inference/research
verifiedhigh · 0.88updated 2026-10-03

LLM-FP4: 4-Bit Floating-Point Quantized Transformers (Li et al.)

paper

Fields

id
sources/qlab-llm-fp4-paper
URL
https://arxiv.org/abs/2310.16836
Publisher
arXiv
Kind
paper
Published
2023-10-25
Accessed
2026-10-03
notes
Single version (v1). Abstract read directly. FP4 here is a research PTQ format, NOT NVFP4 - predates Blackwell by two years and has no relation to the NVFP4 hardware format.

Notes

Single version (v1). Abstract read directly. FP4 here is a research PTQ format, NOT NVFP4 - predates Blackwell by two years and has no relation to the NVFP4 hardware format.

Cross-references

This record references no other records.