fix: remove quantization, use float16 only

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+from transformers import AutoModelForCausalLM, AutoTokenizer
 from peft import PeftModel
 
 # Model configurations
@@ -11,16 +11,9 @@ print("Loading tokenizer...")
 tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
 tokenizer.pad_token = tokenizer.eos_token
 
-# Configure quantization for 4-bit
-bnb_config = BitsAndBytesConfig(
-    load_in_4bit=True,
-    bnb_4bit_compute_dtype=torch.float16
-)
-
 print("Loading base model...")
 model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
-    quantization_config=bnb_config,
     device_map="auto",
     torch_dtype=torch.float16
 )

requirements.txt

@@ -1,7 +1,5 @@
 torch>=2.0.0
 transformers>=4.30.0
 accelerate>=0.20.0
-bitsandbytes>=0.41.1
 peft==0.6.2
 gradio>=4.8.0
-scipy>=1.11.0