wip

app.py

@@ -1,12 +1,21 @@
 import gradio as gr
+import spaces
 from huggingface_hub import InferenceClient
+import torch
+from transformers import AutoModelForCausalLM, ChameleonProcessor, AutoTokenizer, StoppingCriteria, StoppingCriteriaList, TextIteratorStreamer
+from threading import Thread
+from PIL import Image
+import requests
 
-"""
-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_path = "facebook/chameleon-7b"
+# model = ChameleonForCausalLM.from_pretrained(model_path, torch_dtype=torch.bfloat16, low_cpu_mem_usage=True, device_map="auto")
+# processor = ChameleonProcessor.from_pretrained(model_path)
+model = AutoModelForCausalLM.from_pretrained(model_path, torch_dtype=torch.bfloat16, low_cpu_mem_usage=True, device_map="auto")
+processor = ChameleonProcessor.from_pretrained(model_path)
+tokenizer = processor.tokenizer
 
+@spaces.GPU
 def respond(
     message,
     history: list[tuple[str, str]],
@@ -15,35 +24,42 @@ def respond(
     temperature,
     top_p,
 ):
-    messages = [{"role": "system", "content": system_message}]
+    # 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 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})
+    # 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
+    prompt = "I'm very intrigued by this work of art:<image>Please tell me about the artist."
+    image = Image.open(requests.get("https://uploads4.wikiart.org/images/paul-klee/death-for-the-idea-1915.jpg!Large.jpg", stream=True).raw)
+
+    inputs = processor(prompt, images=[image], return_tensors="pt").to(model.device, dtype=torch.bfloat16)
+    # out = model.generate(**inputs, max_new_tokens=40, do_sample=False)
+
+    streamer = TextIteratorStreamer(tokenizer)
+    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=20)
+
+    thread = Thread(target=model.generate, kwargs=generation_kwargs)
+    thread.start()
+
+    partial_message = ""
+    for new_token in streamer:
+        partial_message += new_token
+        yield partial_message
 
-        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,
+    multimodal=True,
     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"),

requirements.txt

@@ -1 +1,3 @@
-huggingface_hub==0.22.2
+huggingface_hub>=0.23.2
+git+https://github.com/huggingface/transformers.git@chameleon
+accelerate