sync offline system prompts with training data: no_thinking -> no system message, deep_thinking -> <think>/<answer> tags

modeling_moss_vl.py

@@ -55,12 +55,12 @@ from transformers.generation.streamers import TextIteratorStreamer
 
 _OFFLINE_SYSTEM_PROMPTS = {
     "no_thinking": {
-        "text_image": "You are a helpful AI assistant. Respond to the user's request based on the provided text and/or images.",
-        "video": "You are a helpful AI assistant specializing in video analysis. Respond to the user's request based on the provided video content.",
+        "text_image": None,
+        "video": None,
     },
     "deep_thinking": {
-        "text_image": "A conversation between User and Assistant. The user makes a request, and the assistant responds to it based on the provided text and/or images. The assistant first thinks about the reasoning process in the mind and then provides the user with the answer. The reasoning process and answer are enclosed within <thinking></thinking> and <answer></answer> tags, respectively, i.e., <thinking>reasoning process here</thinking><answer>answer here</answer>.",
-        "video": "A conversation between User and Assistant specializing in video analysis. The user makes a request, and the assistant responds to it based on the provided video content. The assistant first thinks about the reasoning process in the mind and then provides the user with the answer. The reasoning process and answer are enclosed within <thinking></thinking> and <answer></answer> tags, respectively, i.e., <thinking>reasoning process here</thinking><answer>answer here</answer>.",
+        "text_image": "A conversation between User and Assistant. The user asks a question, and the Assistant solves it. The assistant first thinks about the reasoning process in the mind and then provides the user with the answer. The reasoning process and answer are enclosed within <think>...</think> and <answer>...</answer> tags, respectively, i.e., <think> reasoning process here </think> <answer> answer here </answer>.",
+        "video": "A conversation between User and Assistant. The user asks a question, and the Assistant solves it. The assistant first thinks about the reasoning process in the mind and then provides the user with the answer. The reasoning process and answer are enclosed within <think>...</think> and <answer>...</answer> tags, respectively, i.e., <think> reasoning process here </think> <answer> answer here </answer>.",
     },
 }