mirror of
https://github.com/LostRuins/koboldcpp.git
synced 2025-09-11 09:34:37 +00:00
Merge commit '017cc5f446' into concedo_experimental
# Conflicts: # .github/ISSUE_TEMPLATE/010-bug-compilation.yml # .github/ISSUE_TEMPLATE/019-bug-misc.yml # CODEOWNERS # examples/batched-bench/batched-bench.cpp # examples/batched/batched.cpp # examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp # examples/gritlm/gritlm.cpp # examples/llama-bench/llama-bench.cpp # examples/passkey/passkey.cpp # examples/quantize-stats/quantize-stats.cpp # examples/run/run.cpp # examples/simple-chat/simple-chat.cpp # examples/simple/simple.cpp # examples/tokenize/tokenize.cpp # ggml/CMakeLists.txt # ggml/src/ggml-metal/CMakeLists.txt # ggml/src/ggml-vulkan/CMakeLists.txt # scripts/sync-ggml.last # src/llama.cpp # tests/test-autorelease.cpp # tests/test-model-load-cancel.cpp # tests/test-tokenizer-0.cpp # tests/test-tokenizer-1-bpe.cpp # tests/test-tokenizer-1-spm.cpp
This commit is contained in:
Concedo
commit
dcfa1eca4e
45 changed files with 806 additions and 229 deletions
|
|
@ -70,6 +70,7 @@ const char * llm_type_name(llm_type type) {
|
|||
case MODEL_70B: return "70B";
|
||||
case MODEL_236B: return "236B";
|
||||
case MODEL_314B: return "314B";
|
||||
case MODEL_671B: return "671B";
|
||||
case MODEL_SMALL: return "0.1B";
|
||||
case MODEL_MEDIUM: return "0.4B";
|
||||
case MODEL_LARGE: return "0.8B";
|
||||
|
|
@ -129,6 +130,14 @@ static std::string llama_model_ftype_name(llama_ftype ftype) {
|
|||
}
|
||||
}
|
||||
|
||||
static const char * llama_expert_gating_func_name(llama_expert_gating_func_type type) {
|
||||
switch (type) {
|
||||
case LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX: return "softmax";
|
||||
case LLAMA_EXPERT_GATING_FUNC_TYPE_SIGMOID: return "sigmoid";
|
||||
default: return "unknown";
|
||||
}
|
||||
}
|
||||
|
||||
std::string llama_model_arch_name (const llama_model & model) {
|
||||
return llm_arch_name(model.arch);
|
||||
}
|
||||
|
|
@ -793,6 +802,16 @@ void llm_load_hparams(llama_model_loader & ml, llama_model & model) {
|
|||
default: model.type = e_model::MODEL_UNKNOWN;
|
||||
}
|
||||
} break;
|
||||
case LLM_ARCH_COHERE2:
|
||||
{
|
||||
ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa);
|
||||
ml.get_key(LLM_KV_LOGIT_SCALE, hparams.f_logit_scale);
|
||||
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
|
||||
switch (hparams.n_layer) {
|
||||
case 32: model.type = e_model::MODEL_8B; break;
|
||||
default: model.type = e_model::MODEL_UNKNOWN;
|
||||
}
|
||||
} break;
|
||||
case LLM_ARCH_DBRX:
|
||||
{
|
||||
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
|
||||
|
|
@ -930,11 +949,19 @@ void llm_load_hparams(llama_model_loader & ml, llama_model & model) {
|
|||
ml.get_key(LLM_KV_EXPERT_FEED_FORWARD_LENGTH, hparams.n_ff_exp);
|
||||
ml.get_key(LLM_KV_EXPERT_SHARED_COUNT, hparams.n_expert_shared);
|
||||
ml.get_key(LLM_KV_EXPERT_WEIGHTS_SCALE, hparams.expert_weights_scale);
|
||||
ml.get_key(LLM_KV_EXPERT_WEIGHTS_NORM, hparams.expert_weights_norm, false);
|
||||
ml.get_key(LLM_KV_EXPERT_GATING_FUNC, hparams.expert_gating_func, false);
|
||||
if (hparams.expert_gating_func == LLAMA_EXPERT_GATING_FUNC_TYPE_NONE) {
|
||||
// for compatibility with existing DeepSeek V2 and V2.5 GGUFs
|
||||
// that have no expert_gating_func model parameter set
|
||||
hparams.expert_gating_func = LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX;
|
||||
}
|
||||
ml.get_key(LLM_KV_ROPE_SCALING_YARN_LOG_MUL, hparams.rope_yarn_log_mul);
|
||||
|
||||
switch (hparams.n_layer) {
|
||||
case 27: model.type = e_model::MODEL_16B; break;
|
||||
case 60: model.type = e_model::MODEL_236B; break;
|
||||
case 61: model.type = e_model::MODEL_671B; break;
|
||||
default: model.type = e_model::MODEL_UNKNOWN;
|
||||
}
|
||||
} break;
|
||||
|
|
@ -1266,6 +1293,10 @@ void llm_load_vocab(llama_model_loader & ml, llama_model & model) {
|
|||
tokenizer_pre == "deepseek-coder") {
|
||||
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_CODER;
|
||||
vocab.tokenizer_clean_spaces = false;
|
||||
} else if (
|
||||
tokenizer_pre == "deepseek-v3") {
|
||||
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEEPSEEK3_LLM;
|
||||
vocab.tokenizer_clean_spaces = false;
|
||||
} else if (
|
||||
tokenizer_pre == "falcon") {
|
||||
vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_FALCON;
|
||||
|
|
@ -1912,24 +1943,24 @@ void llm_load_print_meta(llama_model_loader & ml, llama_model & model) {
|
|||
LLAMA_LOG_INFO("%s: general.name = %s\n", __func__, model.name.c_str());
|
||||
|
||||
// special tokens
|
||||
if (vocab.special_bos_id != -1) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].text.c_str() ); }
|
||||
if (vocab.special_eos_id != -1) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_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() ); }
|
||||
if (vocab.special_eom_id != -1) { LLAMA_LOG_INFO( "%s: EOM token = %d '%s'\n", __func__, vocab.special_eom_id, vocab.id_to_token[vocab.special_eom_id].text.c_str() ); }
|
||||
if (vocab.special_unk_id != -1) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].text.c_str() ); }
|
||||
if (vocab.special_sep_id != -1) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].text.c_str() ); }
|
||||
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.special_bos_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: BOS token = %d '%s'\n", __func__, vocab.special_bos_id, vocab.id_to_token[vocab.special_bos_id].text.c_str() ); }
|
||||
if (vocab.special_eos_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: EOS token = %d '%s'\n", __func__, vocab.special_eos_id, vocab.id_to_token[vocab.special_eos_id].text.c_str() ); }
|
||||
if (vocab.special_eot_id != LLAMA_TOKEN_NULL) { 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() ); }
|
||||
if (vocab.special_eom_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: EOM token = %d '%s'\n", __func__, vocab.special_eom_id, vocab.id_to_token[vocab.special_eom_id].text.c_str() ); }
|
||||
if (vocab.special_unk_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: UNK token = %d '%s'\n", __func__, vocab.special_unk_id, vocab.id_to_token[vocab.special_unk_id].text.c_str() ); }
|
||||
if (vocab.special_sep_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: SEP token = %d '%s'\n", __func__, vocab.special_sep_id, vocab.id_to_token[vocab.special_sep_id].text.c_str() ); }
|
||||
if (vocab.special_pad_id != LLAMA_TOKEN_NULL) { 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 != LLAMA_TOKEN_NULL) { 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 != LLAMA_TOKEN_NULL) { 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.linefeed_id != LLAMA_TOKEN_NULL) { 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_fim_pre_id != -1) { LLAMA_LOG_INFO( "%s: FIM PRE token = %d '%s'\n", __func__, vocab.special_fim_pre_id, vocab.id_to_token[vocab.special_fim_pre_id].text.c_str() ); }
|
||||
if (vocab.special_fim_suf_id != -1) { LLAMA_LOG_INFO( "%s: FIM SUF token = %d '%s'\n", __func__, vocab.special_fim_suf_id, vocab.id_to_token[vocab.special_fim_suf_id].text.c_str() ); }
|
||||
if (vocab.special_fim_mid_id != -1) { LLAMA_LOG_INFO( "%s: FIM MID token = %d '%s'\n", __func__, vocab.special_fim_mid_id, vocab.id_to_token[vocab.special_fim_mid_id].text.c_str() ); }
|
||||
if (vocab.special_fim_pad_id != -1) { LLAMA_LOG_INFO( "%s: FIM PAD token = %d '%s'\n", __func__, vocab.special_fim_pad_id, vocab.id_to_token[vocab.special_fim_pad_id].text.c_str() ); }
|
||||
if (vocab.special_fim_rep_id != -1) { LLAMA_LOG_INFO( "%s: FIM REP token = %d '%s'\n", __func__, vocab.special_fim_rep_id, vocab.id_to_token[vocab.special_fim_rep_id].text.c_str() ); }
|
||||
if (vocab.special_fim_sep_id != -1) { LLAMA_LOG_INFO( "%s: FIM SEP token = %d '%s'\n", __func__, vocab.special_fim_sep_id, vocab.id_to_token[vocab.special_fim_sep_id].text.c_str() ); }
|
||||
if (vocab.special_fim_pre_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: FIM PRE token = %d '%s'\n", __func__, vocab.special_fim_pre_id, vocab.id_to_token[vocab.special_fim_pre_id].text.c_str() ); }
|
||||
if (vocab.special_fim_suf_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: FIM SUF token = %d '%s'\n", __func__, vocab.special_fim_suf_id, vocab.id_to_token[vocab.special_fim_suf_id].text.c_str() ); }
|
||||
if (vocab.special_fim_mid_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: FIM MID token = %d '%s'\n", __func__, vocab.special_fim_mid_id, vocab.id_to_token[vocab.special_fim_mid_id].text.c_str() ); }
|
||||
if (vocab.special_fim_pad_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: FIM PAD token = %d '%s'\n", __func__, vocab.special_fim_pad_id, vocab.id_to_token[vocab.special_fim_pad_id].text.c_str() ); }
|
||||
if (vocab.special_fim_rep_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: FIM REP token = %d '%s'\n", __func__, vocab.special_fim_rep_id, vocab.id_to_token[vocab.special_fim_rep_id].text.c_str() ); }
|
||||
if (vocab.special_fim_sep_id != LLAMA_TOKEN_NULL) { LLAMA_LOG_INFO( "%s: FIM SEP token = %d '%s'\n", __func__, vocab.special_fim_sep_id, vocab.id_to_token[vocab.special_fim_sep_id].text.c_str() ); }
|
||||
|
||||
for (const auto & id : vocab.special_eog_ids) {
|
||||
LLAMA_LOG_INFO( "%s: EOG token = %d '%s'\n", __func__, id, vocab.id_to_token[id].text.c_str() );
|
||||
|
|
@ -1951,6 +1982,8 @@ void llm_load_print_meta(llama_model_loader & ml, llama_model & model) {
|
|||
LLAMA_LOG_INFO("%s: n_ff_exp = %d\n", __func__, hparams.n_ff_exp);
|
||||
LLAMA_LOG_INFO("%s: n_expert_shared = %d\n", __func__, hparams.n_expert_shared);
|
||||
LLAMA_LOG_INFO("%s: expert_weights_scale = %.1f\n", __func__, hparams.expert_weights_scale);
|
||||
LLAMA_LOG_INFO("%s: expert_weights_norm = %d\n", __func__, hparams.expert_weights_norm);
|
||||
LLAMA_LOG_INFO("%s: expert_gating_func = %s\n", __func__, llama_expert_gating_func_name((enum llama_expert_gating_func_type) hparams.expert_gating_func));
|
||||
LLAMA_LOG_INFO("%s: rope_yarn_log_mul = %.4f\n", __func__, hparams.rope_yarn_log_mul);
|
||||
}
|
||||
|
||||
|
|
@ -1996,6 +2029,10 @@ struct llama_model_params llama_model_default_params() {
|
|||
}
|
||||
|
||||
void llama_free_model(struct llama_model * model) {
|
||||
llama_model_free(model);
|
||||
}
|
||||
|
||||
void llama_model_free(struct llama_model * model) {
|
||||
delete model;
|
||||
}
|
||||
|
||||
|
|
@ -2051,6 +2088,7 @@ enum llama_rope_type llama_rope_type(const struct llama_model * model) {
|
|||
case LLM_ARCH_MINICPM:
|
||||
case LLM_ARCH_XVERSE:
|
||||
case LLM_ARCH_COMMAND_R:
|
||||
case LLM_ARCH_COHERE2:
|
||||
case LLM_ARCH_OLMO:
|
||||
case LLM_ARCH_ARCTIC:
|
||||
case LLM_ARCH_DEEPSEEK:
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue