ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

support chatglm4 in lookup (#11855)

Browse source
This commit is contained in:
Yina Chen 2024-08-21 10:53:17 +03:00 committed by GitHub
parent 0236de3ac2
commit cc27321441
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 13 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
python/llm/src/ipex_llm/transformers/speculative.py
View file

@ -493,6 +493,13 @@ def _crop_past_key_values(self, past_key_values, new_cache_size, _enable_ipex=Fa
for k, v in past_key_values
]
elif self.config.model_type == "chatglm":
if self.config.num_layers == 40 and hasattr(self.config, 'rope_ratio'):
past_key_values = [
(k[:, :, :-(new_cache_size), :],
v[:, :, :-(new_cache_size), :])
for k, v in past_key_values
]
else:
# for chatglm, cache shape is [sl, bs, nh, hn]
past_key_values = [
(k[:-(new_cache_size), :, :, :],