Upload 2 files

app.py

@@ -1,64 +1,67 @@
-import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-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(model="https://w6zpj9tjouv19b9t.us-east-1.aws.endpoints.huggingface.cloud")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    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})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-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)",
-        ),
-    ],
-)
-
-
-if __name__ == "__main__":
-    demo.launch()
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer, StoppingCriteria, StoppingCriteriaList, TextIteratorStreamer
+import torch
+from threading import Thread
+
+model_name = "fzmnm/TinyStoriesAdv_v2_92M"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+model.eval()
+
+model.generation_config.pad_token_id = tokenizer.eos_token_id
+
+max_tokens = 512
+
+def build_input_str(message: str, history: 'list[list[str]]'):
+    history_str = ""
+    for entity in history:
+        if entity['role'] == 'user':
+            history_str += f"问：{entity['content']}\n\n"
+        elif entity['role'] == 'assistant':
+            history_str += f"答：{entity['content']}\n\n"
+    return history_str + f"问：{message}\n\n"
+
+def stop_criteria(input_str):
+    return input_str.endswith("\n") and len(input_str.strip()) > 0
+
+class StopOnTokens(StoppingCriteria):
+    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
+        input_str = tokenizer.decode(input_ids[0], skip_special_tokens=True)
+        return stop_criteria(input_str)
+    
+def chat(message, history):
+    input_str = build_input_str(message, history)
+    input_ids = tokenizer.encode(input_str, return_tensors="pt")
+    input_ids = input_ids[:, -max_tokens:]
+    streamer = TextIteratorStreamer(
+        tokenizer, 
+        timeout=10, 
+        skip_prompt=True, 
+        skip_special_tokens=True)
+    stopping_criteria = StoppingCriteriaList([StopOnTokens()])
+    generate_kwargs = dict(
+        input_ids=input_ids,
+        streamer=streamer,
+        stopping_criteria=stopping_criteria,
+        max_new_tokens=512,
+        top_p=0.9,
+        do_sample=True,
+        temperature=0.7
+    )
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
+
+    output_str = ""
+    for new_str in streamer:
+        output_str += new_str
+        yield output_str
+
+app = gr.ChatInterface(
+    fn=chat, 
+    type='messages',
+    examples=['什么是鹦鹉？', '什么是大象？', '谁是李白？', '什么是黑洞？'],
+    title='聊天机器人',
+)
+
+app.launch()

requirements.txt

@@ -1 +1,3 @@
-huggingface_hub==0.25.2
+huggingface_hub==0.25.2
+transformers
+gradio