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

Format

Browse source
This commit is contained in:
Andrei Betlen 2023-07-28 01:53:08 -04:00
parent abc538fcd5
commit a9b9f0397c
1 changed files with 5 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
llama_cpp/llama.py
View file

@ -277,7 +277,9 @@ class Llama:
if self.tensor_split is not None:
FloatArray = (ctypes.c_float * len(self.tensor_split))(*self.tensor_split)
self._p_tensor_split = ctypes.POINTER(ctypes.c_float)(FloatArray) # keep a reference to the array so it is not gc'd
self._p_tensor_split = ctypes.POINTER(ctypes.c_float)(
FloatArray
) # keep a reference to the array so it is not gc'd
self.params.tensor_split = self._p_tensor_split
self.params.rope_freq_base = rope_freq_base
@ -959,9 +961,7 @@ class Llama:
for token in remaining_tokens:
token_end_position += len(self.detokenize([token]))
# Check if stop sequence is in the token
if token_end_position >= (
remaining_length - first_stop_position
):
if token_end_position >= (remaining_length - first_stop_position):
break
logprobs_or_none: Optional[CompletionLogprobs] = None
if logprobs is not None: