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

* gguf-py: remove redundant logs

Browse source 
* llama: remove the init_mapping_prefetch custom parameter
This commit is contained in:
root 2024-03-26 08:32:30 +00:00
parent c14d4e8723
commit 458c1d16b0
2 changed files with 1 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

1
convert-hf-to-gguf.py
View file

@ -212,7 +212,6 @@ class Model(ABC):
try: try:
return cls._model_classes[arch] return cls._model_classes[arch]
except KeyError: except KeyError:
print(f"{cls._model_classes}")
raise NotImplementedError(f'Architecture {arch!r} not supported!') from None raise NotImplementedError(f'Architecture {arch!r} not supported!') from None
def _is_model_safetensors(self) -> bool: def _is_model_safetensors(self) -> bool:

4
llama.cpp
View file

@ -4375,7 +4375,6 @@ static bool llm_load_tensors(
LLAMA_LOG_INFO("%s: ggml ctx size = %7.2f MiB\n", __func__, model.ctxs.size()*ctx_size/1024.0/1024.0); LLAMA_LOG_INFO("%s: ggml ctx size = %7.2f MiB\n", __func__, model.ctxs.size()*ctx_size/1024.0/1024.0);
bool init_mapping_prefetch = true;
// create tensors for the weights // create tensors for the weights
{ {
const int64_t n_embd = hparams.n_embd; const int64_t n_embd = hparams.n_embd;
@ -5230,7 +5229,6 @@ static bool llm_load_tensors(
} break; } break;
case LLM_ARCH_XVERSE: case LLM_ARCH_XVERSE:
{ {
init_mapping_prefetch = false;
model.tok_embd = ml.create_tensor(ctx_input, tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}); model.tok_embd = ml.create_tensor(ctx_input, tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab});
{ {
model.output_norm = ml.create_tensor(ctx_output, tn(LLM_TENSOR_OUTPUT_NORM, "weight"), {n_embd}); model.output_norm = ml.create_tensor(ctx_output, tn(LLM_TENSOR_OUTPUT_NORM, "weight"), {n_embd});
@ -5289,7 +5287,7 @@ static bool llm_load_tensors(
ml.done_getting_tensors(); ml.done_getting_tensors();
ml.init_mappings(init_mapping_prefetch, use_mlock ? &model.mlock_mmaps : nullptr); ml.init_mappings(true, use_mlock ? &model.mlock_mmaps : nullptr);
model.mappings.reserve(ml.mappings.size()); model.mappings.reserve(ml.mappings.size());
// create the backend buffers // create the backend buffers