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

llama : fix MiniCPM (#5392)

Browse source 
* fix bug for norm_rms_eps missing

* to align with the same order as convert.py for model write

* fix: undo HF models permute tensor

* update for flake8 lint
This commit is contained in:
runfuture 2024-02-08 18:36:19 +08:00 • committed by GitHub
parent a6e514a85f
commit 4aa43fab56
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 63 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
llama.cpp
View file

@ -2947,6 +2947,8 @@ static void llm_load_hparams(
} break;
case LLM_ARCH_MINICPM:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
switch (hparams.n_layer) {
case 40: model.type = e_model::MODEL_2B; break;
default: model.type = e_model::MODEL_UNKNOWN;