---
license: mit
---

# Compressed LLM Model Zone

The models are prepared by [Visual Informatics Group @ University of Texas at Austin (VITA-group)](https://vita-group.github.io/).

License: [MIT License](https://opensource.org/license/mit/)

Setup environment
```shell
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install transformers==4.31.0
pip install huggingface_hub accelerate
```

How to use
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
base_model = 'llama-2-7b'
comp_degree = 0.1
comp_method = 'sparsegpt_unstructured'
model_path = f'vita-group/comp-{arch}_{comp_method}_s{comp_degree}'
model = AutoModelForCausalLM.from_pretrained(
        model_path, 
        torch_dtype=torch.float16, 
        low_cpu_mem_usage=True, 
        device_map="auto"
    )
tokenizer = AutoTokenizer.from_pretrained('meta-llama/Llama-2-7b')
input_ids = tokenizer('Hello! I am a VITA-compressed-LLM chatbot!', return_tensors='pt').input_ids
outputs = model.generate(input_ids)
```


|    | Base Model   | Model Size   | Compression Method     | Compression Degree                                                                    |
|---:|:-------------|:-------------|:-----------------------|:--------------------------------------------------------------------------------------|
|  0 | Llama-2      | 7b           | magnitude_unstructured | [s0.1](https://huggingface.co/vita-group/comp-llama-2-7b_magnitude_unstructured_s0.1) |
|  1 | Llama-2      | 7b           | magnitude_unstructured | [s0.2](https://huggingface.co/vita-group/comp-llama-2-7b_magnitude_unstructured_s0.2) |
|  2 | Llama-2      | 7b           | magnitude_unstructured | [s0.3](https://huggingface.co/vita-group/comp-llama-2-7b_magnitude_unstructured_s0.3) |
|  3 | Llama-2      | 7b           | magnitude_unstructured | [s0.5](https://huggingface.co/vita-group/comp-llama-2-7b_magnitude_unstructured_s0.5) |
|  4 | Llama-2      | 7b           | magnitude_unstructured | [s0.6](https://huggingface.co/vita-group/comp-llama-2-7b_magnitude_unstructured_s0.6) |
|  5 | Llama-2      | 7b           | sparsegpt_unstructured | [s0.1](https://huggingface.co/vita-group/comp-llama-2-7b_sparsegpt_unstructured_s0.1) |
|  6 | Llama-2      | 7b           | sparsegpt_unstructured | [s0.2](https://huggingface.co/vita-group/comp-llama-2-7b_sparsegpt_unstructured_s0.2) |
|  7 | Llama-2      | 7b           | sparsegpt_unstructured | [s0.3](https://huggingface.co/vita-group/comp-llama-2-7b_sparsegpt_unstructured_s0.3) |
|  8 | Llama-2      | 7b           | sparsegpt_unstructured | [s0.5](https://huggingface.co/vita-group/comp-llama-2-7b_sparsegpt_unstructured_s0.5) |
|  9 | Llama-2      | 7b           | sparsegpt_unstructured | [s0.6](https://huggingface.co/vita-group/comp-llama-2-7b_sparsegpt_unstructured_s0.6) |
| 10 | Llama-2      | 7b           | wanda_unstructured     | [s0.1](https://huggingface.co/vita-group/comp-llama-2-7b_wanda_unstructured_s0.1)     |
| 11 | Llama-2      | 7b           | wanda_unstructured     | [s0.2](https://huggingface.co/vita-group/comp-llama-2-7b_wanda_unstructured_s0.2)     |
| 12 | Llama-2      | 7b           | wanda_unstructured     | [s0.3](https://huggingface.co/vita-group/comp-llama-2-7b_wanda_unstructured_s0.3)     |
| 13 | Llama-2      | 7b           | wanda_unstructured     | [s0.5](https://huggingface.co/vita-group/comp-llama-2-7b_wanda_unstructured_s0.5)     |
| 14 | Llama-2      | 7b           | wanda_unstructured     | [s0.6](https://huggingface.co/vita-group/comp-llama-2-7b_wanda_unstructured_s0.6)     |