ы

app.py

@@ -2,7 +2,6 @@ import os
 import gradio as gr
 import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
-from accelerate import dispatch_model
 
 # Токен, если модель приватная
 hf_token = os.getenv("HF_TOKEN", None)
@@ -20,16 +19,11 @@ model = AutoModelForCausalLM.from_pretrained(
     model_path,
     token=hf_token,
     torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+    device_map="auto",
     trust_remote_code=True,
 )
 
-# Dispatch model to devices with offloading
 os.makedirs("offload", exist_ok=True)
-model = dispatch_model(
-    model,
-    device_map="auto",
-    offload_state_dict=True,
-)
 
 # Создаём пайплайн
 pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)