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

Revert "llama_cpp server: delete some ignored / unused parameters"

Browse source 
This reverts commit b47b9549d5.
This commit is contained in:
Andrei Betlen 2023-05-07 02:02:34 -04:00
parent c382d8f86a
commit 86753976c4
2 changed files with 16 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
llama_cpp/llama_types.py
View file

@ -60,6 +60,8 @@ class Completion(TypedDict):
class ChatCompletionMessage(TypedDict): class ChatCompletionMessage(TypedDict):
role: Literal["assistant", "user", "system"] role: Literal["assistant", "user", "system"]
content: str content: str
user: NotRequired[str]
class ChatCompletionChoice(TypedDict): class ChatCompletionChoice(TypedDict):
index: int index: int

17
llama_cpp/server/app.py
View file

@ -185,7 +185,13 @@ def create_completion(
completion_or_chunks = llama( completion_or_chunks = llama(
**request.dict( **request.dict(
exclude={ exclude={
"model" "model",
"n",
"frequency_penalty",
"presence_penalty",
"best_of",
"logit_bias",
"user",
} }
) )
) )
@ -221,7 +227,7 @@ CreateEmbeddingResponse = create_model_from_typeddict(llama_cpp.Embedding)
def create_embedding( def create_embedding(
request: CreateEmbeddingRequest, llama: llama_cpp.Llama = Depends(get_llama) request: CreateEmbeddingRequest, llama: llama_cpp.Llama = Depends(get_llama)
): ):
return llama.create_embedding(**request.dict(exclude={"model"})) return llama.create_embedding(**request.dict(exclude={"model", "user"}))
class ChatCompletionRequestMessage(BaseModel): class ChatCompletionRequestMessage(BaseModel):
@ -283,7 +289,12 @@ def create_chat_completion(
completion_or_chunks = llama.create_chat_completion( completion_or_chunks = llama.create_chat_completion(
**request.dict( **request.dict(
exclude={ exclude={
"model" "model",
"n",
"presence_penalty",
"frequency_penalty",
"logit_bias",
"user",
} }
), ),
) )