LLaMA3-Quantization is the official implementation of our paper How Good Are Low-bit Quantized LLAMA3 Models? An Empirical Study [PDF]. Created by researchers from The University of Hong Kong, Beihang University and ETH Zürich.
Meta's LLaMa family has become one of the most powerful open-source Large Language Model (LLM) series. Notably, LLaMa3 models have recently been released and achieve impressive performance across various with super-large scale pre-training on over 15T tokens of data. Given the wide application of low-bit quantization for LLMs in resource-limited scenarios, we explore LLaMa3's capabilities when quantized to low bit-width. This exploration holds the potential to unveil new insights and challenges for low-bit quantization of LLaMa3 and other forthcoming LLMs, especially in addressing performance degradation problems that suffer in LLM compression. Specifically, we evaluate the 10 existing post-training quantization and LoRA-finetuning methods of LLaMa3 on 1-8 bits and diverse datasets to comprehensively reveal LLaMa3's low-bit quantization performance. Our experiment results indicate that LLaMa3 still suffers non-negligent degradation in these scenarios, especially in ultra-low bit-width. This highlights the significant performance gap under low bit-width that needs to be bridged in future developments. We expect that this empirical study will prove valuable in advancing future models, pushing the LLMs to lower bit-width with higher accuracy for being practical. Our project is released on https://github.com/Macaronlin/LLaMA3-Quantization and quantized LLaMa3 models are released in https://huggingface.co/LLMQ.
We provide full script to evaluate various quantization methods in ./scripts/
. We use LLaMa-3-8B in IR-QLoRA method as an example here:
python main.py \
--model meta-llama/Meta-Llama-3-8B \
--peft LLMQ/LLaMA-3-8B-IR-QLoRA \
--tau_range 0.1 --tau_n 100--blocksize 256 \
--epochs 0 \
--output_dir ./log/llama-3-8b-irqlora \
--wbits 4 \
--tasks piqa,arc_easy,arc_challenge,hellaswag,winogrande
-
Evaluation results of post-training quantization on LLAMA3-8B model.
-
Evaluation results of post-training quantization on LLAMA3-70B model.
GPTQ: Accurate Post-training Compression for Generative Pretrained Transformers
AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration
RPTQ: Reorder-Based Post-Training Quantization for Large Language Models
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
PB-LLM: Partially Binarized Large Language Models
BiLLM: Pushing the Limit of Post-Training Quantization for LLMs
SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models
QLoRA: Efficient Finetuning of Quantized LLMs
IR-QLoRA: Accurate LoRA-Finetuning Quantization of LLMs via Information Retention