.

meta_ui.py

@@ -91,9 +91,9 @@ def backend(text, sampling, api_suffix):
             fr = choice.get("finish_reason")  # 如果后续需要 finish reason
 
         log(f"🟢 [{now()}] HTTP {r.status_code}  tokens={ctok}  finish={fr}\n"
-            f"🟢 resp800={r.text[:800]!r}")
+            f"🟢 resp={r.text!r}") 
         if r.status_code != 200:
-            return f"[HTTP {r.status_code}] {r.text[:300]}"
+            return f"[HTTP {r.status_code}] {r.text}"
         return txt or "[⚠ 空]"
     except Exception as e:
         log(f"[❌ 请求异常] {e}")
@@ -174,82 +174,6 @@ def chat(
         return
 
     
-
-# # ────────────────── Chat 回调 ──────────────────
-# def chat(
-#     user_msg, history,
-#     max_new, temp, top_p, top_k,
-#     rep_pen, pres_pen, stop_raw,
-#     api_suffix, log_state
-# ):
-#     from queue import Queue, Empty
-
-#     # 解析传入的 ChatInput 格式
-#     user = user_msg["text"] if isinstance(user_msg, dict) and "text" in user_msg else user_msg
-
-#     # 构造 OpenAI 风格 messages，仅用于 /v1/chat/completions
-#     if api_suffix == "/v1/chat/completions":
-#         messages = []
-#         messages = history[:]  # 正确使用 OpenAI 格式
-#         messages.append({"role": "user", "content": user})
-#         prompt_input = messages
-
-#     else:
-#         prompt_input = user  # 原来的单轮文本 prompt
-
-#     stop = [s.strip() for s in stop_raw.split(",") if s.strip()] or None
-#     samp = {
-#         ("max_tokens" if api_suffix == "/v1/completions" else "max_new_tokens"): int(max_new),
-#         "temperature": temp,
-#         "top_p": top_p,
-#         "top_k": int(top_k),
-#         "repetition_penalty": rep_pen,
-#         "presence_penalty": pres_pen,
-#         **({"stop": stop} if stop else {})
-#     }
-
-#     result_q = Queue()
-
-#     def worker():
-#         out = backend(prompt_input, samp, api_suffix)
-#         result_q.put(out)
-
-#     thread = threading.Thread(target=worker, daemon=True) 
-#     thread.start()      
-
-#     if api_suffix == "/v1/chat/completions":
-#         while True:
-#             if not thread.is_alive() and result_q.empty():
-#                 break
-#             try:
-#                 result = result_q.get(timeout=0.1)
-#             except Empty:
-#                 continue
-
-#             if isinstance(result, str):
-#                 result = {"text": result}
-#             elif not isinstance(result, dict) or "text" not in result:
-#                 result = {"text": str(result)}
-
-#             history.append({"role": "assistant", "content": result["text"]})
-#             yield result["text"], None  # ✅ 显示模型输出，同时更新 history
-#         return
-#     else:
-#         while thread.is_alive():
-#             try:
-#                 result = result_q.get(timeout=0.1)
-#                 break
-#             except Empty:
-#                 continue
-
-#         if isinstance(result, str):
-#             result = {"text": result}
-#         elif not isinstance(result, dict) or "text" not in result:
-#             result = {"text": str(result)}
-
-#         yield result["text"], log_state  # ✅ 其它接口只输出文本，不更新 history
-#         return
-    
 # ────────────────── Gradio UI ──────────────────
 with gr.Blocks(title="调试界面") as demo:
     gr.Markdown(f"## 💬 调试界面  \n权重 **{MODEL_PATH.name}**")