vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

context : add decode/encode

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2025-02-10 16:11:17 +02:00
parent 879ba82777
commit ef358ee78f
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
3 changed files with 526 additions and 522 deletions
Download patch file Download diff file Expand all files Collapse all files

630
src/llama-context.cpp
View file

@ -9,6 +9,121 @@
#include <stdexcept>
#include <cinttypes>
// llama output (TMP)
// Make sure enough space is available for outputs.
// Returns max number of outputs for which space was reserved.
static size_t llama_output_reserve(struct llama_context & lctx, size_t n_outputs) {
const auto & cparams = lctx.cparams;
const auto & hparams = lctx.model.hparams;
const auto & vocab = lctx.model.vocab;
const size_t n_outputs_max = std::max(n_outputs, (size_t) cparams.n_seq_max);
const auto n_batch = cparams.n_batch;
const auto n_vocab = vocab.n_tokens();
const auto n_embd = hparams.n_embd;
// TODO: use a per-batch flag for logits presence instead
const bool has_logits = !cparams.embeddings;
const bool has_embd = cparams.embeddings && (cparams.pooling_type == LLAMA_POOLING_TYPE_NONE);
const size_t logits_size = has_logits ? n_vocab*n_outputs_max : 0;
const size_t embd_size = has_embd ? n_embd*n_outputs_max : 0;
if (lctx.output_ids.empty()) {
// init, never resized afterwards
lctx.output_ids.resize(n_batch);
}
const size_t prev_size = lctx.buf_output ? ggml_backend_buffer_get_size(lctx.buf_output.get()) : 0;
const size_t new_size = (logits_size + embd_size) * sizeof(float);
// alloc only when more than the current capacity is required
// TODO: also consider shrinking the buffer
if (!lctx.buf_output || prev_size < new_size) {
if (lctx.buf_output) {
#ifndef NDEBUG
// This doesn't happen often, but may be annoying in some cases (like the HellaSwag benchmark)
LLAMA_LOG_INFO("%s: reallocating output buffer from size %.02f MiB to %.02f MiB\n", __func__, prev_size / 1024.0 / 1024.0, new_size / 1024.0 / 1024.0);
#endif
lctx.buf_output = nullptr;
lctx.logits = nullptr;
lctx.embd = nullptr;
}
auto * buft = ggml_backend_cpu_buffer_type();
// try to use the host buffer of the device where the output tensor is allocated for faster transfer to system memory
auto * output_dev = lctx.model.dev_output();
auto * output_dev_host_buft = output_dev ? ggml_backend_dev_host_buffer_type(output_dev) : nullptr;
if (output_dev_host_buft) {
buft = output_dev_host_buft;
}
lctx.buf_output.reset(ggml_backend_buft_alloc_buffer(buft, new_size));
if (lctx.buf_output == nullptr) {
LLAMA_LOG_ERROR("%s: failed to allocate output buffer of size %.2f MiB\n", __func__, new_size / (1024.0 * 1024.0));
return 0;
}
}
float * output_base = (float *) ggml_backend_buffer_get_base(lctx.buf_output.get());
lctx.logits = has_logits ? output_base : nullptr;
lctx.embd = has_embd ? output_base + logits_size : nullptr;
lctx.output_size = n_outputs_max;
lctx.logits_size = logits_size;
lctx.embd_size = embd_size;
// set all ids as invalid (negative)
std::fill(lctx.output_ids.begin(), lctx.output_ids.end(), -1);
ggml_backend_buffer_clear(lctx.buf_output.get(), 0);
lctx.n_outputs = 0;
return n_outputs_max;
}
// make the outputs have the same order they had in the user-provided batch
static void llama_output_reorder(struct llama_context & ctx) {
std::vector<size_t> & out_ids = ctx.sbatch.out_ids;
if (!out_ids.empty()) {
const uint32_t n_vocab = ctx.model.vocab.n_tokens();
const uint32_t n_embd = ctx.model.hparams.n_embd;
const int32_t n_outputs = ctx.n_outputs;
GGML_ASSERT((size_t) n_outputs == out_ids.size());
// TODO: is there something more efficient which also minimizes swaps?
// selection sort, to minimize swaps (from https://en.wikipedia.org/wiki/Selection_sort)
for (int32_t i = 0; i < n_outputs - 1; ++i) {
int32_t j_min = i;
for (int32_t j = i + 1; j < n_outputs; ++j) {
if (out_ids[j] < out_ids[j_min]) {
j_min = j;
}
}
if (j_min == i) { continue; }
std::swap(out_ids[i], out_ids[j_min]);
if (ctx.logits_size > 0) {
for (uint32_t k = 0; k < n_vocab; k++) {
std::swap(ctx.logits[i*n_vocab + k], ctx.logits[j_min*n_vocab + k]);
}
}
if (ctx.embd_size > 0) {
for (uint32_t k = 0; k < n_embd; k++) {
std::swap(ctx.embd[i*n_embd + k], ctx.embd[j_min*n_embd + k]);
}
}
}
std::fill(ctx.output_ids.begin(), ctx.output_ids.end(), -1);
for (int32_t i = 0; i < n_outputs; ++i) {
ctx.output_ids[out_ids[i]] = i;
}
out_ids.clear();
}
}
static int32_t llama_relative_position_bucket(llama_pos x, llama_pos y, uint64_t n_buckets, bool bidirectional) {
// TODO move to hparams if a T5 variant appears that uses a different value
const int64_t max_distance = 128;
@ -340,6 +455,20 @@ llama_context::llama_context(
}
struct llama_batch_manager_i {
virtual ~llama_batch_manager_i() = default;
virtual bool is_done() const = 0;
virtual llama_ubatch next() = 0;
virtual bool prepare() = 0;
virtual void restore() = 0;
virtual void update() = 0;
virtual void finalize() = 0;
// TODO: might be temporary
int64_t n_outputs_all = 0;
};
struct llama_batch_manager : public llama_batch_manager_i {
llama_batch_manager(llama_context & lctx, const llama_batch & batch) : lctx(lctx), batch(batch), kv_slot_restorer(lctx.kv_self) {
const auto & model = lctx.model;
@ -398,6 +527,10 @@ struct llama_batch_manager : public llama_batch_manager_i {
~llama_batch_manager() override {
}
virtual bool is_done() const override {
return lctx.sbatch.n_tokens == 0;
}
virtual llama_ubatch next() override {
ubatch = llama_ubatch();
@ -558,6 +691,390 @@ std::unique_ptr<llama_batch_manager_i> llama_context::prepare_batch(const llama_
return std::make_unique<llama_batch_manager>(*this, batch);
}
int llama_context::decode(llama_batch & inp_batch) {
is_encoding = false;
if (inp_batch.n_tokens == 0) {
LLAMA_LOG_ERROR("%s: n_tokens == 0\n", __func__);
return -1;
}
// temporary allocate memory for the input batch if needed
// TODO: this is incorrect for multiple sequences because pos_max() is the maximum across all sequences
llama_batch_allocr batch_allocr(inp_batch, inp_batch.pos ? -1 : pos_max() + 1);
const llama_batch & batch = batch_allocr.batch;
const auto & vocab = model.vocab;
const auto & hparams = model.hparams;
const int32_t n_vocab = vocab.n_tokens();
const int64_t n_embd = hparams.n_embd;
// TODO: try catch
auto bman = prepare_batch(batch);
const auto n_outputs_all = bman->n_outputs_all;
// reserve output buffer
// TODO: move to batch manager?
if (llama_output_reserve(*this, bman->n_outputs_all) < (size_t) n_outputs_all) {
LLAMA_LOG_ERROR("%s: could not reserve space for batch with %" PRId64 " outputs\n", __func__, n_outputs_all);
return -2;
};
int64_t n_outputs_prev = 0;
while (!bman->is_done()) {
llama_ubatch ubatch = bman->next();
if (!bman->prepare()) {
LLAMA_LOG_ERROR("%s: failed to prepare ubatch\n", __func__);
bman->restore();
return -3;
}
ggml_backend_sched_reset(sched.get());
ggml_backend_sched_set_eval_callback(sched.get(), cparams.cb_eval, cparams.cb_eval_user_data);
ggml_cgraph * gf = cb_build_graph(*this, ubatch, false);
// LLAMA_LOG_INFO("graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf->n_nodes, gf->n_leafs);
ggml_backend_sched_alloc_graph(sched.get(), gf);
set_inputs(ubatch);
// the output is always the last tensor in the graph
struct ggml_tensor * t_logits = ggml_graph_node(gf, -1);
struct ggml_tensor * t_embd = ggml_graph_node(gf, -2);
if (n_outputs == 0) {
// no output
t_logits = nullptr;
t_embd = nullptr;
} else if (cparams.embeddings) {
t_logits = nullptr; // do not extract logits for embedding case
t_embd = nullptr;
for (int i = ggml_graph_n_nodes(gf) - 1; i >= 0; --i) {
if (strcmp(ggml_graph_node(gf, i)->name, "result_embd_pooled") == 0) {
t_embd = ggml_graph_node(gf, i);
break;
}
}
GGML_ASSERT(t_embd != nullptr && "missing embeddings tensor");
} else {
t_embd = nullptr; // do not extract embeddings when not needed
GGML_ASSERT(strcmp(t_logits->name, "result_output") == 0 && "missing result_output tensor");
}
const auto compute_status = compute_graph(gf, ubatch.n_tokens > 1);
if (compute_status != GGML_STATUS_SUCCESS) {
bman->restore();
switch (compute_status) {
case GGML_STATUS_ABORTED:
return 2;
case GGML_STATUS_ALLOC_FAILED:
return -2;
case GGML_STATUS_FAILED:
default:
return -3;
}
}
bman->update();
// plot the computation graph in dot format (for debugging purposes)
//if (n_past%100 == 0) {
// ggml_graph_dump_dot(gf, NULL, "llama.dot");
//}
// extract logits
if (t_logits) {
ggml_backend_t backend_res = ggml_backend_sched_get_tensor_backend(sched.get(), t_logits);
GGML_ASSERT(backend_res != nullptr);
GGML_ASSERT(logits != nullptr);
float * logits_out = logits + n_outputs_prev*n_vocab;
const int32_t n_outputs_new = n_outputs;
if (n_outputs_new) {
GGML_ASSERT( n_outputs_prev + n_outputs_new <= n_outputs_all);
GGML_ASSERT((n_outputs_prev + n_outputs_new)*n_vocab <= (int64_t) logits_size);
ggml_backend_tensor_get_async(backend_res, t_logits, logits_out, 0, n_outputs_new*n_vocab*sizeof(float));
}
}
// extract embeddings
if (t_embd) {
ggml_backend_t backend_embd = ggml_backend_sched_get_tensor_backend(sched.get(), t_embd);
GGML_ASSERT(backend_embd != nullptr);
switch (cparams.pooling_type) {
case LLAMA_POOLING_TYPE_NONE:
{
// extract token embeddings
GGML_ASSERT(embd != nullptr);
float * embd_out = embd + n_outputs_prev*n_embd;
const int32_t n_outputs_new = n_outputs;
if (n_outputs_new) {
GGML_ASSERT( n_outputs_prev + n_outputs_new <= n_outputs_all);
GGML_ASSERT((n_outputs_prev + n_outputs_new)*n_embd <= (int64_t) embd_size);
ggml_backend_tensor_get_async(backend_embd, t_embd, embd_out, 0, n_outputs_new*n_embd*sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_MEAN:
case LLAMA_POOLING_TYPE_CLS:
case LLAMA_POOLING_TYPE_LAST:
{
// extract sequence embeddings (cleared before processing each batch)
auto & embd_seq_out = embd_seq;
for (uint32_t s = 0; s < ubatch.n_seqs; ++s) {
const llama_seq_id seq_id = ubatch.seq_id[s][0];
if (embd_seq_out.find(seq_id) != embd_seq_out.end()) {
continue;
}
embd_seq_out[seq_id].resize(n_embd);
ggml_backend_tensor_get_async(backend_embd, t_embd, embd_seq_out[seq_id].data(), (n_embd*seq_id)*sizeof(float), n_embd*sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_RANK:
{
// extract the rerank score - a single float per sequence
auto & embd_seq_out = embd_seq;
for (uint32_t s = 0; s < ubatch.n_seqs; ++s) {
const llama_seq_id seq_id = ubatch.seq_id[s][0];
if (embd_seq_out.find(seq_id) != embd_seq_out.end()) {
continue;
}
embd_seq_out[seq_id].resize(1);
ggml_backend_tensor_get_async(backend_embd, t_embd, embd_seq_out[seq_id].data(), (seq_id)*sizeof(float), sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_UNSPECIFIED:
{
GGML_ABORT("unknown pooling type");
}
}
}
n_outputs_prev += n_outputs;
}
// set output mappings
{
bool sorted_output = true;
GGML_ASSERT(sbatch.out_ids.size() == (size_t) n_outputs_all);
for (size_t i = 0; i < (size_t) n_outputs_all; ++i) {
size_t out_id = sbatch.out_ids[i];
output_ids[out_id] = i;
if (out_id != i) {
sorted_output = false;
}
}
if (sorted_output) {
sbatch.out_ids.clear();
}
}
// set to total number of outputs in the batch, for use in llama_get_logits_ith
n_outputs = n_outputs_all;
// wait for the computation to finish (automatically done when obtaining the model output)
//llama_synchronize(&;
bman->finalize();
// Reset state for the next token before backend sync, to allow the CPU activities in the reset to
// overlap with device computation.
ggml_backend_sched_reset(sched.get());
return 0;
}
int llama_context::encode(llama_batch & inp_batch) {
is_encoding = true;
if (inp_batch.n_tokens == 0) {
LLAMA_LOG_ERROR("%s: n_tokens == 0\n", __func__);
return -1;
}
// temporary allocate memory for the input batch if needed
// TODO: this is incorrect for multiple sequences because pos_max() is the maximum across all sequences
llama_batch_allocr batch_allocr(inp_batch, inp_batch.pos ? -1 : pos_max() + 1);
const llama_batch & batch = batch_allocr.batch;
const uint32_t n_tokens = batch.n_tokens;
const auto & hparams = model.hparams;
GGML_ASSERT((!batch.token && batch.embd) || (batch.token && !batch.embd)); // NOLINT
if (batch.token) {
for (uint32_t i = 0; i < n_tokens; ++i) {
if (batch.token[i] < 0 || (uint32_t) batch.token[i] >= model.vocab.n_tokens()) {
LLAMA_LOG_ERROR("%s: invalid token[%d] = %d\n", __func__, i, batch.token[i]);
return -1;
}
}
}
// micro-batching is not possible for non-causal encoding, so we process the batch in a single shot
GGML_ASSERT(cparams.n_ubatch >= n_tokens && "encoder requires n_ubatch >= n_tokens");
if (t_compute_start_us == 0) {
t_compute_start_us = ggml_time_us();
}
n_queued_tokens += n_tokens;
const int64_t n_embd = hparams.n_embd;
sbatch.from_batch(batch, n_embd, /* simple_split */ true, /* logits_all */ true);
const llama_ubatch ubatch = sbatch.split_simple(n_tokens);
// reserve output buffer
if (llama_output_reserve(*this, n_tokens) < n_tokens) {
LLAMA_LOG_ERROR("%s: could not reserve space for batch with %u outputs\n", __func__, n_tokens);
return -2;
};
for (uint32_t i = 0; i < n_tokens; ++i) {
output_ids[i] = i;
}
inp_embd_enc = NULL;
n_outputs = n_tokens;
//batch_manager->prepare(ubatch);
// TODO: do reserve
GGML_ASSERT(need_reserve == false);
ggml_backend_sched_reset(sched.get());
ggml_backend_sched_set_eval_callback(sched.get(), cparams.cb_eval, cparams.cb_eval_user_data);
ggml_cgraph * gf = cb_build_graph(*this, ubatch, false);
ggml_backend_sched_alloc_graph(sched.get(), gf);
set_inputs(ubatch);
// the output embeddings after the final encoder normalization
struct ggml_tensor * t_embd = nullptr;
// there are two cases here
if (llama_model_has_decoder(&model)) {
// first case is an encoder-decoder T5 model where embeddings are passed to decoder
t_embd = ggml_graph_node(gf, -1);
GGML_ASSERT(strcmp(t_embd->name, "result_norm") == 0 && "missing result_output tensor");
} else {
// second case is an encoder-only T5 model
if (cparams.embeddings) {
// only output embeddings if required
t_embd = ggml_graph_node(gf, -1);
if (strcmp(t_embd->name, "result_embd_pooled") != 0) {
t_embd = ggml_graph_node(gf, -2);
}
GGML_ASSERT(strcmp(t_embd->name, "result_embd_pooled") == 0 && "missing embeddings tensor");
}
}
const auto compute_status = compute_graph(gf, n_tokens > 1);
switch (compute_status) {
case GGML_STATUS_SUCCESS:
break;
case GGML_STATUS_ABORTED:
return 2;
case GGML_STATUS_ALLOC_FAILED:
return -2;
case GGML_STATUS_FAILED:
default:
return -3;
}
// extract embeddings
if (t_embd) {
ggml_backend_t backend_embd = ggml_backend_sched_get_tensor_backend(sched.get(), t_embd);
GGML_ASSERT(backend_embd != nullptr);
if (llama_model_has_decoder(&model)) {
embd_enc.resize(n_tokens*n_embd);
float * embd_out = embd_enc.data();
ggml_backend_tensor_get_async(backend_embd, t_embd, embd_out, 0, n_tokens*n_embd*sizeof(float));
GGML_ASSERT(!ubatch.equal_seqs); // TODO: handle equal splits
// remember the sequence ids used during the encoding - needed for cross attention later
seq_ids_enc.resize(n_tokens);
for (uint32_t i = 0; i < n_tokens; i++) {
for (int s = 0; s < ubatch.n_seq_id[i]; s++) {
llama_seq_id seq_id = ubatch.seq_id[i][s];
seq_ids_enc[i].insert(seq_id);
}
}
} else {
GGML_ASSERT(embd != nullptr);
switch (cparams.pooling_type) {
case LLAMA_POOLING_TYPE_NONE:
{
// extract token embeddings
GGML_ASSERT(embd != nullptr);
float * embd_out = embd;
GGML_ASSERT(n_tokens*n_embd <= (int64_t) embd_size);
ggml_backend_tensor_get_async(backend_embd, t_embd, embd_out, 0, n_tokens*n_embd*sizeof(float));
} break;
case LLAMA_POOLING_TYPE_MEAN:
case LLAMA_POOLING_TYPE_CLS:
case LLAMA_POOLING_TYPE_LAST:
{
// extract sequence embeddings
auto & embd_seq_out = embd_seq;
embd_seq_out.clear();
GGML_ASSERT(!ubatch.equal_seqs); // TODO: handle equal splits
for (uint32_t i = 0; i < n_tokens; i++) {
const llama_seq_id seq_id = ubatch.seq_id[i][0];
if (embd_seq_out.find(seq_id) != embd_seq_out.end()) {
continue;
}
embd_seq_out[seq_id].resize(n_embd);
ggml_backend_tensor_get_async(backend_embd, t_embd, embd_seq_out[seq_id].data(), (n_embd*seq_id)*sizeof(float), n_embd*sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_RANK:
{
// TODO: this likely should be the same logic as in llama_decoder_internal, but better to
// wait for an encoder model that requires this pooling type in order to test it
// https://github.com/ggerganov/llama.cpp/pull/9510
GGML_ABORT("RANK pooling not implemented yet");
}
case LLAMA_POOLING_TYPE_UNSPECIFIED:
{
GGML_ABORT("unknown pooling type");
}
}
}
}
// Reset state for the next token before backend sync, to allow the CPU activities in the reset to
// overlap with device computation.
ggml_backend_sched_reset(sched.get());
return 0;
}
enum ggml_status llama_context::compute_graph(
ggml_cgraph * graph,
bool batched) {
@ -2194,119 +2711,6 @@ ggml_tensor * llama_context::build_rwkv6_time_mix(
return cur;
}
// llama output
size_t llama_output_reserve(struct llama_context & lctx, size_t n_outputs) {
const auto & cparams = lctx.cparams;
const auto & hparams = lctx.model.hparams;
const auto & vocab = lctx.model.vocab;
const size_t n_outputs_max = std::max(n_outputs, (size_t) cparams.n_seq_max);
const auto n_batch = cparams.n_batch;
const auto n_vocab = vocab.n_tokens();
const auto n_embd = hparams.n_embd;
// TODO: use a per-batch flag for logits presence instead
const bool has_logits = !cparams.embeddings;
const bool has_embd = cparams.embeddings && (cparams.pooling_type == LLAMA_POOLING_TYPE_NONE);
const size_t logits_size = has_logits ? n_vocab*n_outputs_max : 0;
const size_t embd_size = has_embd ? n_embd*n_outputs_max : 0;
if (lctx.output_ids.empty()) {
// init, never resized afterwards
lctx.output_ids.resize(n_batch);
}
const size_t prev_size = lctx.buf_output ? ggml_backend_buffer_get_size(lctx.buf_output.get()) : 0;
const size_t new_size = (logits_size + embd_size) * sizeof(float);
// alloc only when more than the current capacity is required
// TODO: also consider shrinking the buffer
if (!lctx.buf_output || prev_size < new_size) {
if (lctx.buf_output) {
#ifndef NDEBUG
// This doesn't happen often, but may be annoying in some cases (like the HellaSwag benchmark)
LLAMA_LOG_INFO("%s: reallocating output buffer from size %.02f MiB to %.02f MiB\n", __func__, prev_size / 1024.0 / 1024.0, new_size / 1024.0 / 1024.0);
#endif
lctx.buf_output = nullptr;
lctx.logits = nullptr;
lctx.embd = nullptr;
}
auto * buft = ggml_backend_cpu_buffer_type();
// try to use the host buffer of the device where the output tensor is allocated for faster transfer to system memory
auto * output_dev = lctx.model.dev_output();
auto * output_dev_host_buft = output_dev ? ggml_backend_dev_host_buffer_type(output_dev) : nullptr;
if (output_dev_host_buft) {
buft = output_dev_host_buft;
}
lctx.buf_output.reset(ggml_backend_buft_alloc_buffer(buft, new_size));
if (lctx.buf_output == nullptr) {
LLAMA_LOG_ERROR("%s: failed to allocate output buffer of size %.2f MiB\n", __func__, new_size / (1024.0 * 1024.0));
return 0;
}
}
float * output_base = (float *) ggml_backend_buffer_get_base(lctx.buf_output.get());
lctx.logits = has_logits ? output_base : nullptr;
lctx.embd = has_embd ? output_base + logits_size : nullptr;
lctx.output_size = n_outputs_max;
lctx.logits_size = logits_size;
lctx.embd_size = embd_size;
// set all ids as invalid (negative)
std::fill(lctx.output_ids.begin(), lctx.output_ids.end(), -1);
ggml_backend_buffer_clear(lctx.buf_output.get(), 0);
lctx.n_outputs = 0;
return n_outputs_max;
}
void llama_output_reorder(struct llama_context & ctx) {
std::vector<size_t> & out_ids = ctx.sbatch.out_ids;
if (!out_ids.empty()) {
const uint32_t n_vocab = ctx.model.vocab.n_tokens();
const uint32_t n_embd = ctx.model.hparams.n_embd;
const int32_t n_outputs = ctx.n_outputs;
GGML_ASSERT((size_t) n_outputs == out_ids.size());
// TODO: is there something more efficient which also minimizes swaps?
// selection sort, to minimize swaps (from https://en.wikipedia.org/wiki/Selection_sort)
for (int32_t i = 0; i < n_outputs - 1; ++i) {
int32_t j_min = i;
for (int32_t j = i + 1; j < n_outputs; ++j) {
if (out_ids[j] < out_ids[j_min]) {
j_min = j;
}
}
if (j_min == i) { continue; }
std::swap(out_ids[i], out_ids[j_min]);
if (ctx.logits_size > 0) {
for (uint32_t k = 0; k < n_vocab; k++) {
std::swap(ctx.logits[i*n_vocab + k], ctx.logits[j_min*n_vocab + k]);
}
}
if (ctx.embd_size > 0) {
for (uint32_t k = 0; k < n_embd; k++) {
std::swap(ctx.embd[i*n_embd + k], ctx.embd[j_min*n_embd + k]);
}
}
}
std::fill(ctx.output_ids.begin(), ctx.output_ids.end(), -1);
for (int32_t i = 0; i < n_outputs; ++i) {
ctx.output_ids[out_ids[i]] = i;
}
out_ids.clear();
}
}
//
// interface implementation
//

32
src/llama-context.h
View file

@ -16,22 +16,7 @@
using llama_loras = std::unordered_map<struct llama_adapter_lora *, float>;
// TODO: this is very WIP - improve
struct llama_batch_manager_i {
virtual ~llama_batch_manager_i() = default;
//bool is_done() const;
virtual llama_ubatch next() = 0;
virtual bool prepare() = 0;
virtual void restore() = 0;
virtual void update() = 0;
virtual void finalize() = 0;
// TODO: might be temporary
int64_t n_outputs_all = 0;
};
struct llama_batch_manager_i;
// TODO: make implementation details private
// TODO: become abstract base class, split the current implementation into different child classes
@ -44,6 +29,8 @@ struct llama_context {
const llama_context_params & params,
build_graph_callback && cb_build_graph);
virtual ~llama_context() = default;
const struct llama_model & model;
llama_cparams cparams;
@ -104,8 +91,10 @@ struct llama_context {
ggml_abort_callback abort_callback = nullptr;
void * abort_callback_data = nullptr;
// TODO: do not pass logits_all explicitly
std::unique_ptr<llama_batch_manager_i> prepare_batch(const llama_batch & batch);
virtual std::unique_ptr<llama_batch_manager_i> prepare_batch(const llama_batch & batch);
virtual int decode(llama_batch & inp_batch);
virtual int encode(llama_batch & inp_batch);
// returns the result of ggml_backend_sched_graph_compute_async execution
enum ggml_status compute_graph(
@ -286,13 +275,6 @@ struct llama_context {
int n_pos_per_token = 1;
};
// Make sure enough space is available for outputs.
// Returns max number of outputs for which space was reserved.
size_t llama_output_reserve(struct llama_context & lctx, size_t n_outputs);
// make the outputs have the same order they had in the user-provided batch
void llama_output_reorder(struct llama_context & ctx);
// For internal test use
// TODO: remove
const std::vector<std::pair<std::string, struct ggml_tensor *>> & llama_internal_get_tensor_map(struct llama_context * ctx);

386
src/llama.cpp
View file

@ -7584,213 +7584,7 @@ static struct ggml_cgraph * llama_build_graph(
static int llama_decode_impl(
llama_context & lctx,
llama_batch inp_batch) {
lctx.is_encoding = false;
if (inp_batch.n_tokens == 0) {
LLAMA_LOG_ERROR("%s: n_tokens == 0\n", __func__);
return -1;
}
// temporary allocate memory for the input batch if needed
// TODO: this is incorrect for multiple sequences because pos_max() is the maximum across all sequences
llama_batch_allocr batch_allocr(inp_batch, inp_batch.pos ? -1 : lctx.pos_max() + 1);
const llama_batch & batch = batch_allocr.batch;
const auto & model = lctx.model;
const auto & vocab = model.vocab;
const auto & cparams = lctx.cparams;
const auto & hparams = lctx.model.hparams;
const int32_t n_vocab = vocab.n_tokens();
const int64_t n_embd = hparams.n_embd;
// TODO: try catch
auto bman = lctx.prepare_batch(batch);
const auto n_outputs_all = bman->n_outputs_all;
// reserve output buffer
// TODO: move to batch manager?
if (llama_output_reserve(lctx, bman->n_outputs_all) < (size_t) n_outputs_all) {
LLAMA_LOG_ERROR("%s: could not reserve space for batch with %" PRId64 " outputs\n", __func__, n_outputs_all);
return -2;
};
int64_t n_outputs_prev = 0;
while (lctx.sbatch.n_tokens > 0) {
llama_ubatch ubatch = bman->next();
if (!bman->prepare()) {
LLAMA_LOG_ERROR("%s: failed to prepare ubatch\n", __func__);
bman->restore();
return -3;
}
ggml_backend_sched_reset(lctx.sched.get());
ggml_backend_sched_set_eval_callback(lctx.sched.get(), lctx.cparams.cb_eval, lctx.cparams.cb_eval_user_data);
ggml_cgraph * gf = llama_build_graph(lctx, ubatch, false);
// LLAMA_LOG_INFO("graph build time: %.3f ms (%d nodes, %d leafs)\n", (ggml_time_us() - t_start_us)/1000.0, gf->n_nodes, gf->n_leafs);
ggml_backend_sched_alloc_graph(lctx.sched.get(), gf);
lctx.set_inputs(ubatch);
// the output is always the last tensor in the graph
struct ggml_tensor * res = ggml_graph_node(gf, -1);
struct ggml_tensor * embd = ggml_graph_node(gf, -2);
if (lctx.n_outputs == 0) {
// no output
res = nullptr;
embd = nullptr;
} else if (cparams.embeddings) {
res = nullptr; // do not extract logits for embedding case
embd = nullptr;
for (int i = ggml_graph_n_nodes(gf) - 1; i >= 0; --i) {
if (strcmp(ggml_graph_node(gf, i)->name, "result_embd_pooled") == 0) {
embd = ggml_graph_node(gf, i);
break;
}
}
GGML_ASSERT(embd != nullptr && "missing embeddings tensor");
} else {
embd = nullptr; // do not extract embeddings when not needed
GGML_ASSERT(strcmp(res->name, "result_output") == 0 && "missing result_output tensor");
}
const auto compute_status = lctx.compute_graph(gf, ubatch.n_tokens > 1);
if (compute_status != GGML_STATUS_SUCCESS) {
bman->restore();
switch (compute_status) {
case GGML_STATUS_ABORTED:
return 2;
case GGML_STATUS_ALLOC_FAILED:
return -2;
case GGML_STATUS_FAILED:
default:
return -3;
}
}
bman->update();
// plot the computation graph in dot format (for debugging purposes)
//if (n_past%100 == 0) {
// ggml_graph_dump_dot(gf, NULL, "llama.dot");
//}
// extract logits
if (res) {
ggml_backend_t backend_res = ggml_backend_sched_get_tensor_backend(lctx.sched.get(), res);
GGML_ASSERT(backend_res != nullptr);
GGML_ASSERT(lctx.logits != nullptr);
float * logits_out = lctx.logits + n_outputs_prev*n_vocab;
const int32_t n_outputs_new = lctx.n_outputs;
if (n_outputs_new) {
GGML_ASSERT( n_outputs_prev + n_outputs_new <= n_outputs_all);
GGML_ASSERT((n_outputs_prev + n_outputs_new)*n_vocab <= (int64_t) lctx.logits_size);
ggml_backend_tensor_get_async(backend_res, res, logits_out, 0, n_outputs_new*n_vocab*sizeof(float));
}
}
// extract embeddings
if (embd) {
ggml_backend_t backend_embd = ggml_backend_sched_get_tensor_backend(lctx.sched.get(), embd);
GGML_ASSERT(backend_embd != nullptr);
switch (cparams.pooling_type) {
case LLAMA_POOLING_TYPE_NONE:
{
// extract token embeddings
GGML_ASSERT(lctx.embd != nullptr);
float * embd_out = lctx.embd + n_outputs_prev*n_embd;
const int32_t n_outputs_new = lctx.n_outputs;
if (n_outputs_new) {
GGML_ASSERT( n_outputs_prev + n_outputs_new <= n_outputs_all);
GGML_ASSERT((n_outputs_prev + n_outputs_new)*n_embd <= (int64_t) lctx.embd_size);
ggml_backend_tensor_get_async(backend_embd, embd, embd_out, 0, n_outputs_new*n_embd*sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_MEAN:
case LLAMA_POOLING_TYPE_CLS:
case LLAMA_POOLING_TYPE_LAST:
{
// extract sequence embeddings (cleared before processing each batch)
auto & embd_seq_out = lctx.embd_seq;
for (uint32_t s = 0; s < ubatch.n_seqs; ++s) {
const llama_seq_id seq_id = ubatch.seq_id[s][0];
if (embd_seq_out.find(seq_id) != embd_seq_out.end()) {
continue;
}
embd_seq_out[seq_id].resize(n_embd);
ggml_backend_tensor_get_async(backend_embd, embd, embd_seq_out[seq_id].data(), (n_embd*seq_id)*sizeof(float), n_embd*sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_RANK:
{
// extract the rerank score - a single float per sequence
auto & embd_seq_out = lctx.embd_seq;
for (uint32_t s = 0; s < ubatch.n_seqs; ++s) {
const llama_seq_id seq_id = ubatch.seq_id[s][0];
if (embd_seq_out.find(seq_id) != embd_seq_out.end()) {
continue;
}
embd_seq_out[seq_id].resize(1);
ggml_backend_tensor_get_async(backend_embd, embd, embd_seq_out[seq_id].data(), (seq_id)*sizeof(float), sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_UNSPECIFIED:
{
GGML_ABORT("unknown pooling type");
}
}
}
n_outputs_prev += lctx.n_outputs;
}
// set output mappings
{
bool sorted_output = true;
GGML_ASSERT(lctx.sbatch.out_ids.size() == (size_t) n_outputs_all);
for (size_t i = 0; i < (size_t) n_outputs_all; ++i) {
size_t out_id = lctx.sbatch.out_ids[i];
lctx.output_ids[out_id] = i;
if (out_id != i) {
sorted_output = false;
}
}
if (sorted_output) {
lctx.sbatch.out_ids.clear();
}
}
// set to total number of outputs in the batch, for use in llama_get_logits_ith
lctx.n_outputs = n_outputs_all;
// wait for the computation to finish (automatically done when obtaining the model output)
//llama_synchronize(&lctx);
bman->finalize();
// Reset state for the next token before backend sync, to allow the CPU activities in the reset to
// overlap with device computation.
ggml_backend_sched_reset(lctx.sched.get());
return 0;
return lctx.decode(inp_batch);
}
// encode a batch of tokens by evaluating the encoder part of the transformer
@ -7805,183 +7599,7 @@ static int llama_decode_impl(
static int llama_encode_impl(
llama_context & lctx,
llama_batch inp_batch) {
lctx.is_encoding = true;
if (inp_batch.n_tokens == 0) {
LLAMA_LOG_ERROR("%s: n_tokens == 0\n", __func__);
return -1;
}
// temporary allocate memory for the input batch if needed
// TODO: this is incorrect for multiple sequences because pos_max() is the maximum across all sequences
llama_batch_allocr batch_allocr(inp_batch, inp_batch.pos ? -1 : lctx.pos_max() + 1);
const llama_batch & batch = batch_allocr.batch;
const uint32_t n_tokens = batch.n_tokens;
const auto & model = lctx.model;
const auto & hparams = model.hparams;
const auto & cparams = lctx.cparams;
GGML_ASSERT((!batch.token && batch.embd) || (batch.token && !batch.embd)); // NOLINT
if (batch.token) {
for (uint32_t i = 0; i < n_tokens; ++i) {
if (batch.token[i] < 0 || (uint32_t) batch.token[i] >= model.vocab.n_tokens()) {
LLAMA_LOG_ERROR("%s: invalid token[%d] = %d\n", __func__, i, batch.token[i]);
return -1;
}
}
}
// micro-batching is not possible for non-causal encoding, so we process the batch in a single shot
GGML_ASSERT(cparams.n_ubatch >= n_tokens && "encoder requires n_ubatch >= n_tokens");
if (lctx.t_compute_start_us == 0) {
lctx.t_compute_start_us = ggml_time_us();
}
lctx.n_queued_tokens += n_tokens;
const int64_t n_embd = hparams.n_embd;
lctx.sbatch.from_batch(batch, n_embd, /* simple_split */ true, /* logits_all */ true);
const llama_ubatch ubatch = lctx.sbatch.split_simple(n_tokens);
// reserve output buffer
if (llama_output_reserve(lctx, n_tokens) < n_tokens) {
LLAMA_LOG_ERROR("%s: could not reserve space for batch with %u outputs\n", __func__, n_tokens);
return -2;
};
for (uint32_t i = 0; i < n_tokens; ++i) {
lctx.output_ids[i] = i;
}
lctx.inp_embd_enc = NULL;
lctx.n_outputs = n_tokens;
//batch_manager->prepare(ubatch);
// TODO: do reserve
GGML_ASSERT(lctx.need_reserve == false);
ggml_backend_sched_reset(lctx.sched.get());
ggml_backend_sched_set_eval_callback(lctx.sched.get(), lctx.cparams.cb_eval, lctx.cparams.cb_eval_user_data);
ggml_cgraph * gf = llama_build_graph(lctx, ubatch, false);
ggml_backend_sched_alloc_graph(lctx.sched.get(), gf);
lctx.set_inputs(ubatch);
// the output embeddings after the final encoder normalization
struct ggml_tensor * embd = nullptr;
// there are two cases here
if (llama_model_has_decoder(&lctx.model)) {
// first case is an encoder-decoder T5 model where embeddings are passed to decoder
embd = ggml_graph_node(gf, -1);
GGML_ASSERT(strcmp(embd->name, "result_norm") == 0 && "missing result_output tensor");
} else {
// second case is an encoder-only T5 model
if (cparams.embeddings) {
// only output embeddings if required
embd = ggml_graph_node(gf, -1);
if (strcmp(embd->name, "result_embd_pooled") != 0) {
embd = ggml_graph_node(gf, -2);
}
GGML_ASSERT(strcmp(embd->name, "result_embd_pooled") == 0 && "missing embeddings tensor");
}
}
const auto compute_status = lctx.compute_graph(gf, n_tokens > 1);
switch (compute_status) {
case GGML_STATUS_SUCCESS:
break;
case GGML_STATUS_ABORTED:
return 2;
case GGML_STATUS_ALLOC_FAILED:
return -2;
case GGML_STATUS_FAILED:
default:
return -3;
}
// extract embeddings
if (embd) {
ggml_backend_t backend_embd = ggml_backend_sched_get_tensor_backend(lctx.sched.get(), embd);
GGML_ASSERT(backend_embd != nullptr);
if (llama_model_has_decoder(&lctx.model)) {
lctx.embd_enc.resize(n_tokens*n_embd);
float * embd_out = lctx.embd_enc.data();
ggml_backend_tensor_get_async(backend_embd, embd, embd_out, 0, n_tokens*n_embd*sizeof(float));
GGML_ASSERT(!ubatch.equal_seqs); // TODO: handle equal splits
// remember the sequence ids used during the encoding - needed for cross attention later
lctx.seq_ids_enc.resize(n_tokens);
for (uint32_t i = 0; i < n_tokens; i++) {
for (int s = 0; s < ubatch.n_seq_id[i]; s++) {
llama_seq_id seq_id = ubatch.seq_id[i][s];
lctx.seq_ids_enc[i].insert(seq_id);
}
}
} else {
GGML_ASSERT(lctx.embd != nullptr);
switch (cparams.pooling_type) {
case LLAMA_POOLING_TYPE_NONE:
{
// extract token embeddings
GGML_ASSERT(lctx.embd != nullptr);
float * embd_out = lctx.embd;
GGML_ASSERT(n_tokens*n_embd <= (int64_t) lctx.embd_size);
ggml_backend_tensor_get_async(backend_embd, embd, embd_out, 0, n_tokens*n_embd*sizeof(float));
} break;
case LLAMA_POOLING_TYPE_MEAN:
case LLAMA_POOLING_TYPE_CLS:
case LLAMA_POOLING_TYPE_LAST:
{
// extract sequence embeddings
auto & embd_seq_out = lctx.embd_seq;
embd_seq_out.clear();
GGML_ASSERT(!ubatch.equal_seqs); // TODO: handle equal splits
for (uint32_t i = 0; i < n_tokens; i++) {
const llama_seq_id seq_id = ubatch.seq_id[i][0];
if (embd_seq_out.find(seq_id) != embd_seq_out.end()) {
continue;
}
embd_seq_out[seq_id].resize(n_embd);
ggml_backend_tensor_get_async(backend_embd, embd, embd_seq_out[seq_id].data(), (n_embd*seq_id)*sizeof(float), n_embd*sizeof(float));
}
} break;
case LLAMA_POOLING_TYPE_RANK:
{
// TODO: this likely should be the same logic as in llama_decoder_internal, but better to
// wait for an encoder model that requires this pooling type in order to test it
// https://github.com/ggerganov/llama.cpp/pull/9510
GGML_ABORT("RANK pooling not implemented yet");
}
case LLAMA_POOLING_TYPE_UNSPECIFIED:
{
GGML_ABORT("unknown pooling type");
}
}
}
}
// Reset state for the next token before backend sync, to allow the CPU activities in the reset to
// overlap with device computation.
ggml_backend_sched_reset(lctx.sched.get());
return 0;
return lctx.encode(inp_batch);
}
//