afnotes commited on
Commit
ebb8aa3
·
1 Parent(s): 8170a49

Edited app.py again :(

Browse files
Files changed (1) hide show
  1. app.py +12 -1
app.py CHANGED
@@ -1,9 +1,20 @@
1
  import gradio as gr
2
  import torch
3
- from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig # Import BitsAndBytesConfig
4
  from peft import PeftModel
5
  import os
6
 
 
 
 
 
 
 
 
 
 
 
 
7
  # --- Configuration ---
8
  BASE_MODEL_ID = "Qwen/Qwen3-4B"
9
  ADAPTER_MODEL_ID = "afnotes/Qwen3-4B-YeI-LoRA"
 
1
  import gradio as gr
2
  import torch
3
+ from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
4
  from peft import PeftModel
5
  import os
6
 
7
+ # Add this GPU check here
8
+ if not torch.cuda.is_available():
9
+ # If CUDA is not available, try to explain and possibly fall back to CPU if memory allows (unlikely for 4B)
10
+ print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
11
+ print("WARNING: CUDA (GPU) is NOT available on this Space instance.")
12
+ print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
13
+ # You might consider removing the quantization_config and trying CPU, but for 4B it will likely crash
14
+ # For now, we'll let it fail if it can't find a GPU as intended.
15
+ else:
16
+ print(f"CUDA is available! PyTorch detected GPU: {torch.cuda.get_device_name(0)}")
17
+
18
  # --- Configuration ---
19
  BASE_MODEL_ID = "Qwen/Qwen3-4B"
20
  ADAPTER_MODEL_ID = "afnotes/Qwen3-4B-YeI-LoRA"