File size: 1,535 Bytes
2df5a65 4aab9a8 2df5a65 |
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 |
---
language:
- en
library_name: transformers
tags:
- auto-gptq
- AutoRound
license: apache-2.0
---
## Model Details
This is [Qwen/Qwen2.5-1.5B](https://huggingface.co/Qwen/Qwen2.5-1.5B) quantized with [AutoRound](https://github.com/intel/auto-round/tree/main) (asymmetric quantization) and serialized with the GPTQ format in 4-bit. The model has been created, tested, and evaluated by The Kaitchup.
Details on the quantization process and how to use the model here:
[The Best Quantization Methods to Run Llama 3.1 on Your GPU](https://newsletter.kaitchup.com/p/the-best-quantization-methods-to)
It is possible to fine-tune an adapter on top of it following the QLoRA methodology. More about this here:
[QLoRA with AutoRound: Cheaper and Better LLM Fine-tuning on Your GPU](https://newsletter.kaitchup.com/p/qlora-with-autoround-cheaper-and)
I used these hyperparameters for quantization:
```
bits, group_size = 4, 128
autoround = AutoRound(model, tokenizer, nsamples=512, iters=1000, low_gpu_mem_usage=False, bits=bits, group_size=group_size)
autoround.quantize()
output_dir = "./tmp_autoround"
autoround.save_quantized(output_dir, format='auto_gptq', inplace=True)
```
Evaluation results (zero-shot evaluation with lm_eval):
- **Developed by:** [The Kaitchup](https://newsletter.kaitchup.com/)
- **Language(s) (NLP):** English
- **License:** Apache 2.0 license |