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

convert : use sentencepiece tokenizer for Mixtral-instruct

Browse source
This commit is contained in:
Radek Pilar 2023-12-12 17:05:37 +01:00
parent cf75991cac
commit d6f74975a4
1 changed files with 7 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

7
convert-hf-to-gguf.py
View file

@ -180,6 +180,8 @@ class Model:
return StableLMModel return StableLMModel
if model_architecture == "QWenLMHeadModel": if model_architecture == "QWenLMHeadModel":
return QwenModel return QwenModel
if model_architecture == "MixtralForCausalLM":
return MixtralModel
return Model return Model
def _is_model_safetensors(self) -> bool: def _is_model_safetensors(self) -> bool:
@ -848,6 +850,11 @@ class StableLMModel(Model):
self.gguf_writer.add_parallel_residual(hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True) self.gguf_writer.add_parallel_residual(hparams["use_parallel_residual"] if "use_parallel_residual" in hparams else True)
self.gguf_writer.add_layer_norm_eps(1e-5) self.gguf_writer.add_layer_norm_eps(1e-5)
class MixtralModel(Model):
def set_vocab(self):
self._set_vocab_sentencepiece()
class QwenModel(Model): class QwenModel(Model):
@staticmethod @staticmethod