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

llama : more consistent names of count variables

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2024-03-11 13:25:38 +02:00
parent caa106d4e0
commit 32daccd755
No known key found for this signature in database
GPG key ID: BF970631944C16B7
6 changed files with 28 additions and 27 deletions
Download patch file Download diff file Expand all files Collapse all files

2
README.md
View file

@ -10,7 +10,7 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others)
### Recent API changes
- [2024 Mar 8] `llama_kv_cache_seq_rm()` returns a `bool` instead of `void`, and new `llama_n_max_seq()` returns the upper limit of acceptable `seq_id` in batches (relevant when dealing with multiple sequences) https://github.com/ggerganov/llama.cpp/pull/5328
- [2024 Mar 8] `llama_kv_cache_seq_rm()` returns a `bool` instead of `void`, and new `llama_n_seq_max()` returns the upper limit of acceptable `seq_id` in batches (relevant when dealing with multiple sequences) https://github.com/ggerganov/llama.cpp/pull/5328
- [2024 Mar 4] Embeddings API updated https://github.com/ggerganov/llama.cpp/pull/5796
- [2024 Mar 3] `struct llama_context_params` https://github.com/ggerganov/llama.cpp/pull/5849

16
common/common.cpp
View file

@ -1786,17 +1786,17 @@ 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);
view.n_cells, view.n_seq_max, 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) {
for (int i = 0; i < view.n_cells; i++, c_curr++, cs_curr += view.n_seq_max) {
if (i % row_size == 0) {
printf("\n%5d: ", i);
}
int seq_count = 0;
for (int j = 0; j < view.n_max_seq; j++) {
for (int j = 0; j < view.n_seq_max; j++) {
if (cs_curr[j] >= 0) { seq_count++; }
}
putchar(slot_chars[std::min(sizeof(slot_chars) - 2, size_t(seq_count))]);
@ -1809,14 +1809,14 @@ 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);
view.n_cells, view.n_seq_max, 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++) {
for (int i = 0; i < view.n_cells; i++, c_curr++, cs_curr += view.n_seq_max) {
for (int j = 0; j < view.n_seq_max; j++) {
if (cs_curr[j] < 0) { continue; }
if (seqs.find(cs_curr[j]) == seqs.end()) {
if (seqs.size() + 1 >= sizeof(slot_chars)) { break; }
@ -1835,11 +1835,11 @@ void dump_kv_cache_view_seqs(const llama_kv_cache_view & view, int row_size) {
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) {
for (int i = 0; i < view.n_cells; i++, c_curr++, cs_curr += view.n_seq_max) {
if (i % row_size == 0) {
printf("\n%5d: ", i);
}
for (int j = 0; j < view.n_max_seq; j++) {
for (int j = 0; j < view.n_seq_max; j++) {
if (cs_curr[j] >= 0) {
const auto & it = seqs.find(cs_curr[j]);
putchar(it != seqs.end() ? int(slot_chars[it->second]) : '+');

1
examples/main/main.cpp
View file

@ -878,6 +878,7 @@ int main(int argc, char ** argv) {
const auto line_pfx = ::llama_tokenize(ctx, params.input_prefix, false, true);
const auto line_inp = ::llama_tokenize(ctx, buffer, false, false);
const auto line_sfx = ::llama_tokenize(ctx, params.input_suffix, false, true);
LOG("input tokens: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, line_inp).c_str());
embd_inp.insert(embd_inp.end(), line_pfx.begin(), line_pfx.end());

6
examples/perplexity/perplexity.cpp
View file

@ -841,7 +841,7 @@ static void hellaswag_score(llama_context * ctx, const gpt_params & params) {
const int n_batch = params.n_batch;
const int max_tasks_per_batch = 32;
const int max_seq = std::min(4*max_tasks_per_batch, (int) llama_n_max_seq(ctx));
const int max_seq = std::min(4*max_tasks_per_batch, (int) llama_n_seq_max(ctx));
llama_batch batch = llama_batch_init(n_ctx, 0, max_seq);
@ -1118,7 +1118,7 @@ static void winogrande_score(llama_context * ctx, const gpt_params & params) {
const int n_batch = params.n_batch;
const int max_tasks_per_batch = 128;
const int max_seq = std::min(2*max_tasks_per_batch, (int) llama_n_max_seq(ctx));
const int max_seq = std::min(2*max_tasks_per_batch, (int) llama_n_seq_max(ctx));
llama_batch batch = llama_batch_init(n_ctx, 0, max_seq);
@ -1470,7 +1470,7 @@ static void multiple_choice_score(llama_context * ctx, const gpt_params & params
const int n_batch = params.n_batch;
const int max_tasks_per_batch = 32;
const int max_seq = std::min(4*max_tasks_per_batch, (int) llama_n_max_seq(ctx));
const int max_seq = std::min(4*max_tasks_per_batch, (int) llama_n_seq_max(ctx));
llama_batch batch = llama_batch_init(n_ctx, 0, max_seq);

18
llama.cpp
View file

@ -13037,7 +13037,7 @@ uint32_t llama_n_batch(const struct llama_context * ctx) {
return ctx->cparams.n_batch;
}
uint32_t llama_n_max_seq(const struct llama_context * ctx) {
uint32_t llama_n_seq_max(const struct llama_context * ctx) {
return ctx->kv_self.size;
}
@ -13201,10 +13201,10 @@ int32_t llama_model_apply_lora_from_file(const struct llama_model * model, const
}
}
struct llama_kv_cache_view llama_kv_cache_view_init(const struct llama_context * ctx, int32_t n_max_seq) {
struct llama_kv_cache_view llama_kv_cache_view_init(const struct llama_context * ctx, int32_t n_seq_max) {
struct llama_kv_cache_view result = {
/*.n_cells = */ 0,
/*.n_max_seq = */ n_max_seq,
/*.n_seq_max = */ n_seq_max,
/*.token_count = */ 0,
/*.used_cells = */ llama_get_kv_cache_used_cells(ctx),
/*.max_contiguous = */ 0,
@ -13232,7 +13232,7 @@ void llama_kv_cache_view_update(const struct llama_context * ctx, struct llama_k
void * p = realloc(view->cells, sizeof(struct llama_kv_cache_view_cell) * view->n_cells);
GGML_ASSERT(p != nullptr && "Failed to alloc kv_cache_view cells");
view->cells = (struct llama_kv_cache_view_cell *)p;
p = realloc(view->cells_sequences, sizeof(llama_seq_id) * view->n_max_seq * view->n_cells);
p = realloc(view->cells_sequences, sizeof(llama_seq_id) * view->n_seq_max * view->n_cells);
GGML_ASSERT(p != nullptr && "Failed to alloc kv_cache_view cells sequences");
view->cells_sequences = (llama_seq_id *)p;
}
@ -13246,7 +13246,7 @@ void llama_kv_cache_view_update(const struct llama_context * ctx, struct llama_k
uint32_t max_contig = 0;
int32_t max_contig_idx = -1;
for (int32_t i = 0; i < int32_t(ctx->kv_self.size); i++, c_curr++, cs_curr += view->n_max_seq) {
for (int32_t i = 0; i < int32_t(ctx->kv_self.size); i++, c_curr++, cs_curr += view->n_seq_max) {
const size_t curr_size = kv_cells[i].seq_id.size();
token_count += curr_size;
c_curr->pos = kv_cells[i].pos + kv_cells[i].delta;
@ -13263,7 +13263,7 @@ void llama_kv_cache_view_update(const struct llama_context * ctx, struct llama_k
int seq_idx = 0;
for (const llama_seq_id it : kv_cells[i].seq_id) {
if (seq_idx >= view->n_max_seq) {
if (seq_idx >= view->n_seq_max) {
break;
}
cs_curr[seq_idx] = it;
@ -13272,7 +13272,7 @@ void llama_kv_cache_view_update(const struct llama_context * ctx, struct llama_k
if (seq_idx != 0) {
used_cells++;
}
for (; seq_idx < view->n_max_seq; seq_idx++) {
for (; seq_idx < view->n_seq_max; seq_idx++) {
cs_curr[seq_idx] = -1;
}
}
@ -13934,12 +13934,12 @@ int32_t llama_tokenize(
const char * text,
int32_t text_len,
llama_token * tokens,
int32_t n_max_tokens,
int32_t n_tokens_max,
bool add_bos,
bool special) {
auto res = llama_tokenize_internal(model->vocab, std::string(text, text_len), add_bos, special);
if (n_max_tokens < (int) res.size()) {
if (n_tokens_max < (int) res.size()) {
// LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
return -((int) res.size());
}

12
llama.h
View file

@ -377,7 +377,7 @@ extern "C" {
LLAMA_API uint32_t llama_n_ctx (const struct llama_context * ctx);
LLAMA_API uint32_t llama_n_batch (const struct llama_context * ctx);
LLAMA_API uint32_t llama_n_max_seq (const struct llama_context * ctx);
LLAMA_API uint32_t llama_n_seq_max (const struct llama_context * ctx);
LLAMA_API enum llama_vocab_type llama_vocab_type(const struct llama_model * model);
LLAMA_API enum llama_rope_type llama_rope_type (const struct llama_model * model);
@ -456,7 +456,7 @@ extern "C" {
// Maximum number of sequences that can exist in a cell. It's not an error
// if there are more sequences in a cell than this value, however they will
// not be visible in the view cells_sequences.
int32_t n_max_seq;
int32_t n_seq_max;
// Number of tokens in the cache. For example, if there are two populated
// cells, the first with 1 sequence id in it and the second with 2 sequence
@ -476,12 +476,12 @@ extern "C" {
// Information for an individual cell.
struct llama_kv_cache_view_cell * cells;
// The sequences for each cell. There will be n_max_seq items per cell.
// The sequences for each cell. There will be n_seq_max items per cell.
llama_seq_id * cells_sequences;
};
// Create an empty KV cache view. (use only for debugging purposes)
LLAMA_API struct llama_kv_cache_view llama_kv_cache_view_init(const struct llama_context * ctx, int32_t n_max_seq);
LLAMA_API struct llama_kv_cache_view llama_kv_cache_view_init(const struct llama_context * ctx, int32_t n_seq_max);
// Free a KV cache view. (use only for debugging purposes)
LLAMA_API void llama_kv_cache_view_free(struct llama_kv_cache_view * view);
@ -708,7 +708,7 @@ extern "C" {
/// @details Convert the provided text into tokens.
/// @param tokens The tokens pointer must be large enough to hold the resulting tokens.
/// @return Returns the number of tokens on success, no more than n_max_tokens
/// @return Returns the number of tokens on success, no more than n_tokens_max
/// @return Returns a negative number on failure - the number of tokens that would have been returned
/// @param special Allow tokenizing special and/or control tokens which otherwise are not exposed and treated as plaintext.
/// Does not insert a leading space.
@ -717,7 +717,7 @@ extern "C" {
const char * text,
int32_t text_len,
llama_token * tokens,
int32_t n_max_tokens,
int32_t n_tokens_max,
bool add_bos,
bool special);