added cuda support

app.py

@@ -1,15 +1,19 @@
 from huggingface_hub import InferenceClient
 import gradio as gr
+import torch
+
+# Sprawdź, czy CUDA jest dostępne
+device = "cuda" if torch.cuda.is_available() else "cpu"
 
 client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
 
 def format_prompt(message, history):
-  prompt = "<s>"
-  for user_prompt, bot_response in history:
-    prompt += f"[INST] {user_prompt} [/INST]"
-    prompt += f" {bot_response}</s> "
-  prompt += f"[INST] {message} [/INST]"
-  return prompt
+    prompt = "<s>"
+    for user_prompt, bot_response in history:
+        prompt += f"[INST] {user_prompt} [/INST]"
+        prompt += f" {bot_response}</s> "
+    prompt += f"[INST] {message} [/INST]"
+    return prompt
 
 def generate(
     prompt, history, temperature=0, max_new_tokens=3500, top_p=0.95, repetition_penalty=1.0,
@@ -26,6 +30,7 @@ def generate(
         repetition_penalty=repetition_penalty,
         do_sample=True,
         seed=42,
+        device=device,  # Dodajemy obsługę CUDA
     )
 
     formatted_prompt = format_prompt(prompt, history)
@@ -38,15 +43,14 @@ def generate(
         yield output
     return output
 
-
 mychatbot = gr.Chatbot(
-    avatar_images=["./user.png", "./botm.png"], bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,)
+    bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,)
 
 demo = gr.ChatInterface(fn=generate,
                         chatbot=mychatbot,
-                        title="Tomoniai's Mixtral 8x7b Chat",
+                        title="Test API :)",
                         retry_btn=None,
                         undo_btn=None
                        )
 
-demo.queue().launch(show_api=True)
+demo.queue().launch(show_api=True)

requirements.txt

@@ -1,3 +1,5 @@
 huggingface_hub
 gradio
 numpy<2
+torch
+bitsandbytes