Add Support for QWEN models from API (QWEN-max, etc.); Improve on the robustness of getting observation

mm_agents/agent.py

@@ -943,8 +943,7 @@ class PromptAgent:
             messages = payload["messages"]
             max_tokens = payload["max_tokens"]
             top_p = payload["top_p"]
-            if payload["temperature"]:
-                logger.warning("Qwen model does not support temperature parameter, it will be ignored.")
+            temperature = payload["temperature"]
 
             qwen_messages = []
 
@@ -961,23 +960,42 @@ class PromptAgent:
 
                 qwen_messages.append(qwen_message)
 
-            response = dashscope.MultiModalConversation.call(
-                model='qwen-vl-plus',
-                messages=messages,
-                max_length=max_tokens,
-                top_p=top_p,
-            )
-            # The response status_code is HTTPStatus.OK indicate success,
-            # otherwise indicate request is failed, you can get error code
-            # and message from code and message.
-            if response.status_code == HTTPStatus.OK:
+            flag = 0
+            while True:
                 try:
-                    return response.json()['output']['choices'][0]['message']['content']
-                except Exception:
-                    return ""
-            else:
-                print(response.code)  # The error code.
-                print(response.message)  # The error message.
+                    if flag > 20:
+                        break
+                    logger.info("Generating content with model: %s", self.model)
+                    response = dashscope.Generation.call(
+                        model=self.model,
+                        messages=qwen_messages,
+                        result_format="message",
+                        max_length=max_tokens,
+                        top_p=top_p,
+                        temperature=temperature
+                    )
+
+                    if response.status_code == HTTPStatus.OK:
+                        break
+                    else:
+                        logger.error('Request id: %s, Status code: %s, error code: %s, error message: %s' % (
+                            response.request_id, response.status_code,
+                            response.code, response.message
+                        ))
+                        raise Exception("Failed to call LLM: " + response.message)
+                except:
+                    if flag == 0:
+                        qwen_messages = [qwen_messages[0]] + qwen_messages[-1:]
+                    else:
+                        for i in range(len(qwen_messages[-1]["content"])):
+                            if "text" in qwen_messages[-1]["content"][i]:
+                                qwen_messages[-1]["content"][i]["text"] = ' '.join(qwen_messages[-1]["content"][i]["text"].split()[:-500])
+                    flag = flag + 1
+
+            try:
+                return response['output']['choices'][0]['message']['content']
+            except Exception as e:
+                print("Failed to call LLM: " + str(e))
                 return ""
 
         else: