Testing with qwen and updating tools

This commit is contained in:
2025-06-01 11:50:12 -05:00
parent 4c4f9b7f2c
commit 7844f26d59
6 changed files with 418 additions and 311 deletions
+18 -14
View File
@@ -8,16 +8,16 @@ from openai import OpenAI
class ChatGPTTelegramInferenceBot(BaseTelegramInferenceBot):
def __init__(self):
super().__init__()
self.client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"))
self.model = "gpt-4o-mini"
self.max_tokens = 16384
self.client = OpenAI(api_key=os.environ.get("OPENAI_API_KEY"), base_url="http://localhost:1234/v1")
self.model = "qwen3-1.7b"
self.max_tokens = 32768
def get_chat_response(self, messages):
response = self.client.chat.completions.create(
model=self.model,
messages=[{"role": "system", "content": self.system_prompt}] + messages,
functions=self.functions,
function_call="auto",
tools=self.functions,
tool_choice = "auto",
max_tokens=self.max_tokens
)
return response
@@ -34,8 +34,8 @@ class ChatGPTTelegramInferenceBot(BaseTelegramInferenceBot):
tool_calls = []
for message_part in response.choices:
if message_part.finish_reason == "function_call":
tool_calls.append(message_part.message.function_call)
if message_part.finish_reason == "tool_calls":
tool_calls.extend(message_part.message.tool_calls)
messages.append(response.choices[0].message)
@@ -44,9 +44,13 @@ class ChatGPTTelegramInferenceBot(BaseTelegramInferenceBot):
tool_use_results = []
while len(tool_calls) > 0:
tool_call = tool_calls.pop(0)
tool_call = tool_calls.pop(0).function
tool_response = self.call_tool(tool_call.name, tool_call.arguments)
tool_use_results.append({"role": "function", "name": tool_call.name, "content": json.dumps(tool_response)})
try:
tool_use_results.append({"role": "tool", "name": tool_call.name, "content": tool_response})
except (TypeError, ValueError) as e:
logging.error(f"Failed to serialize tool response: {e}")
tool_use_results.append({"role": "function", "name": tool_call.name, "content": "Serialization error"})
messages.extend(tool_use_results)
@@ -84,12 +88,12 @@ class ChatGPTTelegramInferenceBot(BaseTelegramInferenceBot):
return "No active processing to abort."
async def switch_model(self):
if self.model == "gpt-4o-mini":
self.model = "gpt-4o"
self.max_tokens = 4096
if self.model == "qwen3-4b":
self.model = "qwen3-30b-a3b"
# self.max_tokens = 4096
else:
self.model = "gpt-4o-mini"
self.max_tokens = 16384
self.model = "qwen3-4b"
# self.max_tokens = 16384
logging.info(f"Switched to model: {self.model}")
return f"Switched to model: {self.model}"