phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

fix(ollama_chat.py): fix default token counting for ollama chat

Browse source
This commit is contained in:
Krrish Dholakia 2024-01-24 20:09:08 -08:00
parent 3e59a02dfb
commit 43f139fafd
2 changed files with 15 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

12
litellm/llms/ollama_chat.py
View file

@ -220,8 +220,10 @@ def get_ollama_response(
model_response["choices"][0]["message"] = response_json["message"]
model_response["created"] = int(time.time())
model_response["model"] = "ollama/" + model
prompt_tokens = response_json.get("prompt_eval_count", len(encoding.encode(prompt))) # type: ignore
completion_tokens = response_json["eval_count"]
prompt_tokens = response_json.get("prompt_eval_count", litellm.token_counter(messages=messages)) # type: ignore
completion_tokens = response_json.get(
"eval_count", litellm.token_counter(text=response_json["message"])
)
model_response["usage"] = litellm.Usage(
prompt_tokens=prompt_tokens,
completion_tokens=completion_tokens,
@ -320,8 +322,10 @@ async def ollama_acompletion(url, data, model_response, encoding, logging_obj):
model_response["choices"][0]["message"] = response_json["message"]
model_response["created"] = int(time.time())
model_response["model"] = "ollama/" + data["model"]
prompt_tokens = response_json.get("prompt_eval_count", len(encoding.encode(prompt))) # type: ignore
completion_tokens = response_json["eval_count"]
prompt_tokens = response_json.get("prompt_eval_count", litellm.token_counter(messages=data["messages"])) # type: ignore
completion_tokens = response_json.get(
"eval_count", litellm.token_counter(text=response_json["message"])
)
model_response["usage"] = litellm.Usage(
prompt_tokens=prompt_tokens,
completion_tokens=completion_tokens,

9
litellm/utils.py
View file

@ -2872,8 +2872,13 @@ def token_counter(
print_verbose(
f"Token Counter - using generic token counter, for model={model}"
)
enc = tokenizer_json["tokenizer"].encode(text)
num_tokens = len(enc)
num_tokens = openai_token_counter(
text=text, # type: ignore
model="gpt-3.5-turbo",
messages=messages,
is_tool_call=is_tool_call,
count_response_tokens=count_response_tokens,
)
else:
num_tokens = len(encoding.encode(text)) # type: ignore
return num_tokens