0x65,
|
||||
0x74, 0x74, 0x69, 0x6e, 0x67, 0x73, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x54,
|
||||
0x61, 0x72, 0x67, 0x65, 0x74, 0x2e, 0x64, 0x69, 0x73, 0x70, 0x61, 0x74,
|
||||
0x63, 0x68, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x28, 0x6e, 0x65, 0x77, 0x20,
|
||||
0x43, 0x75, 0x73, 0x74, 0x6f, 0x6d, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x28,
|
||||
0x22, 0x67, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f,
|
||||
0x73, 0x65, 0x74, 0x74, 0x69, 0x6e, 0x67, 0x73, 0x22, 0x2c, 0x20, 0x7b,
|
||||
0x20, 0x64, 0x65, 0x74, 0x61, 0x69, 0x6c, 0x3a, 0x20, 0x63, 0x68, 0x75,
|
||||
0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x67, 0x65, 0x6e, 0x65,
|
||||
0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73, 0x65, 0x74, 0x74, 0x69,
|
||||
0x6e, 0x67, 0x73, 0x20, 0x7d, 0x29, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x69,
|
||||
0x66, 0x20, 0x28, 0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74,
|
||||
0x61, 0x2e, 0x74, 0x69, 0x6d, 0x69, 0x6e, 0x67, 0x73, 0x29, 0x20, 0x7b,
|
||||
0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x65, 0x76, 0x65,
|
||||
0x6e, 0x74, 0x54, 0x61, 0x72, 0x67, 0x65, 0x74, 0x2e, 0x64, 0x69, 0x73,
|
||||
0x70, 0x61, 0x74, 0x63, 0x68, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x28, 0x6e,
|
||||
0x65, 0x77, 0x20, 0x43, 0x75, 0x73, 0x74, 0x6f, 0x6d, 0x45, 0x76, 0x65,
|
||||
0x6e, 0x74, 0x28, 0x22, 0x74, 0x69, 0x6d, 0x69, 0x6e, 0x67, 0x73, 0x22,
|
||||
0x2c, 0x20, 0x7b, 0x20, 0x64, 0x65, 0x74, 0x61, 0x69, 0x6c, 0x3a, 0x20,
|
||||
0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x74,
|
||||
0x69, 0x6d, 0x69, 0x6e, 0x67, 0x73, 0x20, 0x7d, 0x29, 0x29, 0x3b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x54,
|
||||
0x61, 0x72, 0x67, 0x65, 0x74, 0x2e, 0x64, 0x69, 0x73, 0x70, 0x61, 0x74,
|
||||
0x63, 0x68, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x28, 0x6e, 0x65, 0x77, 0x20,
|
||||
0x43, 0x75, 0x73, 0x74, 0x6f, 0x6d, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x28,
|
||||
0x22, 0x64, 0x6f, 0x6e, 0x65, 0x22, 0x2c, 0x20, 0x7b, 0x20, 0x64, 0x65,
|
||||
0x74, 0x61, 0x69, 0x6c, 0x3a, 0x20, 0x7b, 0x20, 0x63, 0x6f, 0x6e, 0x74,
|
||||
0x65, 0x6e, 0x74, 0x20, 0x7d, 0x20, 0x7d, 0x29, 0x29, 0x3b, 0x0a, 0x20,
|
||||
0x20, 0x7d, 0x29, 0x28, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x72, 0x65, 0x74,
|
||||
0x75, 0x72, 0x6e, 0x20, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x54, 0x61, 0x72,
|
||||
0x67, 0x65, 0x74, 0x3b, 0x0a, 0x7d, 0x0a, 0x0a, 0x2f, 0x2f, 0x20, 0x43,
|
||||
0x61, 0x6c, 0x6c, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x2c, 0x20, 0x72,
|
||||
0x65, 0x74, 0x75, 0x72, 0x6e, 0x20, 0x61, 0x20, 0x70, 0x72, 0x6f, 0x6d,
|
||||
0x69, 0x73, 0x65, 0x20, 0x74, 0x68, 0x61, 0x74, 0x20, 0x72, 0x65, 0x73,
|
||||
0x6f, 0x6c, 0x76, 0x65, 0x73, 0x20, 0x74, 0x6f, 0x20, 0x74, 0x68, 0x65,
|
||||
0x20, 0x63, 0x6f, 0x6d, 0x70, 0x6c, 0x65, 0x74, 0x65, 0x64, 0x20, 0x74,
|
||||
0x65, 0x78, 0x74, 0x2e, 0x20, 0x54, 0x68, 0x69, 0x73, 0x20, 0x64, 0x6f,
|
||||
0x65, 0x73, 0x20, 0x6e, 0x6f, 0x74, 0x20, 0x73, 0x75, 0x70, 0x70, 0x6f,
|
||||
0x72, 0x74, 0x20, 0x73, 0x74, 0x72, 0x65, 0x61, 0x6d, 0x69, 0x6e, 0x67,
|
||||
0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x45, 0x78, 0x61, 0x6d, 0x70,
|
||||
0x6c, 0x65, 0x3a, 0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x50, 0x72, 0x6f, 0x6d, 0x69,
|
||||
0x73, 0x65, 0x28, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x29, 0x2e, 0x74,
|
||||
0x68, 0x65, 0x6e, 0x28, 0x28, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74,
|
||||
0x29, 0x20, 0x3d, 0x3e, 0x20, 0x7b, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x20, 0x64, 0x6f, 0x63, 0x75, 0x6d, 0x65, 0x6e, 0x74,
|
||||
0x2e, 0x77, 0x72, 0x69, 0x74, 0x65, 0x28, 0x63, 0x6f, 0x6e, 0x74, 0x65,
|
||||
0x6e, 0x74, 0x29, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x20, 0x7d,
|
||||
0x29, 0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x6f, 0x72, 0x0a, 0x2f, 0x2f, 0x0a, 0x2f, 0x2f, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x63, 0x6f, 0x6e, 0x74, 0x65,
|
||||
0x6e, 0x74, 0x20, 0x3d, 0x20, 0x61, 0x77, 0x61, 0x69, 0x74, 0x20, 0x6c,
|
||||
0x6c, 0x61, 0x6d, 0x61, 0x50, 0x72, 0x6f, 0x6d, 0x69, 0x73, 0x65, 0x28,
|
||||
0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x29, 0x0a, 0x2f, 0x2f, 0x20, 0x20,
|
||||
0x20, 0x20, 0x20, 0x64, 0x6f, 0x63, 0x75, 0x6d, 0x65, 0x6e, 0x74, 0x2e,
|
||||
0x77, 0x72, 0x69, 0x74, 0x65, 0x28, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e,
|
||||
0x74, 0x29, 0x0a, 0x2f, 0x2f, 0x0a, 0x65, 0x78, 0x70, 0x6f, 0x72, 0x74,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61,
|
||||
0x50, 0x72, 0x6f, 0x6d, 0x69, 0x73, 0x65, 0x20, 0x3d, 0x20, 0x28, 0x70,
|
||||
0x72, 0x6f, 0x6d, 0x70, 0x74, 0x2c, 0x20, 0x70, 0x61, 0x72, 0x61, 0x6d,
|
||||
0x73, 0x20, 0x3d, 0x20, 0x7b, 0x7d, 0x2c, 0x20, 0x63, 0x6f, 0x6e, 0x66,
|
||||
0x69, 0x67, 0x20, 0x3d, 0x20, 0x7b, 0x7d, 0x29, 0x20, 0x3d, 0x3e, 0x20,
|
||||
0x7b, 0x0a, 0x20, 0x20, 0x72, 0x65, 0x74, 0x75, 0x72, 0x6e, 0x20, 0x6e,
|
||||
0x65, 0x77, 0x20, 0x50, 0x72, 0x6f, 0x6d, 0x69, 0x73, 0x65, 0x28, 0x61,
|
||||
0x73, 0x79, 0x6e, 0x63, 0x20, 0x28, 0x72, 0x65, 0x73, 0x6f, 0x6c, 0x76,
|
||||
0x65, 0x2c, 0x20, 0x72, 0x65, 0x6a, 0x65, 0x63, 0x74, 0x29, 0x20, 0x3d,
|
||||
0x3e, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x6c, 0x65, 0x74, 0x20,
|
||||
0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x20, 0x3d, 0x20, 0x22, 0x22,
|
||||
0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x74, 0x72, 0x79, 0x20, 0x7b, 0x0a,
|
||||
0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x66, 0x6f, 0x72, 0x20, 0x61, 0x77,
|
||||
0x61, 0x69, 0x74, 0x20, 0x28, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x63,
|
||||
0x68, 0x75, 0x6e, 0x6b, 0x20, 0x6f, 0x66, 0x20, 0x6c, 0x6c, 0x61, 0x6d,
|
||||
0x61, 0x28, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x2c, 0x20, 0x70, 0x61,
|
||||
0x72, 0x61, 0x6d, 0x73, 0x2c, 0x20, 0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67,
|
||||
0x29, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x20, 0x2b, 0x3d, 0x20,
|
||||
0x63, 0x68, 0x75, 0x6e, 0x6b, 0x2e, 0x64, 0x61, 0x74, 0x61, 0x2e, 0x63,
|
||||
0x6f, 0x6e, 0x74, 0x65, 0x6e, 0x74, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20,
|
||||
0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x72, 0x65,
|
||||
0x73, 0x6f, 0x6c, 0x76, 0x65, 0x28, 0x63, 0x6f, 0x6e, 0x74, 0x65, 0x6e,
|
||||
0x74, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x7d, 0x20, 0x63, 0x61,
|
||||
0x74, 0x63, 0x68, 0x20, 0x28, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x29, 0x20,
|
||||
0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x72, 0x65, 0x6a, 0x65,
|
||||
0x63, 0x74, 0x28, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x29, 0x3b, 0x0a, 0x20,
|
||||
0x20, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x7d, 0x29, 0x3b, 0x0a, 0x7d,
|
||||
0x3b, 0x0a, 0x0a, 0x2f, 0x2a, 0x2a, 0x0a, 0x20, 0x2a, 0x20, 0x28, 0x64,
|
||||
0x65, 0x70, 0x72, 0x65, 0x63, 0x61, 0x74, 0x65, 0x64, 0x29, 0x0a, 0x20,
|
||||
0x2a, 0x2f, 0x0a, 0x65, 0x78, 0x70, 0x6f, 0x72, 0x74, 0x20, 0x63, 0x6f,
|
||||
0x6e, 0x73, 0x74, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x43, 0x6f, 0x6d,
|
||||
0x70, 0x6c, 0x65, 0x74, 0x65, 0x20, 0x3d, 0x20, 0x61, 0x73, 0x79, 0x6e,
|
||||
0x63, 0x20, 0x28, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x73, 0x2c, 0x20, 0x63,
|
||||
0x6f, 0x6e, 0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x2c, 0x20, 0x63,
|
||||
0x61, 0x6c, 0x6c, 0x62, 0x61, 0x63, 0x6b, 0x29, 0x20, 0x3d, 0x3e, 0x20,
|
||||
0x7b, 0x0a, 0x20, 0x20, 0x66, 0x6f, 0x72, 0x20, 0x61, 0x77, 0x61, 0x69,
|
||||
0x74, 0x20, 0x28, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x63, 0x68, 0x75,
|
||||
0x6e, 0x6b, 0x20, 0x6f, 0x66, 0x20, 0x6c, 0x6c, 0x61, 0x6d, 0x61, 0x28,
|
||||
0x70, 0x61, 0x72, 0x61, 0x6d, 0x73, 0x2e, 0x70, 0x72, 0x6f, 0x6d, 0x70,
|
||||
0x74, 0x2c, 0x20, 0x70, 0x61, 0x72, 0x61, 0x6d, 0x73, 0x2c, 0x20, 0x7b,
|
||||
0x20, 0x63, 0x6f, 0x6e, 0x74, 0x72, 0x6f, 0x6c, 0x6c, 0x65, 0x72, 0x20,
|
||||
0x7d, 0x29, 0x29, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x63, 0x61,
|
||||
0x6c, 0x6c, 0x62, 0x61, 0x63, 0x6b, 0x28, 0x63, 0x68, 0x75, 0x6e, 0x6b,
|
||||
0x29, 0x3b, 0x0a, 0x20, 0x20, 0x7d, 0x0a, 0x7d, 0x0a, 0x0a, 0x2f, 0x2f,
|
||||
0x20, 0x47, 0x65, 0x74, 0x20, 0x74, 0x68, 0x65, 0x20, 0x6d, 0x6f, 0x64,
|
||||
0x65, 0x6c, 0x20, 0x69, 0x6e, 0x66, 0x6f, 0x20, 0x66, 0x72, 0x6f, 0x6d,
|
||||
0x20, 0x74, 0x68, 0x65, 0x20, 0x73, 0x65, 0x72, 0x76, 0x65, 0x72, 0x2e,
|
||||
0x20, 0x54, 0x68, 0x69, 0x73, 0x20, 0x69, 0x73, 0x20, 0x75, 0x73, 0x65,
|
||||
0x66, 0x75, 0x6c, 0x20, 0x66, 0x6f, 0x72, 0x20, 0x67, 0x65, 0x74, 0x74,
|
||||
0x69, 0x6e, 0x67, 0x20, 0x74, 0x68, 0x65, 0x20, 0x63, 0x6f, 0x6e, 0x74,
|
||||
0x65, 0x78, 0x74, 0x20, 0x77, 0x69, 0x6e, 0x64, 0x6f, 0x77, 0x20, 0x61,
|
||||
0x6e, 0x64, 0x20, 0x73, 0x6f, 0x20, 0x6f, 0x6e, 0x2e, 0x0a, 0x65, 0x78,
|
||||
0x70, 0x6f, 0x72, 0x74, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x6c,
|
||||
0x6c, 0x61, 0x6d, 0x61, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x49, 0x6e, 0x66,
|
||||
0x6f, 0x20, 0x3d, 0x20, 0x61, 0x73, 0x79, 0x6e, 0x63, 0x20, 0x28, 0x63,
|
||||
0x6f, 0x6e, 0x66, 0x69, 0x67, 0x20, 0x3d, 0x20, 0x7b, 0x7d, 0x29, 0x20,
|
||||
0x3d, 0x3e, 0x20, 0x7b, 0x0a, 0x20, 0x20, 0x69, 0x66, 0x20, 0x28, 0x21,
|
||||
0x67, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73,
|
||||
0x65, 0x74, 0x74, 0x69, 0x6e, 0x67, 0x73, 0x29, 0x20, 0x7b, 0x0a, 0x20,
|
||||
0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x74, 0x20, 0x61, 0x70, 0x69,
|
||||
0x5f, 0x75, 0x72, 0x6c, 0x20, 0x3d, 0x20, 0x63, 0x6f, 0x6e, 0x66, 0x69,
|
||||
0x67, 0x2e, 0x61, 0x70, 0x69, 0x5f, 0x75, 0x72, 0x6c, 0x20, 0x7c, 0x7c,
|
||||
0x20, 0x22, 0x22, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x63, 0x6f, 0x6e,
|
||||
0x73, 0x74, 0x20, 0x70, 0x72, 0x6f, 0x70, 0x73, 0x20, 0x3d, 0x20, 0x61,
|
||||
0x77, 0x61, 0x69, 0x74, 0x20, 0x66, 0x65, 0x74, 0x63, 0x68, 0x28, 0x60,
|
||||
0x24, 0x7b, 0x61, 0x70, 0x69, 0x5f, 0x75, 0x72, 0x6c, 0x7d, 0x2f, 0x70,
|
||||
0x72, 0x6f, 0x70, 0x73, 0x60, 0x29, 0x2e, 0x74, 0x68, 0x65, 0x6e, 0x28,
|
||||
0x72, 0x20, 0x3d, 0x3e, 0x20, 0x72, 0x2e, 0x6a, 0x73, 0x6f, 0x6e, 0x28,
|
||||
0x29, 0x29, 0x3b, 0x0a, 0x20, 0x20, 0x20, 0x20, 0x67, 0x65, 0x6e, 0x65,
|
||||
0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73, 0x65, 0x74, 0x74, 0x69,
|
||||
0x6e, 0x67, 0x73, 0x20, 0x3d, 0x20, 0x70, 0x72, 0x6f, 0x70, 0x73, 0x2e,
|
||||
0x64, 0x65, 0x66, 0x61, 0x75, 0x6c, 0x74, 0x5f, 0x67, 0x65, 0x6e, 0x65,
|
||||
0x72, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73, 0x65, 0x74, 0x74, 0x69,
|
||||
0x6e, 0x67, 0x73, 0x3b, 0x0a, 0x20, 0x20, 0x7d, 0x0a, 0x20, 0x20, 0x72,
|
||||
0x65, 0x74, 0x75, 0x72, 0x6e, 0x20, 0x67, 0x65, 0x6e, 0x65, 0x72, 0x61,
|
||||
0x74, 0x69, 0x6f, 0x6e, 0x5f, 0x73, 0x65, 0x74, 0x74, 0x69, 0x6e, 0x67,
|
||||
0x73, 0x3b, 0x0a, 0x7d, 0x0a
|
||||
};
|
||||
unsigned int completion_js_len = 5909;
|
||||
|
|
@ -8,13 +8,3 @@ PUBLIC=$DIR/public
|
|||
echo "download js bundle files"
|
||||
curl https://npm.reversehttp.com/@preact/signals-core,@preact/signals,htm/preact,preact,preact/hooks > $PUBLIC/index.js
|
||||
echo >> $PUBLIC/index.js # add newline
|
||||
|
||||
FILES=$(ls $PUBLIC)
|
||||
|
||||
cd $PUBLIC
|
||||
for FILE in $FILES; do
|
||||
echo "generate $FILE.hpp"
|
||||
|
||||
# use simple flag for old version of xxd
|
||||
xxd -i $FILE > $DIR/$FILE.hpp
|
||||
done
|
||||
|
|
|
|||
File diff suppressed because it is too large
Load diff
File diff suppressed because it is too large
Load diff
File diff suppressed because it is too large
Load diff
|
|
@ -881,11 +881,11 @@
|
|||
.replace(/&/g, '&')
|
||||
.replace(/</g, '<')
|
||||
.replace(/>/g, '>')
|
||||
.replace(/^#{1,6} (.*)$/gim, '<h3>$1</h3>')
|
||||
.replace(/\*\*(.*?)\*\*/g, '<strong>$1</strong>')
|
||||
.replace(/__(.*?)__/g, '<strong>$1</strong>')
|
||||
.replace(/\*(.*?)\*/g, '<em>$1</em>')
|
||||
.replace(/_(.*?)_/g, '<em>$1</em>')
|
||||
.replace(/(^|\n)#{1,6} ([^\n]*)(?=([^`]*`[^`]*`)*[^`]*$)/g, '$1<h3>$2</h3>')
|
||||
.replace(/\*\*(.*?)\*\*(?=([^`]*`[^`]*`)*[^`]*$)/g, '<strong>$1</strong>')
|
||||
.replace(/__(.*?)__(?=([^`]*`[^`]*`)*[^`]*$)/g, '<strong>$1</strong>')
|
||||
.replace(/\*(.*?)\*(?=([^`]*`[^`]*`)*[^`]*$)/g, '<em>$1</em>')
|
||||
.replace(/_(.*?)_(?=([^`]*`[^`]*`)*[^`]*$)/g, '<em>$1</em>')
|
||||
.replace(/```.*?\n([\s\S]*?)```/g, '<pre><code>$1</code></pre>')
|
||||
.replace(/`(.*?)`/g, '<code>$1</code>')
|
||||
.replace(/\n/gim, '<br />');
|
||||
|
|
|
|||
|
|
@ -854,7 +854,7 @@ struct server_context {
|
|||
slot.sparams.penalize_nl = json_value(data, "penalize_nl", default_sparams.penalize_nl);
|
||||
slot.params.n_keep = json_value(data, "n_keep", slot.params.n_keep);
|
||||
slot.params.n_discard = json_value(data, "n_discard", default_params.n_discard);
|
||||
slot.params.seed = json_value(data, "seed", default_params.seed);
|
||||
slot.sparams.seed = json_value(data, "seed", default_sparams.seed);
|
||||
slot.sparams.n_probs = json_value(data, "n_probs", default_sparams.n_probs);
|
||||
slot.sparams.min_keep = json_value(data, "min_keep", default_sparams.min_keep);
|
||||
|
||||
|
|
@ -1028,7 +1028,6 @@ struct server_context {
|
|||
send_error(task, "Failed to parse grammar", ERROR_TYPE_INVALID_REQUEST);
|
||||
return false;
|
||||
}
|
||||
llama_set_rng_seed(ctx, slot.params.seed);
|
||||
}
|
||||
|
||||
slot.command = SLOT_COMMAND_LOAD_PROMPT;
|
||||
|
|
@ -1118,7 +1117,7 @@ struct server_context {
|
|||
|
||||
bool process_token(completion_token_output & result, server_slot & slot) {
|
||||
// remember which tokens were sampled - used for repetition penalties during sampling
|
||||
const std::string token_str = llama_token_to_piece(ctx, result.tok);
|
||||
const std::string token_str = llama_token_to_piece(ctx, result.tok, false);
|
||||
slot.sampled = result.tok;
|
||||
|
||||
// search stop word and delete it
|
||||
|
|
@ -1201,7 +1200,7 @@ struct server_context {
|
|||
});
|
||||
}
|
||||
|
||||
if (result.tok == llama_token_eos(model)) {
|
||||
if (llama_token_is_eog(model, result.tok)) {
|
||||
slot.stopped_eos = true;
|
||||
slot.has_next_token = false;
|
||||
|
||||
|
|
|
|||
|
|
@ -29,7 +29,7 @@ To mitigate it, you can increase values in `n_predict`, `kv_size`.
|
|||
cd ../../..
|
||||
mkdir build
|
||||
cd build
|
||||
cmake ../
|
||||
cmake -DLLAMA_CURL=ON ../
|
||||
cmake --build . --target server
|
||||
```
|
||||
|
||||
|
|
|
|||
57
examples/server/tests/features/results.feature
Normal file
57
examples/server/tests/features/results.feature
Normal file
|
|
@ -0,0 +1,57 @@
|
|||
@llama.cpp
|
||||
@results
|
||||
Feature: Results
|
||||
|
||||
Background: Server startup
|
||||
Given a server listening on localhost:8080
|
||||
And a model file tinyllamas/split/stories15M-00001-of-00003.gguf from HF repo ggml-org/models
|
||||
And a model file test-model-00001-of-00003.gguf
|
||||
And 128 as batch size
|
||||
And 256 KV cache size
|
||||
And 128 max tokens to predict
|
||||
|
||||
Scenario Outline: Multi users completion
|
||||
Given <n_slots> slots
|
||||
And continuous batching
|
||||
Then the server is starting
|
||||
Then the server is healthy
|
||||
|
||||
Given 42 as seed
|
||||
And a prompt:
|
||||
"""
|
||||
Write a very long story about AI.
|
||||
"""
|
||||
|
||||
Given 42 as seed
|
||||
And a prompt:
|
||||
"""
|
||||
Write a very long story about AI.
|
||||
"""
|
||||
|
||||
Given 42 as seed
|
||||
And a prompt:
|
||||
"""
|
||||
Write a very long story about AI.
|
||||
"""
|
||||
|
||||
Given 42 as seed
|
||||
And a prompt:
|
||||
"""
|
||||
Write a very long story about AI.
|
||||
"""
|
||||
|
||||
Given 42 as seed
|
||||
And a prompt:
|
||||
"""
|
||||
Write a very long story about AI.
|
||||
"""
|
||||
|
||||
Given concurrent completion requests
|
||||
Then the server is busy
|
||||
Then the server is idle
|
||||
And all slots are idle
|
||||
Then all predictions are equal
|
||||
Examples:
|
||||
| n_slots |
|
||||
| 1 |
|
||||
| 2 |
|
||||
|
|
@ -61,6 +61,7 @@ def step_server_config(context, server_fqdn, server_port):
|
|||
context.server_metrics = False
|
||||
context.server_process = None
|
||||
context.seed = None
|
||||
context.draft = None
|
||||
context.server_seed = None
|
||||
context.user_api_key = None
|
||||
context.response_format = None
|
||||
|
|
@ -107,6 +108,11 @@ def step_n_gpu_layer(context, ngl):
|
|||
context.n_gpu_layer = ngl
|
||||
|
||||
|
||||
@step('{draft:d} as draft')
|
||||
def step_draft(context, draft):
|
||||
context.draft = draft
|
||||
|
||||
|
||||
@step('{n_ctx:d} KV cache size')
|
||||
def step_n_ctx(context, n_ctx):
|
||||
context.n_ctx = n_ctx
|
||||
|
|
@ -254,6 +260,15 @@ def step_n_tokens_predicted(context, predicted_n):
|
|||
assert_n_tokens_predicted(context.completion, predicted_n)
|
||||
|
||||
|
||||
@step('all predictions are equal')
|
||||
@async_run_until_complete
|
||||
async def step_predictions_equal(context):
|
||||
n_completions = await gather_tasks_results(context)
|
||||
assert n_completions >= 2, "need at least 2 completions"
|
||||
assert_all_predictions_equal(context.tasks_result)
|
||||
context.tasks_result = []
|
||||
|
||||
|
||||
@step('the completion is truncated')
|
||||
def step_assert_completion_truncated(context):
|
||||
step_assert_completion_truncated(context, '')
|
||||
|
|
@ -1020,6 +1035,23 @@ def assert_n_tokens_predicted(completion_response, expected_predicted_n=None, re
|
|||
assert n_predicted == expected_predicted_n, (f'invalid number of tokens predicted:'
|
||||
f' {n_predicted} <> {expected_predicted_n}')
|
||||
|
||||
def assert_all_predictions_equal(completion_responses):
|
||||
content_0 = completion_responses[0]['content']
|
||||
|
||||
if 'DEBUG' in os.environ and os.environ['DEBUG'] == 'ON':
|
||||
print(f"content 0: {content_0}")
|
||||
|
||||
i = 1
|
||||
for response in completion_responses[1:]:
|
||||
content = response['content']
|
||||
|
||||
if 'DEBUG' in os.environ and os.environ['DEBUG'] == 'ON':
|
||||
print(f"content {i}: {content}")
|
||||
|
||||
assert content == content_0, "contents not equal"
|
||||
|
||||
i += 1
|
||||
|
||||
|
||||
async def gather_tasks_results(context):
|
||||
n_tasks = len(context.concurrent_tasks)
|
||||
|
|
@ -1148,6 +1180,8 @@ def start_server_background(context):
|
|||
server_args.extend(['--ubatch-size', context.n_ubatch])
|
||||
if context.n_gpu_layer:
|
||||
server_args.extend(['--n-gpu-layers', context.n_gpu_layer])
|
||||
if context.draft is not None:
|
||||
server_args.extend(['--draft', context.draft])
|
||||
if context.server_continuous_batching:
|
||||
server_args.append('--cont-batching')
|
||||
if context.server_embeddings:
|
||||
|
|
|
|||
|
|
@ -381,10 +381,6 @@ static json oaicompat_completion_params_parse(
|
|||
} else {
|
||||
llama_params["stop"] = json_value(body, "stop", json::array());
|
||||
}
|
||||
// Some chat templates don't use EOS token to stop generation
|
||||
// We must add their end sequences to list of stop words
|
||||
llama_params["stop"].push_back("<|im_end|>"); // chatml
|
||||
llama_params["stop"].push_back("<end_of_turn>"); // gemma
|
||||
|
||||
// Handle "response_format" field
|
||||
if (body.contains("response_format")) {
|
||||
|
|
|
|||
|
|
@ -133,8 +133,8 @@ int main(int argc, char ** argv) {
|
|||
// sample the most likely token
|
||||
const llama_token new_token_id = llama_sample_token_greedy(ctx, &candidates_p);
|
||||
|
||||
// is it an end of stream?
|
||||
if (new_token_id == llama_token_eos(model) || n_cur == n_len) {
|
||||
// is it an end of generation?
|
||||
if (llama_token_is_eog(model, new_token_id) || n_cur == n_len) {
|
||||
LOG_TEE("\n");
|
||||
|
||||
break;
|
||||
|
|
|
|||
|
|
@ -360,7 +360,7 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
}
|
||||
|
||||
if (token_id == llama_token_eos(model_tgt)) {
|
||||
if (llama_token_is_eog(model_tgt, token_id)) {
|
||||
has_eos = true;
|
||||
}
|
||||
++n_predict;
|
||||
|
|
|
|||
6
flake.lock
generated
6
flake.lock
generated
|
|
@ -20,11 +20,11 @@
|
|||
},
|
||||
"nixpkgs": {
|
||||
"locked": {
|
||||
"lastModified": 1712791164,
|
||||
"narHash": "sha256-3sbWO1mbpWsLepZGbWaMovSO7ndZeFqDSdX0hZ9nVyw=",
|
||||
"lastModified": 1713537308,
|
||||
"narHash": "sha256-XtTSSIB2DA6tOv+l0FhvfDMiyCmhoRbNB+0SeInZkbk=",
|
||||
"owner": "NixOS",
|
||||
"repo": "nixpkgs",
|
||||
"rev": "1042fd8b148a9105f3c0aca3a6177fd1d9360ba5",
|
||||
"rev": "5c24cf2f0a12ad855f444c30b2421d044120c66f",
|
||||
"type": "github"
|
||||
},
|
||||
"original": {
|
||||
|
|
|
|||
16
ggml-alloc.c
16
ggml-alloc.c
|
|
@ -371,16 +371,16 @@ struct ggml_gallocr {
|
|||
};
|
||||
|
||||
ggml_gallocr_t ggml_gallocr_new_n(ggml_backend_buffer_type_t * bufts, int n_bufs) {
|
||||
ggml_gallocr_t galloc = (ggml_gallocr_t)calloc(sizeof(struct ggml_gallocr), 1);
|
||||
ggml_gallocr_t galloc = (ggml_gallocr_t)calloc(1, sizeof(struct ggml_gallocr));
|
||||
GGML_ASSERT(galloc != NULL);
|
||||
|
||||
galloc->bufts = calloc(sizeof(ggml_backend_buffer_type_t) * n_bufs, 1);
|
||||
galloc->bufts = calloc(n_bufs, sizeof(ggml_backend_buffer_type_t));
|
||||
GGML_ASSERT(galloc->bufts != NULL);
|
||||
|
||||
galloc->buffers = calloc(sizeof(ggml_backend_buffer_t) * n_bufs, 1);
|
||||
galloc->buffers = calloc(n_bufs, sizeof(ggml_backend_buffer_t) * n_bufs);
|
||||
GGML_ASSERT(galloc->buffers != NULL);
|
||||
|
||||
galloc->buf_tallocs = calloc(sizeof(struct ggml_dyn_tallocr *) * n_bufs, 1);
|
||||
galloc->buf_tallocs = calloc(n_bufs, sizeof(struct ggml_dyn_tallocr *));
|
||||
GGML_ASSERT(galloc->buf_tallocs != NULL);
|
||||
|
||||
for (int i = 0; i < n_bufs; i++) {
|
||||
|
|
@ -646,8 +646,8 @@ bool ggml_gallocr_reserve_n(ggml_gallocr_t galloc, struct ggml_cgraph * graph, c
|
|||
free(galloc->hash_set.keys);
|
||||
free(galloc->hash_values);
|
||||
galloc->hash_set.size = hash_size;
|
||||
galloc->hash_set.keys = calloc(sizeof(struct ggml_tensor *), hash_size);
|
||||
galloc->hash_values = calloc(sizeof(struct hash_node), hash_size);
|
||||
galloc->hash_set.keys = calloc(hash_size, sizeof(struct ggml_tensor *));
|
||||
galloc->hash_values = calloc(hash_size, sizeof(struct hash_node));
|
||||
GGML_ASSERT(galloc->hash_set.keys != NULL);
|
||||
GGML_ASSERT(galloc->hash_values != NULL);
|
||||
} else {
|
||||
|
|
@ -667,7 +667,7 @@ bool ggml_gallocr_reserve_n(ggml_gallocr_t galloc, struct ggml_cgraph * graph, c
|
|||
// set the node_allocs from the hash table
|
||||
if (galloc->n_nodes < graph->n_nodes) {
|
||||
free(galloc->node_allocs);
|
||||
galloc->node_allocs = calloc(sizeof(struct node_alloc), graph->n_nodes);
|
||||
galloc->node_allocs = calloc(graph->n_nodes, sizeof(struct node_alloc));
|
||||
GGML_ASSERT(galloc->node_allocs != NULL);
|
||||
}
|
||||
galloc->n_nodes = graph->n_nodes;
|
||||
|
|
@ -697,7 +697,7 @@ bool ggml_gallocr_reserve_n(ggml_gallocr_t galloc, struct ggml_cgraph * graph, c
|
|||
}
|
||||
if (galloc->n_leafs < graph->n_leafs) {
|
||||
free(galloc->leaf_allocs);
|
||||
galloc->leaf_allocs = calloc(sizeof(galloc->leaf_allocs[0]), graph->n_leafs);
|
||||
galloc->leaf_allocs = calloc(graph->n_leafs, sizeof(galloc->leaf_allocs[0]));
|
||||
GGML_ASSERT(galloc->leaf_allocs != NULL);
|
||||
}
|
||||
galloc->n_leafs = graph->n_leafs;
|
||||
|
|
|
|||
|
|
@ -822,7 +822,11 @@ GGML_CALL static enum ggml_status ggml_backend_cpu_graph_compute(ggml_backend_t
|
|||
GGML_CALL static bool ggml_backend_cpu_supports_op(ggml_backend_t backend, const struct ggml_tensor * op) {
|
||||
switch (op->op) {
|
||||
case GGML_OP_CPY:
|
||||
return op->type != GGML_TYPE_IQ2_XXS && op->type != GGML_TYPE_IQ2_XS && op->type != GGML_TYPE_IQ1_S; // missing type_traits.from_float
|
||||
return
|
||||
op->type != GGML_TYPE_IQ2_XXS &&
|
||||
op->type != GGML_TYPE_IQ2_XS &&
|
||||
op->type != GGML_TYPE_IQ1_S &&
|
||||
op->type != GGML_TYPE_IQ1_M; // missing type_traits.from_float
|
||||
case GGML_OP_MUL_MAT:
|
||||
return op->src[1]->type == GGML_TYPE_F32 || op->src[1]->type == ggml_internal_get_type_traits(op->src[0]->type).vec_dot_type;
|
||||
default:
|
||||
|
|
@ -1721,23 +1725,23 @@ ggml_backend_sched_t ggml_backend_sched_new(
|
|||
GGML_ASSERT(n_backends <= GGML_SCHED_MAX_BACKENDS);
|
||||
GGML_ASSERT(ggml_backend_is_cpu(backends[n_backends - 1])); // last backend must be CPU
|
||||
|
||||
struct ggml_backend_sched * sched = calloc(sizeof(struct ggml_backend_sched), 1);
|
||||
struct ggml_backend_sched * sched = calloc(1, sizeof(struct ggml_backend_sched));
|
||||
|
||||
// initialize hash table
|
||||
sched->hash_set = ggml_hash_set_new(graph_size);
|
||||
sched->tensor_backend_id = calloc(sizeof(sched->tensor_backend_id[0]), sched->hash_set.size);
|
||||
sched->tensor_copies = calloc(sizeof(sched->tensor_copies[0]), sched->hash_set.size);
|
||||
sched->tensor_backend_id = calloc(sched->hash_set.size, sizeof(sched->tensor_backend_id[0]));
|
||||
sched->tensor_copies = calloc(sched->hash_set.size, sizeof(sched->tensor_copies[0]));
|
||||
|
||||
const size_t nodes_size = graph_size + GGML_SCHED_MAX_SPLITS*GGML_SCHED_MAX_SPLIT_INPUTS*2;
|
||||
sched->node_backend_ids = calloc(sizeof(sched->node_backend_ids[0]), nodes_size);
|
||||
sched->leaf_backend_ids = calloc(sizeof(sched->leaf_backend_ids[0]), nodes_size);
|
||||
sched->node_backend_ids = calloc(nodes_size, sizeof(sched->node_backend_ids[0]));
|
||||
sched->leaf_backend_ids = calloc(nodes_size, sizeof(sched->leaf_backend_ids[0]));
|
||||
|
||||
sched->n_backends = n_backends;
|
||||
|
||||
sched->n_copies = parallel ? GGML_SCHED_MAX_COPIES : 1;
|
||||
|
||||
const int initial_splits_capacity = 16;
|
||||
sched->splits = calloc(sizeof(sched->splits[0]), initial_splits_capacity);
|
||||
sched->splits = calloc(initial_splits_capacity, sizeof(sched->splits[0]));
|
||||
sched->splits_capacity = initial_splits_capacity;
|
||||
|
||||
for (int b = 0; b < n_backends; b++) {
|
||||
|
|
@ -1968,10 +1972,10 @@ static void graph_copy_init_tensor(struct ggml_hash_set hash_set, struct ggml_te
|
|||
struct ggml_backend_graph_copy ggml_backend_graph_copy(ggml_backend_t backend, struct ggml_cgraph * graph) {
|
||||
struct ggml_hash_set hash_set = {
|
||||
/* .size = */ graph->visited_hash_table.size,
|
||||
/* .keys = */ calloc(sizeof(hash_set.keys[0]), graph->visited_hash_table.size) // NOLINT
|
||||
/* .keys = */ calloc(graph->visited_hash_table.size, sizeof(hash_set.keys[0])) // NOLINT
|
||||
};
|
||||
struct ggml_tensor ** node_copies = calloc(sizeof(node_copies[0]), hash_set.size); // NOLINT
|
||||
bool * node_init = calloc(sizeof(node_init[0]), hash_set.size);
|
||||
struct ggml_tensor ** node_copies = calloc(hash_set.size, sizeof(node_copies[0])); // NOLINT
|
||||
bool * node_init = calloc(hash_set.size, sizeof(node_init[0]));
|
||||
|
||||
struct ggml_init_params params = {
|
||||
/* .mem_size = */ ggml_tensor_overhead()*hash_set.size + ggml_graph_overhead_custom(graph->size, false),
|
||||
|
|
|
|||
260
ggml-impl.h
260
ggml-impl.h
|
|
@ -45,7 +45,7 @@ extern "C" {
|
|||
// 16-bit float
|
||||
// on Arm, we use __fp16
|
||||
// on x86, we use uint16_t
|
||||
#if defined(__ARM_NEON) && !defined(_MSC_VER)
|
||||
#if defined(__ARM_NEON)
|
||||
|
||||
// if YCM cannot find <arm_neon.h>, make a symbolic link to it, for example:
|
||||
//
|
||||
|
|
@ -53,8 +53,262 @@ extern "C" {
|
|||
//
|
||||
#include <arm_neon.h>
|
||||
|
||||
#ifdef _MSC_VER
|
||||
|
||||
typedef uint16_t ggml_fp16_internal_t;
|
||||
|
||||
#define ggml_vld1q_u32(w,x,y,z) { ((w) + ((uint64_t)(x) << 32)), ((y) + ((uint64_t)(z) << 32)) }
|
||||
|
||||
#else
|
||||
|
||||
typedef __fp16 ggml_fp16_internal_t;
|
||||
|
||||
#define ggml_vld1q_u32(w,x,y,z) { (w), (x), (y), (z) }
|
||||
|
||||
#endif // _MSC_VER
|
||||
|
||||
#if !defined(__aarch64__)
|
||||
|
||||
// 32-bit ARM compatibility
|
||||
|
||||
// vaddvq_s16
|
||||
// vpaddq_s16
|
||||
// vpaddq_s32
|
||||
// vaddvq_s32
|
||||
// vaddvq_f32
|
||||
// vmaxvq_f32
|
||||
// vcvtnq_s32_f32
|
||||
// vzip1_u8
|
||||
// vzip2_u8
|
||||
|
||||
inline static int32_t vaddvq_s16(int16x8_t v) {
|
||||
return
|
||||
(int32_t)vgetq_lane_s16(v, 0) + (int32_t)vgetq_lane_s16(v, 1) +
|
||||
(int32_t)vgetq_lane_s16(v, 2) + (int32_t)vgetq_lane_s16(v, 3) +
|
||||
(int32_t)vgetq_lane_s16(v, 4) + (int32_t)vgetq_lane_s16(v, 5) +
|
||||
(int32_t)vgetq_lane_s16(v, 6) + (int32_t)vgetq_lane_s16(v, 7);
|
||||
}
|
||||
|
||||
inline static int16x8_t vpaddq_s16(int16x8_t a, int16x8_t b) {
|
||||
int16x4_t a0 = vpadd_s16(vget_low_s16(a), vget_high_s16(a));
|
||||
int16x4_t b0 = vpadd_s16(vget_low_s16(b), vget_high_s16(b));
|
||||
return vcombine_s16(a0, b0);
|
||||
}
|
||||
|
||||
inline static int32x4_t vpaddq_s32(int32x4_t a, int32x4_t b) {
|
||||
int32x2_t a0 = vpadd_s32(vget_low_s32(a), vget_high_s32(a));
|
||||
int32x2_t b0 = vpadd_s32(vget_low_s32(b), vget_high_s32(b));
|
||||
return vcombine_s32(a0, b0);
|
||||
}
|
||||
|
||||
inline static int32_t vaddvq_s32(int32x4_t v) {
|
||||
return vgetq_lane_s32(v, 0) + vgetq_lane_s32(v, 1) + vgetq_lane_s32(v, 2) + vgetq_lane_s32(v, 3);
|
||||
}
|
||||
|
||||
inline static float vaddvq_f32(float32x4_t v) {
|
||||
return vgetq_lane_f32(v, 0) + vgetq_lane_f32(v, 1) + vgetq_lane_f32(v, 2) + vgetq_lane_f32(v, 3);
|
||||
}
|
||||
|
||||
inline static float vmaxvq_f32(float32x4_t v) {
|
||||
return
|
||||
MAX(MAX(vgetq_lane_f32(v, 0), vgetq_lane_f32(v, 1)),
|
||||
MAX(vgetq_lane_f32(v, 2), vgetq_lane_f32(v, 3)));
|
||||
}
|
||||
|
||||
inline static int32x4_t vcvtnq_s32_f32(float32x4_t v) {
|
||||
int32x4_t res;
|
||||
|
||||
res[0] = roundf(vgetq_lane_f32(v, 0));
|
||||
res[1] = roundf(vgetq_lane_f32(v, 1));
|
||||
res[2] = roundf(vgetq_lane_f32(v, 2));
|
||||
res[3] = roundf(vgetq_lane_f32(v, 3));
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
inline static uint8x8_t vzip1_u8(uint8x8_t a, uint8x8_t b) {
|
||||
uint8x8_t res;
|
||||
|
||||
res[0] = a[0]; res[1] = b[0];
|
||||
res[2] = a[1]; res[3] = b[1];
|
||||
res[4] = a[2]; res[5] = b[2];
|
||||
res[6] = a[3]; res[7] = b[3];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
inline static uint8x8_t vzip2_u8(uint8x8_t a, uint8x8_t b) {
|
||||
uint8x8_t res;
|
||||
|
||||
res[0] = a[4]; res[1] = b[4];
|
||||
res[2] = a[5]; res[3] = b[5];
|
||||
res[4] = a[6]; res[5] = b[6];
|
||||
res[6] = a[7]; res[7] = b[7];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
// vld1q_s16_x2
|
||||
// vld1q_u8_x2
|
||||
// vld1q_u8_x4
|
||||
// vld1q_s8_x2
|
||||
// vld1q_s8_x4
|
||||
// TODO: double-check these work correctly
|
||||
|
||||
typedef struct ggml_int16x8x2_t {
|
||||
int16x8_t val[2];
|
||||
} ggml_int16x8x2_t;
|
||||
|
||||
inline static ggml_int16x8x2_t ggml_vld1q_s16_x2(const int16_t * ptr) {
|
||||
ggml_int16x8x2_t res;
|
||||
|
||||
res.val[0] = vld1q_s16(ptr + 0);
|
||||
res.val[1] = vld1q_s16(ptr + 8);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_uint8x16x2_t {
|
||||
uint8x16_t val[2];
|
||||
} ggml_uint8x16x2_t;
|
||||
|
||||
inline static ggml_uint8x16x2_t ggml_vld1q_u8_x2(const uint8_t * ptr) {
|
||||
ggml_uint8x16x2_t res;
|
||||
|
||||
res.val[0] = vld1q_u8(ptr + 0);
|
||||
res.val[1] = vld1q_u8(ptr + 16);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_uint8x16x4_t {
|
||||
uint8x16_t val[4];
|
||||
} ggml_uint8x16x4_t;
|
||||
|
||||
inline static ggml_uint8x16x4_t ggml_vld1q_u8_x4(const uint8_t * ptr) {
|
||||
ggml_uint8x16x4_t res;
|
||||
|
||||
res.val[0] = vld1q_u8(ptr + 0);
|
||||
res.val[1] = vld1q_u8(ptr + 16);
|
||||
res.val[2] = vld1q_u8(ptr + 32);
|
||||
res.val[3] = vld1q_u8(ptr + 48);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_int8x16x2_t {
|
||||
int8x16_t val[2];
|
||||
} ggml_int8x16x2_t;
|
||||
|
||||
inline static ggml_int8x16x2_t ggml_vld1q_s8_x2(const int8_t * ptr) {
|
||||
ggml_int8x16x2_t res;
|
||||
|
||||
res.val[0] = vld1q_s8(ptr + 0);
|
||||
res.val[1] = vld1q_s8(ptr + 16);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_int8x16x4_t {
|
||||
int8x16_t val[4];
|
||||
} ggml_int8x16x4_t;
|
||||
|
||||
inline static ggml_int8x16x4_t ggml_vld1q_s8_x4(const int8_t * ptr) {
|
||||
ggml_int8x16x4_t res;
|
||||
|
||||
res.val[0] = vld1q_s8(ptr + 0);
|
||||
res.val[1] = vld1q_s8(ptr + 16);
|
||||
res.val[2] = vld1q_s8(ptr + 32);
|
||||
res.val[3] = vld1q_s8(ptr + 48);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
// NOTE: not tested
|
||||
inline static int8x16_t ggml_vqtbl1q_s8(int8x16_t a, uint8x16_t b) {
|
||||
int8x16_t res;
|
||||
|
||||
res[ 0] = a[b[ 0]];
|
||||
res[ 1] = a[b[ 1]];
|
||||
res[ 2] = a[b[ 2]];
|
||||
res[ 3] = a[b[ 3]];
|
||||
res[ 4] = a[b[ 4]];
|
||||
res[ 5] = a[b[ 5]];
|
||||
res[ 6] = a[b[ 6]];
|
||||
res[ 7] = a[b[ 7]];
|
||||
res[ 8] = a[b[ 8]];
|
||||
res[ 9] = a[b[ 9]];
|
||||
res[10] = a[b[10]];
|
||||
res[11] = a[b[11]];
|
||||
res[12] = a[b[12]];
|
||||
res[13] = a[b[13]];
|
||||
res[14] = a[b[14]];
|
||||
res[15] = a[b[15]];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
// NOTE: not tested
|
||||
inline static uint8x16_t ggml_vqtbl1q_u8(uint8x16_t a, uint8x16_t b) {
|
||||
uint8x16_t res;
|
||||
|
||||
res[ 0] = a[b[ 0]];
|
||||
res[ 1] = a[b[ 1]];
|
||||
res[ 2] = a[b[ 2]];
|
||||
res[ 3] = a[b[ 3]];
|
||||
res[ 4] = a[b[ 4]];
|
||||
res[ 5] = a[b[ 5]];
|
||||
res[ 6] = a[b[ 6]];
|
||||
res[ 7] = a[b[ 7]];
|
||||
res[ 8] = a[b[ 8]];
|
||||
res[ 9] = a[b[ 9]];
|
||||
res[10] = a[b[10]];
|
||||
res[11] = a[b[11]];
|
||||
res[12] = a[b[12]];
|
||||
res[13] = a[b[13]];
|
||||
res[14] = a[b[14]];
|
||||
res[15] = a[b[15]];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
#else
|
||||
|
||||
#define ggml_int16x8x2_t int16x8x2_t
|
||||
#define ggml_uint8x16x2_t uint8x16x2_t
|
||||
#define ggml_uint8x16x4_t uint8x16x4_t
|
||||
#define ggml_int8x16x2_t int8x16x2_t
|
||||
#define ggml_int8x16x4_t int8x16x4_t
|
||||
|
||||
#define ggml_vld1q_s16_x2 vld1q_s16_x2
|
||||
#define ggml_vld1q_u8_x2 vld1q_u8_x2
|
||||
#define ggml_vld1q_u8_x4 vld1q_u8_x4
|
||||
#define ggml_vld1q_s8_x2 vld1q_s8_x2
|
||||
#define ggml_vld1q_s8_x4 vld1q_s8_x4
|
||||
#define ggml_vqtbl1q_s8 vqtbl1q_s8
|
||||
#define ggml_vqtbl1q_u8 vqtbl1q_u8
|
||||
|
||||
#endif // !defined(__aarch64__)
|
||||
|
||||
#if !defined(__ARM_FEATURE_DOTPROD)
|
||||
|
||||
inline static int32x4_t ggml_vdotq_s32(int32x4_t acc, int8x16_t a, int8x16_t b) {
|
||||
const int16x8_t p0 = vmull_s8(vget_low_s8 (a), vget_low_s8 (b));
|
||||
const int16x8_t p1 = vmull_s8(vget_high_s8(a), vget_high_s8(b));
|
||||
|
||||
return vaddq_s32(acc, vaddq_s32(vpaddlq_s16(p0), vpaddlq_s16(p1)));
|
||||
}
|
||||
|
||||
#else
|
||||
|
||||
#define ggml_vdotq_s32(a, b, c) vdotq_s32(a, b, c)
|
||||
|
||||
#endif // !defined(__ARM_FEATURE_DOTPROD)
|
||||
|
||||
#endif // defined(__ARM_NEON)
|
||||
|
||||
#if defined(__ARM_NEON) && !defined(__MSC_VER)
|
||||
|
||||
#define GGML_COMPUTE_FP16_TO_FP32(x) ggml_compute_fp16_to_fp32(x)
|
||||
#define GGML_COMPUTE_FP32_TO_FP16(x) ggml_compute_fp32_to_fp16(x)
|
||||
|
||||
|
|
@ -75,8 +329,6 @@ static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) {
|
|||
|
||||
#else
|
||||
|
||||
typedef uint16_t ggml_fp16_internal_t;
|
||||
|
||||
#ifdef __wasm_simd128__
|
||||
#include <wasm_simd128.h>
|
||||
#else
|
||||
|
|
@ -221,7 +473,7 @@ static inline ggml_fp16_t ggml_compute_fp32_to_fp16(float f) {
|
|||
|
||||
#endif // __F16C__
|
||||
|
||||
#endif // __ARM_NEON
|
||||
#endif // defined(__ARM_NEON) && (!defined(__MSC_VER)
|
||||
|
||||
// precomputed f32 table for f16 (256 KB)
|
||||
// defined in ggml.c, initialized in ggml_init()
|
||||
|
|
|
|||
287
ggml-quants.c
287
ggml-quants.c
|
|
@ -14,41 +14,6 @@
|
|||
#include <stdlib.h> // for qsort
|
||||
#include <stdio.h> // for GGML_ASSERT
|
||||
|
||||
#ifdef __ARM_NEON
|
||||
|
||||
// if YCM cannot find <arm_neon.h>, make a symbolic link to it, for example:
|
||||
//
|
||||
// $ ln -sfn /Library/Developer/CommandLineTools/usr/lib/clang/13.1.6/include/arm_neon.h ./src/
|
||||
//
|
||||
#include <arm_neon.h>
|
||||
|
||||
#else
|
||||
|
||||
#ifdef __wasm_simd128__
|
||||
#include <wasm_simd128.h>
|
||||
#else
|
||||
#if defined(__POWER9_VECTOR__) || defined(__powerpc64__)
|
||||
#include <altivec.h>
|
||||
#undef bool
|
||||
#define bool _Bool
|
||||
#else
|
||||
#if defined(_MSC_VER) || defined(__MINGW32__)
|
||||
#include <intrin.h>
|
||||
#else
|
||||
#if defined(__AVX__) || defined(__AVX2__) || defined(__AVX512F__) || defined(__SSSE3__) || defined(__SSE3__)
|
||||
#if !defined(__riscv)
|
||||
#include <immintrin.h>
|
||||
#endif
|
||||
#endif
|
||||
#endif
|
||||
#endif
|
||||
#endif
|
||||
#endif
|
||||
|
||||
#ifdef __riscv_v_intrinsic
|
||||
#include <riscv_vector.h>
|
||||
#endif
|
||||
|
||||
#undef MIN
|
||||
#undef MAX
|
||||
|
||||
|
|
@ -276,258 +241,6 @@ static inline float hsum_float_4x4(const __m128 a, const __m128 b, const __m128
|
|||
#endif // __AVX__ || __AVX2__ || __AVX512F__
|
||||
#endif // defined(__AVX__) || defined(__AVX2__) || defined(__AVX512F__) || defined(__SSSE3__)
|
||||
|
||||
#if defined(__ARM_NEON)
|
||||
|
||||
#ifdef _MSC_VER
|
||||
|
||||
#define ggml_vld1q_u32(w,x,y,z) { ((w) + ((uint64_t)(x) << 32)), ((y) + ((uint64_t)(z) << 32)) }
|
||||
|
||||
#else
|
||||
|
||||
#define ggml_vld1q_u32(w,x,y,z) { (w), (x), (y), (z) }
|
||||
|
||||
#endif
|
||||
|
||||
#if !defined(__aarch64__)
|
||||
|
||||
// 64-bit compatibility
|
||||
|
||||
// vaddvq_s16
|
||||
// vpaddq_s16
|
||||
// vpaddq_s32
|
||||
// vaddvq_s32
|
||||
// vaddvq_f32
|
||||
// vmaxvq_f32
|
||||
// vcvtnq_s32_f32
|
||||
// vzip1_u8
|
||||
// vzip2_u8
|
||||
|
||||
inline static int32_t vaddvq_s16(int16x8_t v) {
|
||||
return
|
||||
(int32_t)vgetq_lane_s16(v, 0) + (int32_t)vgetq_lane_s16(v, 1) +
|
||||
(int32_t)vgetq_lane_s16(v, 2) + (int32_t)vgetq_lane_s16(v, 3) +
|
||||
(int32_t)vgetq_lane_s16(v, 4) + (int32_t)vgetq_lane_s16(v, 5) +
|
||||
(int32_t)vgetq_lane_s16(v, 6) + (int32_t)vgetq_lane_s16(v, 7);
|
||||
}
|
||||
|
||||
inline static int16x8_t vpaddq_s16(int16x8_t a, int16x8_t b) {
|
||||
int16x4_t a0 = vpadd_s16(vget_low_s16(a), vget_high_s16(a));
|
||||
int16x4_t b0 = vpadd_s16(vget_low_s16(b), vget_high_s16(b));
|
||||
return vcombine_s16(a0, b0);
|
||||
}
|
||||
|
||||
inline static int32x4_t vpaddq_s32(int32x4_t a, int32x4_t b) {
|
||||
int32x2_t a0 = vpadd_s32(vget_low_s32(a), vget_high_s32(a));
|
||||
int32x2_t b0 = vpadd_s32(vget_low_s32(b), vget_high_s32(b));
|
||||
return vcombine_s32(a0, b0);
|
||||
}
|
||||
|
||||
inline static int32_t vaddvq_s32(int32x4_t v) {
|
||||
return vgetq_lane_s32(v, 0) + vgetq_lane_s32(v, 1) + vgetq_lane_s32(v, 2) + vgetq_lane_s32(v, 3);
|
||||
}
|
||||
|
||||
inline static float vaddvq_f32(float32x4_t v) {
|
||||
return vgetq_lane_f32(v, 0) + vgetq_lane_f32(v, 1) + vgetq_lane_f32(v, 2) + vgetq_lane_f32(v, 3);
|
||||
}
|
||||
|
||||
inline static float vmaxvq_f32(float32x4_t v) {
|
||||
return
|
||||
MAX(MAX(vgetq_lane_f32(v, 0), vgetq_lane_f32(v, 1)),
|
||||
MAX(vgetq_lane_f32(v, 2), vgetq_lane_f32(v, 3)));
|
||||
}
|
||||
|
||||
inline static int32x4_t vcvtnq_s32_f32(float32x4_t v) {
|
||||
int32x4_t res;
|
||||
|
||||
res[0] = roundf(vgetq_lane_f32(v, 0));
|
||||
res[1] = roundf(vgetq_lane_f32(v, 1));
|
||||
res[2] = roundf(vgetq_lane_f32(v, 2));
|
||||
res[3] = roundf(vgetq_lane_f32(v, 3));
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
inline static uint8x8_t vzip1_u8(uint8x8_t a, uint8x8_t b) {
|
||||
uint8x8_t res;
|
||||
|
||||
res[0] = a[0]; res[1] = b[0];
|
||||
res[2] = a[1]; res[3] = b[1];
|
||||
res[4] = a[2]; res[5] = b[2];
|
||||
res[6] = a[3]; res[7] = b[3];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
inline static uint8x8_t vzip2_u8(uint8x8_t a, uint8x8_t b) {
|
||||
uint8x8_t res;
|
||||
|
||||
res[0] = a[4]; res[1] = b[4];
|
||||
res[2] = a[5]; res[3] = b[5];
|
||||
res[4] = a[6]; res[5] = b[6];
|
||||
res[6] = a[7]; res[7] = b[7];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
// vld1q_s16_x2
|
||||
// vld1q_u8_x2
|
||||
// vld1q_u8_x4
|
||||
// vld1q_s8_x2
|
||||
// vld1q_s8_x4
|
||||
// TODO: double-check these work correctly
|
||||
|
||||
typedef struct ggml_int16x8x2_t {
|
||||
int16x8_t val[2];
|
||||
} ggml_int16x8x2_t;
|
||||
|
||||
inline static ggml_int16x8x2_t ggml_vld1q_s16_x2(const int16_t * ptr) {
|
||||
ggml_int16x8x2_t res;
|
||||
|
||||
res.val[0] = vld1q_s16(ptr + 0);
|
||||
res.val[1] = vld1q_s16(ptr + 8);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_uint8x16x2_t {
|
||||
uint8x16_t val[2];
|
||||
} ggml_uint8x16x2_t;
|
||||
|
||||
inline static ggml_uint8x16x2_t ggml_vld1q_u8_x2(const uint8_t * ptr) {
|
||||
ggml_uint8x16x2_t res;
|
||||
|
||||
res.val[0] = vld1q_u8(ptr + 0);
|
||||
res.val[1] = vld1q_u8(ptr + 16);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_uint8x16x4_t {
|
||||
uint8x16_t val[4];
|
||||
} ggml_uint8x16x4_t;
|
||||
|
||||
inline static ggml_uint8x16x4_t ggml_vld1q_u8_x4(const uint8_t * ptr) {
|
||||
ggml_uint8x16x4_t res;
|
||||
|
||||
res.val[0] = vld1q_u8(ptr + 0);
|
||||
res.val[1] = vld1q_u8(ptr + 16);
|
||||
res.val[2] = vld1q_u8(ptr + 32);
|
||||
res.val[3] = vld1q_u8(ptr + 48);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_int8x16x2_t {
|
||||
int8x16_t val[2];
|
||||
} ggml_int8x16x2_t;
|
||||
|
||||
inline static ggml_int8x16x2_t ggml_vld1q_s8_x2(const int8_t * ptr) {
|
||||
ggml_int8x16x2_t res;
|
||||
|
||||
res.val[0] = vld1q_s8(ptr + 0);
|
||||
res.val[1] = vld1q_s8(ptr + 16);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
typedef struct ggml_int8x16x4_t {
|
||||
int8x16_t val[4];
|
||||
} ggml_int8x16x4_t;
|
||||
|
||||
inline static ggml_int8x16x4_t ggml_vld1q_s8_x4(const int8_t * ptr) {
|
||||
ggml_int8x16x4_t res;
|
||||
|
||||
res.val[0] = vld1q_s8(ptr + 0);
|
||||
res.val[1] = vld1q_s8(ptr + 16);
|
||||
res.val[2] = vld1q_s8(ptr + 32);
|
||||
res.val[3] = vld1q_s8(ptr + 48);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
// NOTE: not tested
|
||||
inline static int8x16_t ggml_vqtbl1q_s8(int8x16_t a, uint8x16_t b) {
|
||||
int8x16_t res;
|
||||
|
||||
res[ 0] = a[b[ 0]];
|
||||
res[ 1] = a[b[ 1]];
|
||||
res[ 2] = a[b[ 2]];
|
||||
res[ 3] = a[b[ 3]];
|
||||
res[ 4] = a[b[ 4]];
|
||||
res[ 5] = a[b[ 5]];
|
||||
res[ 6] = a[b[ 6]];
|
||||
res[ 7] = a[b[ 7]];
|
||||
res[ 8] = a[b[ 8]];
|
||||
res[ 9] = a[b[ 9]];
|
||||
res[10] = a[b[10]];
|
||||
res[11] = a[b[11]];
|
||||
res[12] = a[b[12]];
|
||||
res[13] = a[b[13]];
|
||||
res[14] = a[b[14]];
|
||||
res[15] = a[b[15]];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
// NOTE: not tested
|
||||
inline static uint8x16_t ggml_vqtbl1q_u8(uint8x16_t a, uint8x16_t b) {
|
||||
uint8x16_t res;
|
||||
|
||||
res[ 0] = a[b[ 0]];
|
||||
res[ 1] = a[b[ 1]];
|
||||
res[ 2] = a[b[ 2]];
|
||||
res[ 3] = a[b[ 3]];
|
||||
res[ 4] = a[b[ 4]];
|
||||
res[ 5] = a[b[ 5]];
|
||||
res[ 6] = a[b[ 6]];
|
||||
res[ 7] = a[b[ 7]];
|
||||
res[ 8] = a[b[ 8]];
|
||||
res[ 9] = a[b[ 9]];
|
||||
res[10] = a[b[10]];
|
||||
res[11] = a[b[11]];
|
||||
res[12] = a[b[12]];
|
||||
res[13] = a[b[13]];
|
||||
res[14] = a[b[14]];
|
||||
res[15] = a[b[15]];
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
#else
|
||||
|
||||
#define ggml_int16x8x2_t int16x8x2_t
|
||||
#define ggml_uint8x16x2_t uint8x16x2_t
|
||||
#define ggml_uint8x16x4_t uint8x16x4_t
|
||||
#define ggml_int8x16x2_t int8x16x2_t
|
||||
#define ggml_int8x16x4_t int8x16x4_t
|
||||
|
||||
#define ggml_vld1q_s16_x2 vld1q_s16_x2
|
||||
#define ggml_vld1q_u8_x2 vld1q_u8_x2
|
||||
#define ggml_vld1q_u8_x4 vld1q_u8_x4
|
||||
#define ggml_vld1q_s8_x2 vld1q_s8_x2
|
||||
#define ggml_vld1q_s8_x4 vld1q_s8_x4
|
||||
#define ggml_vqtbl1q_s8 vqtbl1q_s8
|
||||
#define ggml_vqtbl1q_u8 vqtbl1q_u8
|
||||
|
||||
#endif
|
||||
|
||||
#if !defined(__ARM_FEATURE_DOTPROD)
|
||||
|
||||
inline static int32x4_t ggml_vdotq_s32(int32x4_t acc, int8x16_t a, int8x16_t b) {
|
||||
const int16x8_t p0 = vmull_s8(vget_low_s8 (a), vget_low_s8 (b));
|
||||
const int16x8_t p1 = vmull_s8(vget_high_s8(a), vget_high_s8(b));
|
||||
|
||||
return vaddq_s32(acc, vaddq_s32(vpaddlq_s16(p0), vpaddlq_s16(p1)));
|
||||
}
|
||||
|
||||
#else
|
||||
|
||||
#define ggml_vdotq_s32(a, b, c) vdotq_s32(a, b, c)
|
||||
|
||||
#endif
|
||||
|
||||
#endif
|
||||
|
||||
#if defined(__ARM_NEON) || defined(__wasm_simd128__)
|
||||
#define B1(c,s,n) 0x ## n ## c , 0x ## n ## s
|
||||
#define B2(c,s,n) B1(c,s,n ## c), B1(c,s,n ## s)
|
||||
|
|
|
|||
8
ggml.c
8
ggml.c
|
|
@ -10948,7 +10948,7 @@ static void ggml_compute_forward_mul_mat(
|
|||
#endif
|
||||
|
||||
#if GGML_USE_LLAMAFILE
|
||||
if (nb10 == ggml_type_size(src1->type)) {
|
||||
if (src1_cont) {
|
||||
for (int64_t i13 = 0; i13 < ne13; i13++)
|
||||
for (int64_t i12 = 0; i12 < ne12; i12++)
|
||||
if (!llamafile_sgemm(ne01, ne11, ne00/ggml_blck_size(src0->type),
|
||||
|
|
@ -11001,15 +11001,13 @@ UseGgmlGemm1:;
|
|||
const size_t row_size = ggml_row_size(vec_dot_type, ne10);
|
||||
|
||||
#if GGML_USE_LLAMAFILE
|
||||
if (nb10 == ggml_type_size(src1->type) || src1->type != vec_dot_type) {
|
||||
if (src1->type != vec_dot_type) {
|
||||
for (int64_t i13 = 0; i13 < ne13; i13++)
|
||||
for (int64_t i12 = 0; i12 < ne12; i12++)
|
||||
if (!llamafile_sgemm(ne01, ne11, ne00/ggml_blck_size(src0->type),
|
||||
(const char *)src0->data + i12/r2*nb02 + i13/r3*nb03,
|
||||
nb01/ggml_type_size(src0->type),
|
||||
(const char *)wdata + ggml_row_size(vec_dot_type,
|
||||
nb12/ggml_type_size(src1->type)*i12 +
|
||||
nb13/ggml_type_size(src1->type)*i13),
|
||||
(const char *)wdata + (i12*ne11 + i13*ne12*ne11)*row_size,
|
||||
row_size/ggml_type_size(vec_dot_type),
|
||||
(char *)dst->data + i12*nb2 + i13*nb3,
|
||||
nb1/ggml_type_size(dst->type),
|
||||
|
|
|
|||
|
|
@ -124,6 +124,7 @@ class MODEL_ARCH(IntEnum):
|
|||
QWEN2 = auto()
|
||||
QWEN2MOE = auto()
|
||||
PHI2 = auto()
|
||||
PHI3 = auto()
|
||||
PLAMO = auto()
|
||||
CODESHELL = auto()
|
||||
ORION = auto()
|
||||
|
|
@ -200,6 +201,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
|
|||
MODEL_ARCH.QWEN2: "qwen2",
|
||||
MODEL_ARCH.QWEN2MOE: "qwen2moe",
|
||||
MODEL_ARCH.PHI2: "phi2",
|
||||
MODEL_ARCH.PHI3: "phi3",
|
||||
MODEL_ARCH.PLAMO: "plamo",
|
||||
MODEL_ARCH.CODESHELL: "codeshell",
|
||||
MODEL_ARCH.ORION: "orion",
|
||||
|
|
@ -550,6 +552,20 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
|
|||
MODEL_TENSOR.FFN_DOWN,
|
||||
MODEL_TENSOR.FFN_UP,
|
||||
],
|
||||
MODEL_ARCH.PHI3: [
|
||||
MODEL_TENSOR.TOKEN_EMBD,
|
||||
MODEL_TENSOR.OUTPUT_NORM,
|
||||
MODEL_TENSOR.OUTPUT,
|
||||
MODEL_TENSOR.ATTN_NORM,
|
||||
MODEL_TENSOR.ATTN_QKV,
|
||||
MODEL_TENSOR.ATTN_Q,
|
||||
MODEL_TENSOR.ATTN_K,
|
||||
MODEL_TENSOR.ATTN_V,
|
||||
MODEL_TENSOR.ATTN_OUT,
|
||||
MODEL_TENSOR.FFN_NORM,
|
||||
MODEL_TENSOR.FFN_DOWN,
|
||||
MODEL_TENSOR.FFN_UP,
|
||||
],
|
||||
MODEL_ARCH.CODESHELL: [
|
||||
MODEL_TENSOR.TOKEN_EMBD,
|
||||
MODEL_TENSOR.POS_EMBD,
|
||||
|
|
@ -872,6 +888,7 @@ GGML_QUANT_SIZES = {
|
|||
GGMLQuantizationType.I32: (1, 4),
|
||||
GGMLQuantizationType.I64: (1, 8),
|
||||
GGMLQuantizationType.F64: (1, 8),
|
||||
GGMLQuantizationType.IQ1_M: (256, QK_K // 8 + QK_K // 16 + QK_K // 32),
|
||||
}
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -117,6 +117,7 @@ class TensorNameMap:
|
|||
"h.{bid}.attn.c_attn", # gpt2
|
||||
"transformer.h.{bid}.mixer.Wqkv", # phi2
|
||||
"encoder.layers.{bid}.attn.Wqkv", # nomic-bert
|
||||
"model.layers.{bid}.self_attn.qkv_proj" # phi3
|
||||
),
|
||||
|
||||
# Attention query
|
||||
|
|
@ -234,6 +235,7 @@ class TensorNameMap:
|
|||
"h.{bid}.mlp.c_fc", # gpt2
|
||||
"transformer.h.{bid}.mlp.fc1", # phi2
|
||||
"model.layers.{bid}.mlp.fc1", # phi2
|
||||
"model.layers.{bid}.mlp.gate_up_proj", # phi3
|
||||
"model.layers.layers.{bid}.mlp.up_proj", # plamo
|
||||
"model.layers.{bid}.feed_forward.w3", # internlm2
|
||||
"encoder.layers.{bid}.mlp.fc11", # nomic-bert
|
||||
|
|
|
|||
300
llama.cpp
300
llama.cpp
|
|
@ -210,6 +210,7 @@ enum llm_arch {
|
|||
LLM_ARCH_QWEN2,
|
||||
LLM_ARCH_QWEN2MOE,
|
||||
LLM_ARCH_PHI2,
|
||||
LLM_ARCH_PHI3,
|
||||
LLM_ARCH_PLAMO,
|
||||
LLM_ARCH_CODESHELL,
|
||||
LLM_ARCH_ORION,
|
||||
|
|
@ -245,6 +246,7 @@ static const std::map<llm_arch, const char *> LLM_ARCH_NAMES = {
|
|||
{ LLM_ARCH_QWEN2, "qwen2" },
|
||||
{ LLM_ARCH_QWEN2MOE, "qwen2moe" },
|
||||
{ LLM_ARCH_PHI2, "phi2" },
|
||||
{ LLM_ARCH_PHI3, "phi3" },
|
||||
{ LLM_ARCH_PLAMO, "plamo" },
|
||||
{ LLM_ARCH_CODESHELL, "codeshell" },
|
||||
{ LLM_ARCH_ORION, "orion" },
|
||||
|
|
@ -792,6 +794,23 @@ static const std::map<llm_arch, std::map<llm_tensor, std::string>> LLM_TENSOR_NA
|
|||
{ LLM_TENSOR_FFN_UP, "blk.%d.ffn_up" },
|
||||
},
|
||||
},
|
||||
{
|
||||
LLM_ARCH_PHI3,
|
||||
{
|
||||
{ LLM_TENSOR_TOKEN_EMBD, "token_embd" },
|
||||
{ LLM_TENSOR_OUTPUT_NORM, "output_norm" },
|
||||
{ LLM_TENSOR_OUTPUT, "output" },
|
||||
{ LLM_TENSOR_ATTN_NORM, "blk.%d.attn_norm" },
|
||||
{ LLM_TENSOR_ATTN_QKV, "blk.%d.attn_qkv" },
|
||||
{ LLM_TENSOR_ATTN_Q, "blk.%d.attn_q" },
|
||||
{ LLM_TENSOR_ATTN_K, "blk.%d.attn_k" },
|
||||
{ LLM_TENSOR_ATTN_V, "blk.%d.attn_v" },
|
||||
{ LLM_TENSOR_ATTN_OUT, "blk.%d.attn_output" },
|
||||
{ LLM_TENSOR_FFN_NORM, "blk.%d.ffn_norm" },
|
||||
{ LLM_TENSOR_FFN_DOWN, "blk.%d.ffn_down" },
|
||||
{ LLM_TENSOR_FFN_UP, "blk.%d.ffn_up" },
|
||||
},
|
||||
},
|
||||
{
|
||||
LLM_ARCH_PLAMO,
|
||||
{
|
||||
|
|
@ -1599,12 +1618,12 @@ struct llama_mlock {
|
|||
};
|
||||
using llama_mlocks = std::vector<std::unique_ptr<llama_mlock>>;
|
||||
|
||||
static std::string llama_token_to_piece(const struct llama_context * ctx, llama_token token) {
|
||||
static std::string llama_token_to_piece(const struct llama_context * ctx, llama_token token, bool special) {
|
||||
std::vector<char> result(8, 0);
|
||||
const int n_tokens = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size());
|
||||
const int n_tokens = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size(), special);
|
||||
if (n_tokens < 0) {
|
||||
result.resize(-n_tokens);
|
||||
int check = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size());
|
||||
int check = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size(), special);
|
||||
GGML_ASSERT(check == -n_tokens);
|
||||
}
|
||||
else {
|
||||
|
|
@ -2120,7 +2139,7 @@ struct llama_vocab {
|
|||
id special_prefix_id = -1;
|
||||
id special_suffix_id = -1;
|
||||
id special_middle_id = -1;
|
||||
id special_eot_id = -1;
|
||||
id special_eot_id = -1; // TODO: move above after "eos_id", and here add "file separator" token
|
||||
|
||||
bool add_space_prefix = true;
|
||||
|
||||
|
|
@ -3770,7 +3789,7 @@ static void llm_load_hparams(
|
|||
switch (hparams.n_layer) {
|
||||
case 22: model.type = e_model::MODEL_1B; break;
|
||||
case 26: model.type = e_model::MODEL_3B; break;
|
||||
case 32: model.type = e_model::MODEL_7B; break;
|
||||
case 32: model.type = hparams.n_head == hparams.n_head_kv ? e_model::MODEL_7B : e_model::MODEL_8B; break; // LLaMa 8B v3 uses GQA
|
||||
case 40: model.type = e_model::MODEL_13B; break;
|
||||
case 48: model.type = e_model::MODEL_34B; break;
|
||||
case 60: model.type = e_model::MODEL_30B; break;
|
||||
|
|
@ -3955,6 +3974,16 @@ static void llm_load_hparams(
|
|||
{
|
||||
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
|
||||
|
||||
switch (hparams.n_layer) {
|
||||
case 24: model.type = e_model::MODEL_1B; break;
|
||||
case 32: model.type = e_model::MODEL_3B; break;
|
||||
default: model.type = e_model::MODEL_UNKNOWN;
|
||||
}
|
||||
} break;
|
||||
case LLM_ARCH_PHI3:
|
||||
{
|
||||
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
|
||||
|
||||
switch (hparams.n_layer) {
|
||||
case 24: model.type = e_model::MODEL_1B; break;
|
||||
case 32: model.type = e_model::MODEL_3B; break;
|
||||
|
|
@ -4179,7 +4208,10 @@ static void llm_load_vocab(
|
|||
vocab.special_prefix_id = 67;
|
||||
vocab.special_suffix_id = 69;
|
||||
vocab.special_middle_id = 68;
|
||||
vocab.special_eot_id = 70;
|
||||
// TODO: this is not EOT, it is "file separator" token, needs fix
|
||||
// https://huggingface.co/google/codegemma-7b-it/blob/9b1d9231388358c04d90bd003458f5070d97db44/tokenizer_config.json#L565-L572
|
||||
//vocab.special_eot_id = 70;
|
||||
vocab.special_eot_id = 107;
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -4308,6 +4340,7 @@ static void llm_load_vocab(
|
|||
{ LLM_KV_TOKENIZER_MIDDLE_ID, vocab.special_middle_id },
|
||||
{ LLM_KV_TOKENIZER_EOT_ID, vocab.special_eot_id },
|
||||
};
|
||||
|
||||
for (const auto & it : special_token_types) {
|
||||
const std::string & key = kv(std::get<0>(it));
|
||||
int32_t & id = std::get<1>(it);
|
||||
|
|
@ -4322,7 +4355,6 @@ static void llm_load_vocab(
|
|||
} else {
|
||||
id = new_id;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
// Handle add_bos_token and add_eos_token
|
||||
|
|
@ -4336,6 +4368,28 @@ static void llm_load_vocab(
|
|||
vocab.special_add_eos = int(temp);
|
||||
}
|
||||
}
|
||||
|
||||
// find EOT token: "<|eot_id|>", "<|im_end|>", "<end_of_turn>", etc.
|
||||
//
|
||||
// TODO: convert scripts should provide this token through the KV metadata LLAMA_KV_TOKENIZER_EOT_ID
|
||||
// for now, we apply this workaround to find the EOT token based on its text
|
||||
if (vocab.special_eot_id == -1) {
|
||||
for (const auto & t : vocab.token_to_id) {
|
||||
if (
|
||||
// TODO: gemma "<end_of_turn>" is exported as a normal token, so the following check does not work
|
||||
// need to fix convert script
|
||||
//vocab.id_to_token[t.second].type == LLAMA_TOKEN_TYPE_CONTROL &&
|
||||
(t.first == "<|eot_id|>" ||
|
||||
t.first == "<|im_end|>" ||
|
||||
t.first == "<|end|>" ||
|
||||
t.first == "<end_of_turn>"
|
||||
)
|
||||
) {
|
||||
vocab.special_eot_id = t.second;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// build special tokens cache
|
||||
|
|
@ -4505,7 +4559,12 @@ static void llm_load_print_meta(llama_model_loader & ml, llama_model & model) {
|
|||
if (vocab.special_pad_id != -1) { LLAMA_LOG_INFO( "%s: PAD token = %d '%s'\n", __func__, vocab.special_pad_id, vocab.id_to_token[vocab.special_pad_id].text.c_str() ); }
|
||||
if (vocab.special_cls_id != -1) { LLAMA_LOG_INFO( "%s: CLS token = %d '%s'\n", __func__, vocab.special_cls_id, vocab.id_to_token[vocab.special_cls_id].text.c_str() ); }
|
||||
if (vocab.special_mask_id != -1) { LLAMA_LOG_INFO( "%s: MASK token = %d '%s'\n", __func__, vocab.special_mask_id, vocab.id_to_token[vocab.special_mask_id].text.c_str() ); }
|
||||
|
||||
if (vocab.linefeed_id != -1) { LLAMA_LOG_INFO( "%s: LF token = %d '%s'\n", __func__, vocab.linefeed_id, vocab.id_to_token[vocab.linefeed_id].text.c_str() ); }
|
||||
if (vocab.special_prefix_id != -1) { LLAMA_LOG_INFO( "%s: PRE token = %d '%s'\n", __func__, vocab.special_prefix_id, vocab.id_to_token[vocab.special_prefix_id].text.c_str() ); }
|
||||
if (vocab.special_suffix_id != -1) { LLAMA_LOG_INFO( "%s: SUF token = %d '%s'\n", __func__, vocab.special_suffix_id, vocab.id_to_token[vocab.special_suffix_id].text.c_str() ); }
|
||||
if (vocab.special_middle_id != -1) { LLAMA_LOG_INFO( "%s: MID token = %d '%s'\n", __func__, vocab.special_middle_id, vocab.id_to_token[vocab.special_middle_id].text.c_str() ); }
|
||||
if (vocab.special_eot_id != -1) { LLAMA_LOG_INFO( "%s: EOT token = %d '%s'\n", __func__, vocab.special_eot_id, vocab.id_to_token[vocab.special_eot_id].text.c_str() ); }
|
||||
}
|
||||
|
||||
// Returns false if cancelled by progress_callback
|
||||
|
|
@ -5346,6 +5405,33 @@ static bool llm_load_tensors(
|
|||
layer.ffn_up_b = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_FFN_UP, "bias", i), {n_ff});
|
||||
}
|
||||
} break;
|
||||
case LLM_ARCH_PHI3:
|
||||
{
|
||||
model.tok_embd = ml.create_tensor(ctx_input, tn(LLM_TENSOR_TOKEN_EMBD, "weight"), { n_embd, n_vocab });
|
||||
|
||||
// output
|
||||
{
|
||||
model.output_norm = ml.create_tensor(ctx_output, tn(LLM_TENSOR_OUTPUT_NORM, "weight"), { n_embd });
|
||||
model.output = ml.create_tensor(ctx_output_split, tn(LLM_TENSOR_OUTPUT, "weight"), { n_embd, n_vocab });
|
||||
}
|
||||
|
||||
for (int i = 0; i < n_layer; ++i) {
|
||||
ggml_context* ctx_layer = ctx_for_layer(i);
|
||||
ggml_context* ctx_split = ctx_for_layer_split(i);
|
||||
|
||||
auto& layer = model.layers[i];
|
||||
|
||||
layer.attn_norm = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_ATTN_NORM, "weight", i), { n_embd });
|
||||
|
||||
layer.wqkv = ml.create_tensor(ctx_split, tn(LLM_TENSOR_ATTN_QKV, "weight", i), { n_embd, n_embd + 2 * n_embd_gqa }, false);
|
||||
layer.wo = ml.create_tensor(ctx_split, tn(LLM_TENSOR_ATTN_OUT, "weight", i), { n_embd, n_embd });
|
||||
|
||||
layer.ffn_norm = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_FFN_NORM, "weight", i), { n_embd });
|
||||
|
||||
layer.ffn_down = ml.create_tensor(ctx_split, tn(LLM_TENSOR_FFN_DOWN, "weight", i), { n_ff, n_embd });
|
||||
layer.ffn_up = ml.create_tensor(ctx_split, tn(LLM_TENSOR_FFN_UP, "weight", i), { n_embd, 2 * n_ff });
|
||||
}
|
||||
} break;
|
||||
case LLM_ARCH_PLAMO:
|
||||
{
|
||||
model.tok_embd = ml.create_tensor(ctx_input, tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab});
|
||||
|
|
@ -6325,7 +6411,7 @@ static struct ggml_tensor * llm_build_kqv(
|
|||
|
||||
cur = ggml_flash_attn_ext(ctx, q, k, v, kq_mask, kq_scale);
|
||||
|
||||
if (model.arch == LLM_ARCH_PHI2) {
|
||||
if (model.arch == LLM_ARCH_PHI2 || model.arch == LLM_ARCH_PHI3) {
|
||||
ggml_flash_attn_ext_set_prec(cur, GGML_PREC_F32);
|
||||
}
|
||||
|
||||
|
|
@ -6334,7 +6420,7 @@ static struct ggml_tensor * llm_build_kqv(
|
|||
struct ggml_tensor * kq = ggml_mul_mat(ctx, k, q);
|
||||
cb(kq, "kq", il);
|
||||
|
||||
if (model.arch == LLM_ARCH_PHI2) {
|
||||
if (model.arch == LLM_ARCH_PHI2 || model.arch == LLM_ARCH_PHI3) {
|
||||
// for this arch, we need to perform the KQ multiplication with F32 precision, otherwise we get NaNs
|
||||
// ref: https://github.com/ggerganov/llama.cpp/pull/4490#issuecomment-1859055847
|
||||
ggml_mul_mat_set_prec(kq, GGML_PREC_F32);
|
||||
|
|
@ -9002,12 +9088,140 @@ struct llm_build_context {
|
|||
|
||||
cur = ggml_add(ctx0, cur, model.output_b);
|
||||
cb(cur, "result_output", -1);
|
||||
ggml_build_forward_expand(gf, cur);
|
||||
return gf;
|
||||
}
|
||||
|
||||
struct ggml_cgraph * build_phi3() {
|
||||
struct ggml_cgraph * gf = ggml_new_graph_custom(ctx0, LLAMA_MAX_NODES, false);
|
||||
|
||||
const int64_t n_embd_head = hparams.n_embd_head_v;
|
||||
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
|
||||
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
|
||||
|
||||
struct ggml_tensor * cur;
|
||||
struct ggml_tensor * inpL;
|
||||
|
||||
inpL = llm_build_inp_embd(ctx0, lctx, hparams, batch, model.tok_embd, cb);
|
||||
|
||||
// inp_pos - contains the positions
|
||||
struct ggml_tensor * inp_pos = build_inp_pos();
|
||||
|
||||
// KQ_mask (mask for 1 head, it will be broadcasted to all heads)
|
||||
struct ggml_tensor * KQ_mask = build_inp_KQ_mask();
|
||||
|
||||
for (int il = 0; il < n_layer; ++il) {
|
||||
auto residual = inpL;
|
||||
|
||||
// self-attention
|
||||
{
|
||||
struct ggml_tensor* attn_norm_output = llm_build_norm(ctx0, inpL, hparams,
|
||||
model.layers[il].attn_norm,
|
||||
NULL,
|
||||
LLM_NORM_RMS, cb, il);
|
||||
cb(attn_norm_output, "attn_norm", il);
|
||||
|
||||
struct ggml_tensor * Qcur = nullptr;
|
||||
struct ggml_tensor * Kcur = nullptr;
|
||||
struct ggml_tensor * Vcur = nullptr;
|
||||
|
||||
if (model.layers[il].wqkv) {
|
||||
cur = ggml_mul_mat(ctx0, model.layers[il].wqkv, attn_norm_output);
|
||||
cb(cur, "wqkv", il);
|
||||
|
||||
Qcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd, n_tokens, cur->nb[1], 0 * sizeof(float) * (n_embd)));
|
||||
Kcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd_gqa, n_tokens, cur->nb[1], 1 * sizeof(float) * (n_embd)));
|
||||
Vcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd_gqa, n_tokens, cur->nb[1], 1 * sizeof(float) * (n_embd + n_embd_gqa)));
|
||||
}
|
||||
else {
|
||||
Qcur = ggml_add(ctx0, ggml_mul_mat(ctx0, model.layers[il].wq, attn_norm_output), model.layers[il].bq);
|
||||
Kcur = ggml_add(ctx0, ggml_mul_mat(ctx0, model.layers[il].wk, attn_norm_output), model.layers[il].bk);
|
||||
Vcur = ggml_add(ctx0, ggml_mul_mat(ctx0, model.layers[il].wv, attn_norm_output), model.layers[il].bv);
|
||||
}
|
||||
|
||||
cb(Qcur, "Qcur", il);
|
||||
cb(Kcur, "Kcur", il);
|
||||
cb(Vcur, "Vcur", il);
|
||||
|
||||
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
|
||||
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
|
||||
|
||||
Qcur = ggml_rope_custom(
|
||||
ctx0, Qcur, inp_pos, n_rot, rope_type, 0, n_orig_ctx,
|
||||
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
|
||||
);
|
||||
cb(Qcur, "Qcur", il);
|
||||
|
||||
Qcur = ggml_scale(ctx0, Qcur, 1.0f / sqrtf(float(n_embd_head)));
|
||||
cb(Qcur, "Qcur", il);
|
||||
|
||||
Kcur = ggml_rope_custom(
|
||||
ctx0, Kcur, inp_pos, n_rot, rope_type, 0, n_orig_ctx,
|
||||
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow
|
||||
);
|
||||
cb(Kcur, "Kcur", il);
|
||||
|
||||
cur = llm_build_kv(ctx0, model, hparams, cparams, kv_self, gf,
|
||||
model.layers[il].wo, model.layers[il].bo,
|
||||
Kcur, Vcur, Qcur, KQ_mask, nullptr, n_tokens, kv_head, n_kv, 1.0f, cb, il);
|
||||
}
|
||||
|
||||
if (il == n_layer - 1) {
|
||||
// skip computing output for unused tokens
|
||||
struct ggml_tensor* inp_out_ids = build_inp_out_ids();
|
||||
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
|
||||
residual = ggml_get_rows(ctx0, residual, inp_out_ids);
|
||||
}
|
||||
|
||||
cur = ggml_add(ctx0, cur, residual);
|
||||
residual = cur;
|
||||
|
||||
cur = llm_build_norm(ctx0, cur, hparams,
|
||||
model.layers[il].ffn_norm, NULL,
|
||||
LLM_NORM_RMS, cb, il);
|
||||
cb(cur, "ffn_norm", il);
|
||||
|
||||
// FF
|
||||
// special-case: the up and gate tensors are merged into a single tensor
|
||||
// TOOD: support into llm_build_ffn
|
||||
{
|
||||
struct ggml_tensor* up = ggml_mul_mat(ctx0, model.layers[il].ffn_up, cur);
|
||||
cb(up, "ffn_up", il);
|
||||
|
||||
auto g = ggml_cont(ctx0, ggml_view_2d(ctx0, up, up->ne[0] / 2, up->ne[1], ggml_row_size(up->type, up->ne[0]), 0));
|
||||
auto y = ggml_cont(ctx0, ggml_view_2d(ctx0, up, up->ne[0] / 2, up->ne[1], ggml_row_size(up->type, up->ne[0]), up->nb[1] / 2));
|
||||
|
||||
y = ggml_mul(ctx0, y, ggml_silu(ctx0, g));
|
||||
cb(y, "ffn_gate", il);
|
||||
|
||||
auto down = ggml_mul_mat(ctx0, model.layers[il].ffn_down, y);
|
||||
cb(down, "ffn_down", il);
|
||||
|
||||
cur = down;
|
||||
cb(cur, "ffn_out", il);
|
||||
}
|
||||
|
||||
cur = ggml_add(ctx0, residual, cur);
|
||||
cb(cur, "l_out", il);
|
||||
|
||||
inpL = cur;
|
||||
}
|
||||
|
||||
cur = llm_build_norm(ctx0, inpL, hparams,
|
||||
model.output_norm,
|
||||
NULL,
|
||||
LLM_NORM_RMS, cb, -1);
|
||||
cb(cur, "result_norm", -1);
|
||||
|
||||
cur = ggml_mul_mat(ctx0, model.output, cur);
|
||||
cb(cur, "result_output", -1);
|
||||
|
||||
ggml_build_forward_expand(gf, cur);
|
||||
|
||||
return gf;
|
||||
}
|
||||
|
||||
|
||||
struct ggml_cgraph * build_plamo() {
|
||||
struct ggml_cgraph * gf = ggml_new_graph(ctx0);
|
||||
|
||||
|
|
@ -10509,6 +10723,10 @@ static struct ggml_cgraph * llama_build_graph(
|
|||
{
|
||||
result = llm.build_phi2();
|
||||
} break;
|
||||
case LLM_ARCH_PHI3:
|
||||
{
|
||||
result = llm.build_phi3();
|
||||
} break;
|
||||
case LLM_ARCH_PLAMO:
|
||||
{
|
||||
result = llm.build_plamo();
|
||||
|
|
@ -13336,16 +13554,14 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
|
|||
GGML_ASSERT(ctx);
|
||||
const int64_t t_start_sample_us = ggml_time_us();
|
||||
|
||||
bool allow_eos = false;
|
||||
bool allow_eog = false;
|
||||
for (const auto & stack : grammar->stacks) {
|
||||
if (stack.empty()) {
|
||||
allow_eos = true;
|
||||
allow_eog = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
const llama_token eos = llama_token_eos(&ctx->model);
|
||||
|
||||
std::vector<std::pair<std::vector<uint32_t>, llama_partial_utf8>> candidates_decoded;
|
||||
candidates_decoded.reserve(candidates->size);
|
||||
std::vector<llama_grammar_candidate> candidates_grammar;
|
||||
|
|
@ -13353,9 +13569,10 @@ void llama_sample_grammar(struct llama_context * ctx, llama_token_data_array * c
|
|||
|
||||
for (size_t i = 0; i < candidates->size; ++i) {
|
||||
const llama_token id = candidates->data[i].id;
|
||||
const std::string piece = llama_token_to_piece(ctx, id);
|
||||
if (id == eos) {
|
||||
if (!allow_eos) {
|
||||
const std::string piece = llama_token_to_piece(ctx, id, false);
|
||||
|
||||
if (llama_token_is_eog(&ctx->model, id)) {
|
||||
if (!allow_eog) {
|
||||
candidates->data[i].logit = -INFINITY;
|
||||
}
|
||||
} else if (piece.empty() || piece[0] == 0) {
|
||||
|
|
@ -13518,7 +13735,7 @@ llama_token llama_sample_token_greedy(struct llama_context * ctx, llama_token_da
|
|||
return result;
|
||||
}
|
||||
|
||||
llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
|
||||
llama_token llama_sample_token_with_rng(struct llama_context * ctx, llama_token_data_array * candidates, std::mt19937 & rng) {
|
||||
GGML_ASSERT(ctx);
|
||||
|
||||
const int64_t t_start_sample_us = ggml_time_us();
|
||||
|
|
@ -13531,7 +13748,6 @@ llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_arra
|
|||
}
|
||||
|
||||
std::discrete_distribution<> dist(probs.begin(), probs.end());
|
||||
auto & rng = ctx->rng;
|
||||
int idx = dist(rng);
|
||||
|
||||
llama_token result = candidates->data[idx].id;
|
||||
|
|
@ -13541,10 +13757,14 @@ llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_arra
|
|||
return result;
|
||||
}
|
||||
|
||||
llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_array * candidates) {
|
||||
return llama_sample_token_with_rng(ctx, candidates, ctx->rng);
|
||||
}
|
||||
|
||||
void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar * grammar, llama_token token) {
|
||||
const int64_t t_start_sample_us = ggml_time_us();
|
||||
|
||||
if (token == llama_token_eos(&ctx->model)) {
|
||||
if (llama_token_is_eog(&ctx->model, token)) {
|
||||
for (const auto & stack : grammar->stacks) {
|
||||
if (stack.empty()) {
|
||||
return;
|
||||
|
|
@ -13553,7 +13773,7 @@ void llama_grammar_accept_token(struct llama_context * ctx, struct llama_grammar
|
|||
GGML_ASSERT(false);
|
||||
}
|
||||
|
||||
const std::string piece = llama_token_to_piece(ctx, token);
|
||||
const std::string piece = llama_token_to_piece(ctx, token, false);
|
||||
|
||||
// Note terminating 0 in decoded string
|
||||
const auto decoded = decode_utf8(piece, grammar->partial_utf8);
|
||||
|
|
@ -15454,6 +15674,7 @@ enum llama_rope_type llama_rope_type(const struct llama_model * model) {
|
|||
case LLM_ARCH_QWEN2:
|
||||
case LLM_ARCH_QWEN2MOE:
|
||||
case LLM_ARCH_PHI2:
|
||||
case LLM_ARCH_PHI3:
|
||||
case LLM_ARCH_GEMMA:
|
||||
case LLM_ARCH_STARCODER2:
|
||||
return LLAMA_ROPE_TYPE_NEOX;
|
||||
|
|
@ -16969,6 +17190,13 @@ llama_token_type llama_token_get_type(const struct llama_model * model, llama_to
|
|||
return model->vocab.id_to_token[token].type;
|
||||
}
|
||||
|
||||
bool llama_token_is_eog(const struct llama_model * model, llama_token token) {
|
||||
return token != -1 && (
|
||||
token == llama_token_eos(model) ||
|
||||
token == llama_token_eot(model)
|
||||
);
|
||||
}
|
||||
|
||||
llama_token llama_token_bos(const struct llama_model * model) {
|
||||
return model->vocab.special_bos_id;
|
||||
}
|
||||
|
|
@ -17046,7 +17274,7 @@ static std::string llama_decode_text(const std::string & text) {
|
|||
}
|
||||
|
||||
// does not write null-terminator to buf
|
||||
int32_t llama_token_to_piece(const struct llama_model * model, llama_token token, char * buf, int32_t length) {
|
||||
int32_t llama_token_to_piece(const struct llama_model * model, llama_token token, char * buf, int32_t length, bool special) {
|
||||
if (0 <= token && token < llama_n_vocab(model)) {
|
||||
switch (llama_vocab_get_type(model->vocab)) {
|
||||
case LLAMA_VOCAB_TYPE_WPM:
|
||||
|
|
@ -17061,7 +17289,9 @@ int32_t llama_token_to_piece(const struct llama_model * model, llama_token token
|
|||
}
|
||||
memcpy(buf, result.c_str(), result.length());
|
||||
return result.length();
|
||||
} else if (llama_is_user_defined_token(model->vocab, token)) {
|
||||
} else if (
|
||||
(llama_is_user_defined_token(model->vocab, token)) ||
|
||||
(llama_is_control_token (model->vocab, token) && special)) {
|
||||
std::string result = model->vocab.id_to_token[token].text;
|
||||
if (length < (int) result.length()) {
|
||||
return -(int) result.length();
|
||||
|
|
@ -17074,8 +17304,6 @@ int32_t llama_token_to_piece(const struct llama_model * model, llama_token token
|
|||
}
|
||||
memcpy(buf, "\xe2\x96\x85", 3);
|
||||
return 3;
|
||||
} else if (llama_is_control_token(model->vocab, token)) {
|
||||
;
|
||||
} else if (llama_is_byte_token(model->vocab, token)) {
|
||||
if (length < 1) {
|
||||
return -1;
|
||||
|
|
@ -17096,15 +17324,15 @@ int32_t llama_token_to_piece(const struct llama_model * model, llama_token token
|
|||
}
|
||||
memcpy(buf, result.c_str(), result.length());
|
||||
return result.length();
|
||||
} else if (llama_is_user_defined_token(model->vocab, token)) {
|
||||
} else if (
|
||||
(llama_is_user_defined_token(model->vocab, token)) ||
|
||||
(llama_is_control_token (model->vocab, token) && special)) {
|
||||
std::string result = model->vocab.id_to_token[token].text;
|
||||
if (length < (int) result.length()) {
|
||||
return -(int) result.length();
|
||||
}
|
||||
memcpy(buf, result.c_str(), result.length());
|
||||
return result.length();
|
||||
} else if (llama_is_control_token(model->vocab, token)) {
|
||||
;
|
||||
}
|
||||
break;
|
||||
}
|
||||
|
|
@ -17302,6 +17530,24 @@ static int32_t llama_chat_apply_template_internal(
|
|||
if (add_ass) {
|
||||
ss << "<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>";
|
||||
}
|
||||
} else if (tmpl == "llama3" || (tmpl.find("<|start_header_id|>") != std::string::npos && tmpl.find("<|end_header_id|>") != std::string::npos)) {
|
||||
// Llama 3
|
||||
for (auto message : chat) {
|
||||
std::string role(message->role);
|
||||
ss << "<|start_header_id|>" << role << "<|end_header_id|>\n\n" << trim(message->content) << "<|eot_id|>";
|
||||
}
|
||||
if (add_ass) {
|
||||
ss << "<|start_header_id|>assistant<|end_header_id|>\n\n";
|
||||
}
|
||||
} else if (tmpl == "phi3" || (tmpl.find("<|assistant|>") != std::string::npos && tmpl.find("<|end|>") != std::string::npos )) {
|
||||
// Phi 3
|
||||
for (auto message : chat) {
|
||||
std::string role(message->role);
|
||||
ss << "<|" << role << "|>\n" << trim(message->content) << "<|end|>\n";
|
||||
}
|
||||
if (add_ass) {
|
||||
ss << "<|assistant|>\n";
|
||||
}
|
||||
} else {
|
||||
// template not supported
|
||||
return -1;
|
||||
|
|
|
|||
18
llama.h
18
llama.h
|
|
@ -784,6 +784,9 @@ extern "C" {
|
|||
|
||||
LLAMA_API enum llama_token_type llama_token_get_type(const struct llama_model * model, llama_token token);
|
||||
|
||||
// Check if the token is supposed to end generation (end-of-generation, eg. EOS, EOT, etc.)
|
||||
LLAMA_API bool llama_token_is_eog(const struct llama_model * model, llama_token token);
|
||||
|
||||
// Special tokens
|
||||
LLAMA_API llama_token llama_token_bos(const struct llama_model * model); // beginning-of-sentence
|
||||
LLAMA_API llama_token llama_token_eos(const struct llama_model * model); // end-of-sentence
|
||||
|
|
@ -797,7 +800,7 @@ extern "C" {
|
|||
// Returns -1 if unknown, 1 for true or 0 for false.
|
||||
LLAMA_API int32_t llama_add_eos_token(const struct llama_model * model);
|
||||
|
||||
// codellama infill tokens
|
||||
// Codellama infill tokens
|
||||
LLAMA_API llama_token llama_token_prefix(const struct llama_model * model); // Beginning of infill prefix
|
||||
LLAMA_API llama_token llama_token_middle(const struct llama_model * model); // Beginning of infill middle
|
||||
LLAMA_API llama_token llama_token_suffix(const struct llama_model * model); // Beginning of infill suffix
|
||||
|
|
@ -826,11 +829,13 @@ extern "C" {
|
|||
// Uses the vocabulary in the provided context.
|
||||
// Does not write null terminator to the buffer.
|
||||
// User code is responsible to remove the leading whitespace of the first non-BOS token when decoding multiple tokens.
|
||||
// @param special If true, special tokens are rendered in the output.
|
||||
LLAMA_API int32_t llama_token_to_piece(
|
||||
const struct llama_model * model,
|
||||
llama_token token,
|
||||
char * buf,
|
||||
int32_t length);
|
||||
int32_t length,
|
||||
bool special);
|
||||
|
||||
/// Apply chat template. Inspired by hf apply_chat_template() on python.
|
||||
/// Both "model" and "custom_template" are optional, but at least one is required. "custom_template" has higher precedence than "model"
|
||||
|
|
@ -983,7 +988,7 @@ extern "C" {
|
|||
struct llama_context * ctx,
|
||||
llama_token_data_array * candidates);
|
||||
|
||||
/// @details Randomly selects a token from the candidates based on their probabilities.
|
||||
/// @details Randomly selects a token from the candidates based on their probabilities using the RNG of ctx.
|
||||
LLAMA_API llama_token llama_sample_token(
|
||||
struct llama_context * ctx,
|
||||
llama_token_data_array * candidates);
|
||||
|
|
@ -1070,8 +1075,9 @@ extern "C" {
|
|||
// Internal API to be implemented by llama.cpp and used by tests/benchmarks only
|
||||
#ifdef LLAMA_API_INTERNAL
|
||||
|
||||
#include <vector>
|
||||
#include <random>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
|
||||
struct ggml_tensor;
|
||||
|
||||
|
|
@ -1108,6 +1114,10 @@ std::pair<std::vector<uint32_t>, llama_partial_utf8> decode_utf8(
|
|||
const std::string & src,
|
||||
llama_partial_utf8 partial_start);
|
||||
|
||||
// Randomly selects a token from the candidates based on their probabilities using given std::mt19937.
|
||||
// This is a temporary workaround in order to fix race conditions when sampling with multiple sequences.
|
||||
llama_token llama_sample_token_with_rng(struct llama_context * ctx, llama_token_data_array * candidates, std::mt19937 & rng);
|
||||
|
||||
#endif // LLAMA_API_INTERNAL
|
||||
|
||||
#endif // LLAMA_H
|
||||
|
|
|
|||
16
scripts/xxd.cmake
Normal file
16
scripts/xxd.cmake
Normal file
|
|
@ -0,0 +1,16 @@
|
|||
# CMake equivalent of `xxd -i ${INPUT} ${OUTPUT}`
|
||||
# Usage: cmake -DINPUT=examples/server/public/index.html -DOUTPUT=examples/server/index.html.hpp -P scripts/xxd.cmake
|
||||
|
||||
SET(INPUT "" CACHE STRING "Input File")
|
||||
SET(OUTPUT "" CACHE STRING "Output File")
|
||||
|
||||
get_filename_component(filename "${INPUT}" NAME)
|
||||
string(REGEX REPLACE "\\.|-" "_" name "${filename}")
|
||||
|
||||
file(READ "${INPUT}" hex_data HEX)
|
||||
string(REGEX REPLACE "([0-9a-f][0-9a-f])" "0x\\1," hex_sequence "${hex_data}")
|
||||
|
||||
string(LENGTH ${hex_data} hex_len)
|
||||
math(EXPR len "${hex_len} / 2")
|
||||
|
||||
file(WRITE "${OUTPUT}" "unsigned char ${name}[] = {${hex_sequence}};\nunsigned int ${name}_len = ${len};\n")
|
||||
|
|
@ -46,7 +46,11 @@ int main(void) {
|
|||
// No template included in tokenizer_config.json, so this template likely needs to be manually set.
|
||||
"{%- for message in messages %}{%- if message['role'] == 'system' -%}{{-'SYSTEM: ' + message['content'] + '\n' -}}{%- else -%}{%- if message['role'] == 'user' -%}{{-'USER: ' + message['content'] + '\n'-}}{%- else -%}{{-'ASSISTANT: ' + message['content'] + '</s>\n' -}}{%- endif -%}{%- endif -%}{%- endfor -%}{%- if add_generation_prompt -%}{{-'ASSISTANT:'-}}{%- endif -%}",
|
||||
// CohereForAI/c4ai-command-r-plus
|
||||
"{{ bos_token }}{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% elif false == true %}{% set loop_messages = messages %}{% set system_message = 'You are Command-R, a brilliant, sophisticated, AI-assistant trained to assist human users by providing thorough responses. You are trained by Cohere.' %}{% else %}{% set loop_messages = messages %}{% set system_message = false %}{% endif %}{% if system_message != false %}{{ '<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>' + system_message + '<|END_OF_TURN_TOKEN|>' }}{% endif %}{% for message in loop_messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|START_OF_TURN_TOKEN|><|USER_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}{% elif message['role'] == 'assistant' %}{{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>' }}{% endif %}"
|
||||
"{{ bos_token }}{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% elif false == true %}{% set loop_messages = messages %}{% set system_message = 'You are Command-R, a brilliant, sophisticated, AI-assistant trained to assist human users by providing thorough responses. You are trained by Cohere.' %}{% else %}{% set loop_messages = messages %}{% set system_message = false %}{% endif %}{% if system_message != false %}{{ '<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>' + system_message + '<|END_OF_TURN_TOKEN|>' }}{% endif %}{% for message in loop_messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|START_OF_TURN_TOKEN|><|USER_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}{% elif message['role'] == 'assistant' %}{{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>' }}{% endif %}",
|
||||
// Llama-3
|
||||
"{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}",
|
||||
// Phi-3
|
||||
"{{ bos_token }}{% for message in messages %}{{'<|' + message['role'] + '|>' + ' ' + message['content'] + '<|end|> ' }}{% endfor %}{% if add_generation_prompt %}{{ '<|assistant|> ' }}{% else %}{{ eos_token }}{% endif %}"
|
||||
};
|
||||
std::vector<std::string> expected_output = {
|
||||
// teknium/OpenHermes-2.5-Mistral-7B
|
||||
|
|
@ -73,6 +77,10 @@ int main(void) {
|
|||
"SYSTEM: You are a helpful assistant\nUSER: Hello\nASSISTANT: Hi there</s>\nUSER: Who are you\nASSISTANT: I am an assistant </s>\nUSER: Another question\nASSISTANT:",
|
||||
// CohereForAI/c4ai-command-r-plus
|
||||
"<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>You are a helpful assistant<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>Hello<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>Hi there<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>Who are you<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>I am an assistant<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>Another question<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>",
|
||||
// Llama 3
|
||||
"<|start_header_id|>system<|end_header_id|>\n\nYou are a helpful assistant<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nHello<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\nHi there<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nWho are you<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\nI am an assistant<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nAnother question<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n",
|
||||
// Phi 3
|
||||
"<|system|>\nYou are a helpful assistant<|end|>\n<|user|>\nHello<|end|>\n<|assistant|>\nHi there<|end|>\n<|user|>\nWho are you<|end|>\n<|assistant|>\nI am an assistant<|end|>\n<|user|>\nAnother question<|end|>\n<|assistant|>\n",
|
||||
};
|
||||
std::vector<char> formatted_chat(1024);
|
||||
int32_t res;
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue