upl codes

app.py

@@ -1,64 +1,85 @@
+import torch
 import gradio as gr
-from huggingface_hub import InferenceClient
+from threading import Thread
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
+MODEL_ID = "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B"
+MODEL_NAME = MODEL_ID.split("/")[-1]
+CONTEXT_LENGTH = 16000
+DESCRIPTION = f"This is {MODEL_NAME} model designed for testing thinking for general AI tasks. <br>当前仅提供 HuggingFace 版部署实例，有算力的可自行克隆至本地或复刻至购买了 GPU 环境的账号测试"
 
-def respond(
+
+def predict(
     message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
+    history,
+    system_prompt,
     temperature,
+    max_new_tokens,
+    top_k,
+    repetition_penalty,
     top_p,
 ):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
+    # Format history with a given chat template
+    stop_tokens = ["<|endoftext|>", "<|im_end|>", "|im_end|"]
+    instruction = "<|im_start|>system\n" + system_prompt + "\n<|im_end|>\n"
+    for user, assistant in history:
+        instruction += f"<|im_start|>user\n{user}\n<|im_end|>\n<|im_start|>assistant\n{assistant}\n<|im_end|>\n"
 
-    response = ""
+    instruction += f"<|im_start|>user\n{message}\n<|im_end|>\n<|im_start|>assistant\n"
+    print(instruction)
+    streamer = TextIteratorStreamer(
+        tokenizer,
+        skip_prompt=True,
+        skip_special_tokens=True,
+    )
+    enc = tokenizer(instruction, return_tensors="pt", padding=True, truncation=True)
+    input_ids, attention_mask = enc.input_ids, enc.attention_mask
+    if input_ids.shape[1] > CONTEXT_LENGTH:
+        input_ids = input_ids[:, -CONTEXT_LENGTH:]
+        attention_mask = attention_mask[:, -CONTEXT_LENGTH:]
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+    generate_kwargs = dict(
+        input_ids=input_ids.to(device),
+        attention_mask=attention_mask.to(device),
+        streamer=streamer,
+        do_sample=True,
         temperature=temperature,
+        max_new_tokens=max_new_tokens,
+        top_k=top_k,
+        repetition_penalty=repetition_penalty,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
+    )
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
+    outputs = []
+    for new_token in streamer:
+        outputs.append(new_token)
+        if new_token in stop_tokens:
+            break
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+        yield "".join(outputs)
 
 
 if __name__ == "__main__":
-    demo.launch()
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+    model = AutoModelForCausalLM.from_pretrained(MODEL_ID, device_map="auto")
+    # Create Gradio interface
+    gr.ChatInterface(
+        predict,
+        title=f"{MODEL_NAME} 部署实例",
+        description=DESCRIPTION,
+        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False),
+        additional_inputs=[
+            gr.Textbox(
+                "You are a useful assistant. first recognize user request and then reply carfuly and thinking",
+                label="System prompt",
+            ),
+            gr.Slider(0, 1, 0.6, label="Temperature"),
+            gr.Slider(0, 32000, 10000, label="Max new tokens"),
+            gr.Slider(1, 80, 40, label="Top K sampling"),
+            gr.Slider(0, 2, 1.1, label="Repetition penalty"),
+            gr.Slider(0, 1, 0.95, label="Top P sampling"),
+        ],
+    ).queue().launch()