Update app.py

app.py

@@ -1,62 +1,149 @@
+import os
+import time
+#import spaces
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 import gradio as gr
-from huggingface_hub import InferenceClient
+from threading import Thread
 
+MODEL_LIST = ["GoidaAlignment/GOIDA-0.5B"]
+HF_TOKEN = os.environ.get("HF_TOKEN", None)
+
+TITLE = "<h1><center>Я СКАЗАЛ ГОООЙДА!</center></h1>"
+
+PLACEHOLDER = """
+<center>
+<p>ГООООЙДА!!</p>
+</center>
 """
-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("GoidaAlignment/GOIDA-0.5B")
 
+# pip install transformers
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+device = "cpu" # for GPU usage or "cpu" for CPU usage
+
+tokenizer0 = AutoTokenizer.from_pretrained(MODEL_LIST[0])
+model0 = AutoModelForCausalLM.from_pretrained(MODEL_LIST[0]).to(device)
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    max_tokens,
-    temperature,
-    top_p,
+tokenizer1 = AutoTokenizer.from_pretrained(MODEL_LIST[1])
+model1 = AutoModelForCausalLM.from_pretrained(MODEL_LIST[1]).to(device)
+
+tokenizer2 = AutoTokenizer.from_pretrained(MODEL_LIST[2])
+model2 = AutoModelForCausalLM.from_pretrained(MODEL_LIST[2]).to(device)
+
+#@spaces.GPU()
+def stream_chat(
+    message: str, 
+    history: list, 
+    temperature: float = 0.8, 
+    max_new_tokens: int = 1024, 
+    top_p: float = 1.0, 
+    top_k: int = 20, 
+    penalty: float = 1.2,
+    choice: str = "GoidaAlignment/GOIDA-0.5B"
 ):
-    messages = []
+    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 = []
+    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 = ""
+    
+    model = model2
+    tokenizer = tokenizer2
 
-    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
+    input_text=tokenizer.apply_chat_template(conversation,  add_generation_prompt=True, tokenize=False)
+    inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
+    streamer = TextIteratorStreamer(tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True)
+    
+    generate_kwargs = dict(
+        input_ids=inputs, 
+        max_new_tokens = max_new_tokens,
+        do_sample = False if temperature == 0 else True,
+        top_p = top_p,
+        top_k = top_k,
+        temperature = temperature,
+        streamer=streamer,
+    )
 
-        response += token
-        yield response
+    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
 
+            
+    #print(tokenizer.decode(outputs[0]))
 
-"""
-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.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)",
-        ),
-    ],
-)
+chatbot = gr.Chatbot(height=600, placeholder=PLACEHOLDER)
+
+with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
+    gr.HTML(TITLE)
+    gr.ChatInterface(
+        fn=stream_chat,
+        chatbot=chatbot,
+        fill_height=True,
+        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
+        additional_inputs=[
+            gr.Slider(
+                minimum=0,
+                maximum=1,
+                step=0.1,
+                value=0.4,
+                label="Temperature",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=128,
+                maximum=8192,
+                step=1,
+                value=1024,
+                label="Max new tokens",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=0.0,
+                maximum=1.0,
+                step=0.1,
+                value=1.0,
+                label="top_p",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=1,
+                maximum=20,
+                step=1,
+                value=20,
+                label="top_k",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=0.0,
+                maximum=2.0,
+                step=0.1,
+                value=1.2,
+                label="Repetition penalty",
+                render=False,
+            ),
+            gr.Radio(
+                ["GoidaAlignment/GOIDA-0.5B"],
+                value="494M",
+                label="Load Model",
+                render=False,
+            ),
+        ],
+        cache_examples=False,
+    )
 
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()