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

convert : fix convert for refact models

Browse source
This commit is contained in:
Georgi Gerganov 2024-05-11 10:31:52 +03:00
parent 0faf92e74c
commit 03e940cdec
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
2 changed files with 16 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

12
convert-hf-to-gguf.py
View file

@ -1013,6 +1013,18 @@ class StarCoderModel(Model):
class RefactModel(Model):
model_arch = gguf.MODEL_ARCH.REFACT
def set_vocab(self):
super().set_vocab()
# TODO: how to determine special FIM tokens automatically?
special_vocab = gguf.SpecialVocab(self.dir_model, load_merges=False,
special_token_types = ['prefix', 'suffix', 'middle', 'fsep', 'eot'])
special_vocab._set_special_token("prefix", 1)
special_vocab._set_special_token("suffix", 3)
special_vocab._set_special_token("middle", 2)
special_vocab._set_special_token("fsep", 4) # is this correct?
special_vocab.add_to_gguf(self.gguf_writer)
def set_gguf_parameters(self):
hidden_dim = self.hparams["n_embd"]
inner_dim = 4 * hidden_dim