File size: 1,736 Bytes
2c61adb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d4e3895
2c61adb
 
 
 
 
 
 
d4e3895
2c61adb
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
---
language: en
license: apache-2.0
library_name: transformers
---

# SQFT Fine-tuned Model: sqft-sparsepeft-phi-3-mini-4k-30-math-heu

- Base Model: [IntelLabs/sqft-phi-3-mini-4k-30-base](https://huggingface.co./IntelLabs/sqft-phi-3-mini-4k-30-base)
- Sparsity: 30%
- Quantization: No
- Finetune Method: SQFT + SparsePEFT
- Finetune data: 10K instruction-following math reasoning training dataset from [LLM-Adapters](https://github.com/AGI-Edgerunners/LLM-Adapters) ([math_10k.json](https://github.com/AGI-Edgerunners/LLM-Adapters/blob/main/ft-training_set/math_10k.json))
- Sub-Adapter: Heuristic

### Evaluation

```bash
git clone https://github.com/IntelLabs/Hardware-Aware-Automated-Machine-Learning.git haaml && cd haaml/SQFT

MODEL_NAME=IntelLabs/sqft-sparsepeft-phi-3-mini-4k-30-math-heu
OUTPUT_DIR=./results
python eval/evaluate_math.py --base_model_path ${MODEL_NAME} --output_dir ${OUTPUT_DIR}
```

Refer to our [repo](https://github.com/IntelLabs/Hardware-Aware-Automated-Machine-Learning/tree/main/SQFT) for the environment information to run this command.

## Model Sources

- **Repository:** [https://github.com/IntelLabs/Hardware-Aware-Automated-Machine-Learning/tree/main/SQFT](https://github.com/IntelLabs/Hardware-Aware-Automated-Machine-Learning/tree/main/SQFT)
- **Paper:** [SQFT: Low-cost Model Adaptation in Low-precision Sparse Foundation Models](https://arxiv.org/abs/2410.03750)

## Citation

```bash
@article{munoz2024sqft,
  title = {SQFT: Low-cost Model Adaptation in Low-precision Sparse Foundation Models},
  author={J. Pablo Munoz and Jinjie Yuan and Nilesh Jain},
  journal={The 2024 Conference on Empirical Methods in Natural Language Processing (Findings)},
  year={2024}
}
```

## License

Apache-2.0