ROCm: Model quantization techniques (AMD Quark, GPTQ, bitsandbytes on Instinct)
repo
Fields
id
sources/quant-rocm-quant-docPublisher
AMD
Kind
repo
Published
2025-05-30
Accessed
2026-10-03
notes
States Quark FP8/INT8 support and the two-level INT4-FP8 scheme on Instinct.
Notes
States Quark FP8/INT8 support and the two-level INT4-FP8 scheme on Instinct.
Cross-references
This record references no other records.