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

llama : KV cache view API + better KV cache management (#4170)

Browse source 
* llama : keep track of used KV cells + better KV cache management

* llama : zero KV cache used upon clear

ggml-ci

* llama : allow exporting a view of the KV cache (#4180)

* Allow exporting a view of the KV cache

* Allow dumping the sequences per cell in common

* Track max contiguous cells value and position as well

* Fix max contiguous empty cells index calculation

Make dump functions deal with lengths or sequences counts > 10 better

* Fix off by one error in dump_kv_cache_view

* Add doc comments for KV cache view functions

Eliminate cell sequence struct; use llama_seq_id directly

Minor cleanups

* common : add -dkvc arg for enabling kv cache dumps

---------

Co-authored-by: Kerfuffle <44031344+KerfuffleV2@users.noreply.github.com>
This commit is contained in:
Georgi Gerganov 2023-11-23 19:07:56 +02:00 • committed by GitHub
parent d103d935c0
commit 6b0a7420d0
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 277 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

79
common/common.cpp
View file

@ -12,6 +12,7 @@
#include <regex>
#include <sstream>
#include <string>
#include <unordered_map>
#include <unordered_set>
#include <vector>
#include <cinttypes>
@ -495,6 +496,8 @@ bool gpt_params_parse_ex(int argc, char ** argv, gpt_params & params) {
params.chatml = true;
} else if (arg == "--infill") {
params.infill = true;
} else if (arg == "-dkvc" || arg == "--dump-kv-cache") {
params.dump_kv_cache = true;
} else if (arg == "--multiline-input") {
params.multiline_input = true;
} else if (arg == "--simple-io") {
@ -835,6 +838,8 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
#endif // GGML_USE_CUBLAS
#endif
printf(" --verbose-prompt print prompt before generation\n");
printf(" -dkvc, --dump-kv-cache\n");
printf(" verbose print of the KV cache\n");
printf(" --simple-io use basic IO for better compatibility in subprocesses and limited consoles\n");
printf(" --lora FNAME apply LoRA adapter (implies --no-mmap)\n");
printf(" --lora-scaled FNAME S apply LoRA adapter with user defined scaling S (implies --no-mmap)\n");
@ -1386,3 +1391,77 @@ void dump_non_result_info_yaml(FILE * stream, const gpt_params & params, const l
fprintf(stream, "typical_p: %f # default: 1.0\n", sparams.typical_p);
fprintf(stream, "verbose_prompt: %s # default: false\n", params.verbose_prompt ? "true" : "false");
}
//
// KV cache utils
//
void dump_kv_cache_view(const llama_kv_cache_view & view, int row_size) {
static const char slot_chars[] = ".123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz+";
printf("=== Dumping KV cache. total cells %d, max sequences per cell %d, populated cells %d, total tokens in cache %d, largest empty slot=%d @ %d",
view.n_cells, view.n_max_seq, view.used_cells, view.token_count, view.max_contiguous, view.max_contiguous_idx);
llama_kv_cache_view_cell * c_curr = view.cells;
llama_seq_id * cs_curr = view.cells_sequences;
for (int i = 0; i < view.n_cells; i++, c_curr++, cs_curr += view.n_max_seq) {
if (i % row_size == 0) {
printf("\n%5d: ", i);
}
int seq_count = 0;
for (int j = 0; j < view.n_max_seq; j++) {
if (cs_curr[j] >= 0) { seq_count++; }
}
putchar(slot_chars[std::min(sizeof(slot_chars) - 2, size_t(seq_count))]);
}
printf("\n=== Done dumping\n");
}
void dump_kv_cache_view_seqs(const llama_kv_cache_view & view, int row_size) {
static const char slot_chars[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
printf("=== Dumping KV cache. total cells %d, max sequences per cell %d, populated cells %d, total tokens in cache %d, largest empty slot=%d @ %d\n",
view.n_cells, view.n_max_seq, view.used_cells, view.token_count, view.max_contiguous, view.max_contiguous_idx);
std::unordered_map<llama_seq_id, size_t> seqs;
llama_kv_cache_view_cell * c_curr = view.cells;
llama_seq_id * cs_curr = view.cells_sequences;
for (int i = 0; i < view.n_cells; i++, c_curr++, cs_curr += view.n_max_seq) {
for (int j = 0; j < view.n_max_seq; j++) {
if (cs_curr[j] < 0) { continue; }
if (seqs.find(cs_curr[j]) == seqs.end()) {
if (seqs.size() + 1 >= sizeof(slot_chars)) { break; }
seqs[cs_curr[j]] = seqs.size();
}
}
if (seqs.size() + 1 >= sizeof(slot_chars)) { break; }
}
printf("=== Sequence legend: ");
for (const auto & it : seqs) {
printf("%zu=%d, ", it.second, it.first);
}
printf("'+'=other sequence ids");
c_curr = view.cells;
cs_curr = view.cells_sequences;
for (int i = 0; i < view.n_cells; i++, c_curr++, cs_curr += view.n_max_seq) {
if (i % row_size == 0) {
printf("\n%5d: ", i);
}
for (int j = 0; j < view.n_max_seq; j++) {
if (cs_curr[j] >= 0) {
const auto & it = seqs.find(cs_curr[j]);
putchar(it != seqs.end() ? int(slot_chars[it->second]) : '+');
} else {
putchar('.');
}
}
putchar(' ');
}
printf("\n=== Done dumping\n");
}