A two-step power-of-two quantization method improves low-bit LLaMA perplexity and dequantization speed, but reporting inconsistencies and missing code temper the result.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
PoTPTQ: A Two-step Power-of-Two Post-training for LLMs
A two-step power-of-two quantization method improves low-bit LLaMA perplexity and dequantization speed, but reporting inconsistencies and missing code temper the result.