alihan/llama-cpp-python
1
0
Fork 0
mirror of https://github.com/abetlen/llama-cpp-python.git synced 2023-09-07 17:34:22 +03:00
Code Issues Packages Projects Releases Wiki Activity

Add field to disable reseting between generations

Browse Source
This commit is contained in:
Andrei Betlen
2023-04-13 00:28:00 -04:00
parent 22fa5a621f
commit 6595ad84bf
1 changed files with 4 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
llama_cpp/llama.py
View File

@@ -218,6 +218,7 @@ class Llama:
top_p: float, top_p: float,
temp: float, temp: float,
repeat_penalty: float, repeat_penalty: float,
reset: bool = True,
) -> Generator[ ) -> Generator[
llama_cpp.llama_token, Optional[Sequence[llama_cpp.llama_token]], None llama_cpp.llama_token, Optional[Sequence[llama_cpp.llama_token]], None
]: ]:
@@ -235,12 +236,14 @@ class Llama:
top_p: The top-p sampling parameter. top_p: The top-p sampling parameter.
temp: The temperature parameter. temp: The temperature parameter.
repeat_penalty: The repeat penalty parameter. repeat_penalty: The repeat penalty parameter.
reset: Whether to reset the model state.
Yields: Yields:
The generated tokens. The generated tokens.
""" """
assert self.ctx is not None assert self.ctx is not None
self.reset() if reset:
self.reset()
while True: while True:
self.eval(tokens) self.eval(tokens)
token = self.sample( token = self.sample(