From ce08744b9e04875d42ffe49efd9205b51d435d88 Mon Sep 17 00:00:00 2001 From: yym68686 Date: Thu, 24 Oct 2024 00:08:10 +0800 Subject: [PATCH] =?UTF-8?q?=F0=9F=90=9B=20Bug:=20Fixed=20the=20bug=20where?= =?UTF-8?q?=20Claude=20did=20not=20correctly=20pass=20the=20model=20name?= =?UTF-8?q?=20when=20using=20tools.?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- setup.py | 2 +- src/ModelMerge/models/chatgpt.py | 2 +- src/ModelMerge/models/claude.py | 6 +++--- 3 files changed, 5 insertions(+), 5 deletions(-) diff --git a/setup.py b/setup.py index 176108c..f9b6384 100644 --- a/setup.py +++ b/setup.py @@ -4,7 +4,7 @@ setup( name="modelmerge", - version="0.11.57", + version="0.11.58", description="modelmerge is a multi-large language model API aggregator.", long_description=Path.open(Path("README.md"), encoding="utf-8").read(), long_description_content_type="text/markdown", diff --git a/src/ModelMerge/models/chatgpt.py b/src/ModelMerge/models/chatgpt.py index 1244dff..9dff77a 100644 --- a/src/ModelMerge/models/chatgpt.py +++ b/src/ModelMerge/models/chatgpt.py @@ -44,7 +44,7 @@ class chatgpt(BaseLLM): def __init__( self, api_key: str = None, - engine: str = os.environ.get("GPT_ENGINE") or "gpt-3.5-turbo", + engine: str = os.environ.get("GPT_ENGINE") or "gpt-4o", api_url: str = (os.environ.get("API_URL") or "https://api.openai.com/v1/chat/completions"), system_prompt: str = "You are ChatGPT, a large language model trained by OpenAI. Respond conversationally", proxy: str = None, diff --git a/src/ModelMerge/models/claude.py b/src/ModelMerge/models/claude.py index ffddcd5..6ea5643 100644 --- a/src/ModelMerge/models/claude.py +++ b/src/ModelMerge/models/claude.py @@ -171,7 +171,7 @@ class claude3(BaseLLM): def __init__( self, api_key: str = None, - engine: str = os.environ.get("GPT_ENGINE") or "claude-3-opus-20240229", + engine: str = os.environ.get("GPT_ENGINE") or "claude-3-5-sonnet-20241022", api_url: str = (os.environ.get("CLAUDE_API_URL") or "https://api.anthropic.com/v1/messages"), system_prompt: str = "You are ChatGPT, a large language model trained by OpenAI. Respond conversationally", temperature: float = 0.5, @@ -449,7 +449,7 @@ def ask_stream( response_role = "assistant" if self.conversation[convo_id][-1]["role"] == "function" and self.conversation[convo_id][-1]["name"] == "get_search_results": mess = self.conversation[convo_id].pop(-1) - yield from self.ask_stream(function_response, response_role, convo_id=convo_id, function_name=function_call_name, total_tokens=total_tokens, tools_id=tools_id, function_full_response=function_full_response, api_key=kwargs.get('api_key', self.api_key), plugins=kwargs.get("plugins", PLUGINS), system_prompt=system_prompt) + yield from self.ask_stream(function_response, response_role, convo_id=convo_id, function_name=function_call_name, total_tokens=total_tokens, model=model or self.engine, tools_id=tools_id, function_full_response=function_full_response, api_key=kwargs.get('api_key', self.api_key), plugins=kwargs.get("plugins", PLUGINS), system_prompt=system_prompt) else: if self.conversation[convo_id][-1]["role"] == "function" and self.conversation[convo_id][-1]["name"] == "get_search_results": mess = self.conversation[convo_id].pop(-1) @@ -609,7 +609,7 @@ async def ask_stream_async( response_role = "assistant" if self.conversation[convo_id][-1]["role"] == "function" and self.conversation[convo_id][-1]["name"] == "get_search_results": mess = self.conversation[convo_id].pop(-1) - async for chunk in self.ask_stream_async(function_response, response_role, convo_id=convo_id, function_name=function_call_name, total_tokens=total_tokens, tools_id=tools_id, function_full_response=function_full_response, api_key=kwargs.get('api_key', self.api_key), plugins=kwargs.get("plugins", PLUGINS), system_prompt=system_prompt): + async for chunk in self.ask_stream_async(function_response, response_role, convo_id=convo_id, function_name=function_call_name, total_tokens=total_tokens, model=model or self.engine, tools_id=tools_id, function_full_response=function_full_response, api_key=kwargs.get('api_key', self.api_key), plugins=kwargs.get("plugins", PLUGINS), system_prompt=system_prompt): yield chunk # yield from self.ask_stream(function_response, response_role, convo_id=convo_id, function_name=function_call_name, total_tokens=total_tokens, tools_id=tools_id, function_full_response=function_full_response) else: