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

save and load head_count_kv in lora checkpoints

Browse source
This commit is contained in:
xaedes 2023-09-16 20:20:23 +02:00
parent 7aa9ea7f20
commit 48d3509190
No known key found for this signature in database
GPG key ID: 30030EDD817EA2B1
1 changed files with 6 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

7
examples/finetune/finetune.cpp
View file

@ -174,6 +174,7 @@ static const char * LLM_KV_EMBEDDING_LENGTH = "%s.embedding_length";
static const char * LLM_KV_BLOCK_COUNT = "%s.block_count";
static const char * LLM_KV_FEED_FORWARD_LENGTH = "%s.feed_forward_length";
static const char * LLM_KV_ATTENTION_HEAD_COUNT = "%s.attention.head_count";
static const char * LLM_KV_ATTENTION_HEAD_COUNT_KV = "%s.attention.head_count_kv";
static const char * LLM_KV_ATTENTION_LAYERNORM_RMS_EPS = "%s.attention.layer_norm_rms_epsilon";
static const char * LLM_KV_ROPE_DIMENSION_COUNT = "%s.rope.dimension_count";
static const char * LLM_KV_ROPE_FREQ_BASE = "%s.rope.freq_base"; // TODO load in llama.cpp
@ -865,8 +866,11 @@ static void load_llama_lora_gguf(struct gguf_context * fctx, struct ggml_context
GGUF_GET_KEY(fctx, model->hparams.n_head, gguf_get_val_u32, GGUF_TYPE_UINT32, true, kv(LLM_KV_ATTENTION_HEAD_COUNT));
GGUF_GET_KEY(fctx, model->hparams.n_layer, gguf_get_val_u32, GGUF_TYPE_UINT32, true, kv(LLM_KV_BLOCK_COUNT));
model->hparams.n_head_kv = model->hparams.n_head;
GGUF_GET_KEY(fctx, model->hparams.n_head_kv, gguf_get_val_u32, GGUF_TYPE_UINT32, false, kv(LLM_KV_ATTENTION_HEAD_COUNT_KV));
model->hparams.n_rot = model->hparams.n_embd / model->hparams.n_head;
GGUF_GET_KEY(fctx, model->hparams.n_rot, gguf_get_val_u32, GGUF_TYPE_UINT32, false, kv(LLM_KV_ROPE_DIMENSION_COUNT));
GGUF_GET_KEY(fctx, model->hparams.n_rot, gguf_get_val_u32, GGUF_TYPE_UINT32, false, kv(LLM_KV_ROPE_DIMENSION_COUNT));
float rope_freq_scale = 1.0f;
GGUF_GET_KEY(fctx, lora->hparams.f_norm_rms_eps, gguf_get_val_f32, GGUF_TYPE_FLOAT32, false, kv(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS));
@ -939,6 +943,7 @@ static void save_llama_lora_gguf(struct gguf_context * fctx, struct my_llama_mod
gguf_set_val_u32(fctx, kv(LLM_KV_EMBEDDING_LENGTH), model->hparams.n_embd);
gguf_set_val_u32(fctx, kv(LLM_KV_FEED_FORWARD_LENGTH), model->hparams.n_ff);
gguf_set_val_u32(fctx, kv(LLM_KV_ATTENTION_HEAD_COUNT), model->hparams.n_head);
gguf_set_val_u32(fctx, kv(LLM_KV_ATTENTION_HEAD_COUNT_KV), model->hparams.n_head_kv);
gguf_set_val_u32(fctx, kv(LLM_KV_BLOCK_COUNT), model->hparams.n_layer);
gguf_set_val_u32(fctx, kv(LLM_KV_ROPE_DIMENSION_COUNT), model->hparams.n_rot);
gguf_set_val_f32(fctx, kv(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS), lora->hparams.f_norm_rms_eps);