First trial

app.py

@@ -1,64 +1,73 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import spaces
 
-"""
-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")
+from threading import Thread
+from torch import bfloat16
+from transformers import Qwen2VLForConditionalGeneration, Qwen2VLProcessor, TextIteratorStreamer, AutoProcessor
+from qwen_vl_utils import process_vision_info
 
+model_path = "Pectics/Softie-VL-7B-250123"
 
+model = Qwen2VLForConditionalGeneration.from_pretrained(
+    model_path,
+    torch_dtype=bfloat16,
+    #attn_implementation="flash_attention_2",
+    device_map="auto",
+)
+min_pixels = 256 * 28 * 28
+max_pixels = 1280 * 28 * 28
+processor: Qwen2VLProcessor = AutoProcessor.from_pretrained(model_path, min_pixels=min_pixels, max_pixels=max_pixels)
+
+@spaces.GPU
 def respond(
     message,
-    history: list[tuple[str, str]],
+    history,
     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]})
-
+    for m in history:
+        messages.append({"role": m["role"], "content": m["content"]})
     messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+    
+    text_inputs = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    image_inputs, video_inputs = process_vision_info(messages)
+    inputs = processor(
+        text=[text_inputs],
+        images=image_inputs,
+        videos=video_inputs,
+        padding=True,
+        return_tensors="pt",
+    ).to("cuda")
+    
+    streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)
+    kwargs = dict(
+        **inputs,
+        streamer=streamer,
+        max_new_tokens=max_tokens,
         temperature=temperature,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
+    )
+    thread = Thread(target=model.generate, kwargs=kwargs)
+    thread.start()
+    
+    response = ""
+    for token in streamer:
         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(
+app = gr.ChatInterface(
     respond,
+    type="messages",
     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)",
-        ),
+        gr.Textbox(value="You are Softie, a helpful assistant.", label="系统设定"),
+        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="最大生成长度"),
+        gr.Slider(minimum=0.01, maximum=4.0, value=0.75, step=0.01, label="温度系数（Temperature）"),
+        gr.Slider(minimum=0.01, maximum=1.0, value=0.5, step=0.01, label="核取样系数（Top-p）"),
     ],
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    app.launch()

requirements.txt

@@ -1 +1,6 @@
-huggingface_hub==0.25.2
+torch
+torchvision
+transformers
+accelerate
+qwen-vl-utils
+gradio