Upload 2 files

app.py

@@ -1,18 +1,20 @@
 import gradio as gr
 import torch
 from transformers import Qwen2ForCausalLM, AutoTokenizer
+from peft import PeftModel
 
+base_model_id = "Qwen/Qwen2.5-3B-Instruct"   # base
+adapter_path = "Castial/FinQwen-2.5b-v2"    # your fine-tuned adapter
 
-checkpoint = "Castial/FinQwen-2.5b-v2"
 
-# Load model + tokenizer
-model = Qwen2ForCausalLM.from_pretrained(
-    checkpoint,
+tokenizer = AutoTokenizer.from_pretrained(base_model_id)
+base_model = AutoModelForCausalLM.from_pretrained(
+    base_model_id,
     device_map="auto",
-    dtype="auto",
+    torch_dtype="auto",
 )
-tokenizer = AutoTokenizer.from_pretrained(checkpoint)
 
+model = PeftModel.from_pretrained(base_model, adapter_path)
 
 def respond(
     message,
@@ -80,5 +82,5 @@ chatbot = gr.ChatInterface(
 with gr.Blocks() as demo:
     chatbot.render()
 
-if name == "main":
+if __name__ == "__main__":
     demo.launch()

requirements.txt

@@ -2,5 +2,5 @@ torch
 transformers
 gradio
 accelerate
-
+peft