Disabling 4-bit quantization

app.py

@@ -1,16 +1,15 @@
 import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
-import bitsandbytes as bnb
 import accelerate
 print("Bitsandbytes version:", bnb.__version__)
 print("Accelerate version:", accelerate.__version__)
 
 # Load the original model
-original_model = AutoModelForCausalLM.from_pretrained("unsloth/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit")
+original_model = AutoModelForCausalLM.from_pretrained("unsloth/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit",load_in_4bit=False)
 original_tokenizer = AutoTokenizer.from_pretrained("unsloth/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit")
 
 # Load the fine-tuned model
-fine_tuned_model = AutoModelForCausalLM.from_pretrained("kas1/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit-John1")
+fine_tuned_model = AutoModelForCausalLM.from_pretrained("kas1/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit-John1",load_in_4bit=False)
 fine_tuned_tokenizer = AutoTokenizer.from_pretrained("kas1/DeepSeek-R1-Distill-Llama-8B-unsloth-bnb-4bit-John1")
 
 # Function to generate responses from both models

requirements.txt

@@ -1,5 +1,4 @@
 torch
 transformers
 gradio
-accelerate>=0.26.0
-bitsandbytes
+accelerate>=0.26.0