Start trying to zero-GPU

app.py

@@ -1,49 +1,105 @@
+import sys
+from threading import Thread
+
 import gradio as gr
-from huggingface_hub import InferenceClient
+import spaces
+from transformers import (
+    AutoModelForCausalLM,
+    AutoTokenizer,
+    TextIteratorStreamer,
+    BitsAndBytesConfig,
+)
+
+
+import torch
+
+
+MODEL = "microsoft/Phi-3.5-mini-instruct"
+
+if torch.cuda.is_available():
+    device = "cuda"
+elif sys.platform == "darwin" and torch.backends.mps.is_available():
+    device = "mps"
+else:
+    device = "cpu"
 
-"""
-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("microsoft/Phi-3.5-mini-instruct")
 
+# TODO understand this
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.bfloat16,
+    bnb_4bit_use_double_quant=True,
+    bnb_4bit_quant_type="nf4",
+)
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL,
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    quantization_config=quantization_config,
+)
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
+
+@spaces.GPU()
+def stream_chat(
+    message: str,
+    history: list,
+    system_prompt: str,
+    temperature: float = 0.8,
+    max_new_tokens: int = 1024,
+    top_p: float = 1.0,
+    top_k: int = 20,
+    penalty: float = 1.2,
 ):
-    messages = [{"role": "system", "content": system_message}]
+    print(f"message: {message}")
+    print(f"history: {history}")
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    conversation = [{"role": "system", "content": system_prompt}]
+    for prompt, answer in history:
+        conversation.extend(
+            [
+                {"role": "user", "content": prompt},
+                {"role": "assistant", "content": answer},
+            ]
+        )
 
-    messages.append({"role": "user", "content": message})
+    conversation.append({"role": "user", "content": message})
 
-    response = ""
+    input_ids = tokenizer.apply_chat_template(
+        conversation, add_generation_prompt=True, return_tensors="pt"
+    ).to(model.device)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
+    streamer = TextIteratorStreamer(
+        tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True
+    )
+
+    generate_kwargs = dict(
+        input_ids=input_ids,
+        max_new_tokens=max_new_tokens,
+        do_sample=False if temperature == 0 else True,
         top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        top_k=top_k,
+        temperature=temperature,
+        eos_token_id=[128001, 128008, 128009],
+        streamer=streamer,
+    )
+
+    with torch.no_grad():
+        thread = Thread(target=model.generate, kwargs=generate_kwargs)
+        thread.start()
+
+    buffer = ""
+    for new_text in streamer:
+        buffer += new_text
+        yield buffer
 
-        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,
+    stream_chat,
     additional_inputs=[
         gr.Textbox(value="You are a belligerent Chatbot.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),

requirements.txt

@@ -1 +1,2 @@
-huggingface_hub==0.22.2
+torch
+transformers