ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-03-02 10:00:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

spec : add self speculative decoding, ngram and refactor (#1261)

Browse Source 
* spec : add self speculative decoding and ngram-mod and refactor

common : use common_ prefix for common library function

llama : use LLAMA_TOKEN_NULL

spec : add self speculative decoding (no draft model required) + refactor

spec : add ngram-mod

spec : various improvements ton ngram-map + docs

spec : fix the check-rate logic of ngram-simple

common : add common_speculative_is_compat()

spec : simplify time measurement using common_time_meas

refactor common_sampler_init

refactor common_token_to_piece

refactor and fix cur_p bug

clean up

* spec : remove check rate

* spec: show warnings instead of abort

---------

Co-authored-by: firecoperana <firecoperana>
Co-authored-by: Sascha Rogmann <59577610+srogmann@users.noreply.github.com>
This commit is contained in:
firecoperana
2026-02-13 12:04:55 -06:00
committed by GitHub
parent 1fdbc0dafe
commit 1cb7e1bf39
54 changed files with 2652 additions and 779 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
src/llama.cpp
View File

@@ -2175,7 +2175,7 @@ static int llama_model_load(const std::string & fname, llama_model & model, llam
model.ftype == LLAMA_FTYPE_MOSTLY_Q4_1
)
)) {
// TODO(cebtenzzre): propagate this error outside of llama_load_model_from_file
// TODO(cebtenzzre): propagate this error outside of llama_model_load_from_file
LLAMA_LOG_WARN("%s: disabling Kompute due to unsupported model arch or quantization\n", __func__);
params.n_gpu_layers = 0;
}
@@ -4214,7 +4214,7 @@ static std::string create_rpc_name(std::string endpoint, uint32_t device) {
return dev_name;
}
struct llama_model * llama_load_model_from_file(
struct llama_model * llama_model_load_from_file(
const char * path_model,
struct llama_model_params params) {
ggml_time_init();
@@ -4397,7 +4397,7 @@ static void llama_repack_up_gate_exps(llama_context & lctx) {
}
}
struct llama_context * llama_new_context_with_model(
struct llama_context * llama_init_from_model(
struct llama_model * model,
struct llama_context_params params) {
@@ -4973,8 +4973,8 @@ uint32_t llama_n_seq_max(const struct llama_context * ctx) {
return ctx->kv_self.size;
}
enum llama_vocab_type llama_vocab_type(const struct llama_model * model) {
return model->vocab.get_type();
enum llama_vocab_type llama_vocab_type(const struct llama_vocab * vocab) {
return vocab->get_type();
}
const struct llama_vocab* llama_get_model_vocab(const struct llama_model* model) {
@@ -6878,15 +6878,16 @@ int32_t llama_token_to_piece(
return model->vocab.token_to_piece(token, buf, length, lstrip, special);
}
int32_t llama_detokenize(
const struct llama_model * model,
const llama_token * tokens,
int32_t n_tokens,
char * text,
int32_t text_len_max,
bool remove_special,
bool unparse_special) {
return model->vocab.detokenize(tokens, n_tokens, text, text_len_max, remove_special, unparse_special);
const struct llama_vocab * vocab,
const llama_token * tokens,
int32_t n_tokens,
char * text,
int32_t text_len_max,
bool remove_special,
bool unparse_special) {
return vocab->detokenize(tokens, n_tokens, text, text_len_max, remove_special, unparse_special);
}
//
@@ -7655,9 +7656,13 @@ void llama_set_rng_seed(struct llama_context * ctx, uint32_t seed) {
}
void llama_sample_softmax(struct llama_context * ctx, llama_token_data_array * candidates) {
llama_sample_softmax_impl(ctx ? &ctx->sampling : nullptr, candidates);
llama_sample_softmax_impl(ctx ? &ctx->sampling : nullptr, candidates, /* normalize */ true);
}
void llama_sample_dist(struct llama_context * ctx, llama_token_data_array * candidates) {
llama_sample_softmax_impl(ctx ? &ctx->sampling : nullptr, candidates, /* normalize */ false);
}
void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * candidates, int32_t k, size_t min_keep) {
llama_sample_top_k_impl(ctx ? &ctx->sampling : nullptr, candidates, k, min_keep);
}