Update space

.build_config

@@ -1,6 +0,0 @@
-# .build_config
-build:
-  cuda: "11.8"
-  python_version: "3.10"
-  system_packages:
-    - "nvidia-cuda-toolkit"

README.md

@@ -8,16 +8,6 @@ sdk_version: 4.19.2
 app_file: app.py
 pinned: false
 python_version: "3.10"
-cuda: "11.8"
-license: "mit"
-duplicated_from: unsloth/Llama-3.2-3B-Instruct-bnb-4bit
-models:
-  - unsloth/Llama-3.2-3B-Instruct-bnb-4bit
-  - emeses/lab2_model
-tags:
-  - llama
-  - chat
-  - gradio
 ---
 
 # LLaMA Chat Interface

app.py

@@ -17,16 +17,17 @@ bnb_config = BitsAndBytesConfig(
     load_in_4bit=True,
     bnb_4bit_compute_dtype=torch.float16,
     bnb_4bit_quant_type="nf4",
-    bnb_4bit_use_double_quant=True,
+    bnb_4bit_use_double_quant=True
 )
 
 # Load model and tokenizer
 base_model = AutoModelForCausalLM.from_pretrained(
     "unsloth/Llama-3.2-3B-Instruct-bnb-4bit",
+    quantization_config=bnb_config,
     device_map="auto",
-    torch_dtype=torch.float16,
-    quantization_config=bnb_config
+    torch_dtype=torch.float16
 )
+
 model = PeftModel.from_pretrained(base_model, "emeses/lab2_model")
 tokenizer = AutoTokenizer.from_pretrained("unsloth/Llama-3.2-3B-Instruct-bnb-4bit")
 

requirements.txt

@@ -1,8 +1,8 @@
-huggingface_hub==0.25.2
-gradio==4.19.2
-transformers
-peft==0.7.1
 torch==2.1.2
+transformers==4.36.2
 accelerate==0.26.1
 bitsandbytes==0.41.3
-torch==2.1.2
+peft==0.7.1
+huggingface_hub==0.25.2
+gradio==4.19.2
+sentencepiece