any4: Learned 4-bit Numeric Representation for LLMs
This work addresses the challenge of efficient model compression for LLMs, enabling faster inference with reduced memory usage, though it is incremental as it builds on existing quantization techniques.
The paper tackles the problem of 4-bit weight quantization for large language models (LLMs) by introducing any4, a learned numeric representation that achieves higher accuracy than existing 4-bit types like int4, fp4, and nf4 across various model sizes and families, and is competitive with preprocessing-based methods like AWQ and GPTQ.
We present any4, a learned 4-bit weight quantization solution for large language models (LLMs) providing arbitrary numeric representations without requiring pre-processing of weights or activations. any4 yields higher accuracy compared to other related 4-bit numeric representation types: int4, fp4 and nf4, as evaluated on a range of model sizes, generations and families (Llama 2, Llama 3, Mistral and Mixtral). While any4 does not require preprocessing of weights or activations, it is also competitive with orthogonal techniques that require such preprocessing (e.g., AWQ and GPTQ). We also experiment with any3 and any2 and show competitiveness at lower bits. Additionally, we show that we can calibrate using a single curated diverse sample rather than hundreds of samples from a dataset as done in most quantization approaches. We also open source tinygemm, a latency optimized GPU matrix multiplication library for LLMs, that implements any4 using a GPU-efficient lookup table strategy along with other common quantization methods. We open source our code at https://github.com/facebookresearch/any4 .