feat(model): Passing stop parameter to proxyllm (#2077)

dbgpt/model/proxy/llms/chatgpt.py

@@ -39,6 +39,7 @@ async def chatgpt_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     async for r in client.generate_stream(request):
         yield r
@@ -188,6 +189,8 @@ class OpenAILLMClient(ProxyLLMClient):
             payload["temperature"] = request.temperature
         if request.max_new_tokens:
             payload["max_tokens"] = request.max_new_tokens
+        if request.stop:
+            payload["stop"] = request.stop
         return payload
 
     async def generate(

dbgpt/model/proxy/llms/deepseek.py

@@ -27,6 +27,7 @@ async def deepseek_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     async for r in client.generate_stream(request):
         yield r

dbgpt/model/proxy/llms/gemini.py

@@ -46,6 +46,7 @@ def gemini_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     for r in client.sync_generate_stream(request):
         yield r

dbgpt/model/proxy/llms/moonshot.py

@@ -26,6 +26,7 @@ async def moonshot_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     async for r in client.generate_stream(request):
         yield r

dbgpt/model/proxy/llms/spark.py

@@ -47,6 +47,7 @@ def spark_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     for r in client.sync_generate_stream(request):
         yield r

dbgpt/model/proxy/llms/tongyi.py

@@ -21,6 +21,7 @@ def tongyi_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     for r in client.sync_generate_stream(request):
         yield r
@@ -96,6 +97,7 @@ class TongyiLLMClient(ProxyLLMClient):
                 top_p=0.8,
                 stream=True,
                 result_format="message",
+                stop=request.stop,
             )
             for r in res:
                 if r:

dbgpt/model/proxy/llms/yi.py

@@ -26,6 +26,7 @@ async def yi_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     async for r in client.generate_stream(request):
         yield r

dbgpt/model/proxy/llms/zhipu.py

@@ -28,6 +28,7 @@ def zhipu_generate_stream(
         temperature=params.get("temperature"),
         context=context,
         max_new_tokens=params.get("max_new_tokens"),
+        stop=params.get("stop"),
     )
     for r in client.sync_generate_stream(request):
         yield r