LLM quantization
FIMA-Q
Superseded baseline#28 of 80 most-superseded · first seen Jun 13, 2025
Superseded — cited as a baseline and beaten by newer methods
2 papers critique it · 1 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites FIMA-Q as a baseline.
FIMA-Q wu2025fima proposes an efficient Fisher Information Matrix approximation to guide quantization via block-wise reconstruction loss, pushing ViT PTQ toward W3A3 and W4A4 with notable improvement. However, the block-level independence neglects inter-block compensation, leaving further potential unexploited.
“These strategies are largely layer or block-centric and optimize average fidelity. In COD at W4A4, the dominant failure is token-local: token-wise activation heterogeneity lets background tokens dominate the range and increase the zero-bin mass, which layer-wise fitting or reconstruction does not explicitly bound.”
Beaten on benchmarks
Head-to-head results where a newer method reports beating FIMA-Q. Values are copied from the source paper's tables — verify against the cited paper.
Joint Post-Training Quantization of Vision Transformers with Learned Prompt-Guided Data Generation beats FIMA-Q
68.45 vs 4.84
Top-1 accuracy · [ViT-S, W1.58A8]
Joint Post-Training Quantization of Vision Transformers with Learned Prompt-Guided Data Generation
What to use instead
Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.
- COD-TDQWhen W4A4 Breaks Camouflaged Object Detection: Token-Group Dual-Constraint Activation QuantizationApr 18, 2026
- Joint Post-Training Quantization of Vision Transformers with Learned Prompt-Guided Data GenerationJoint Post-Training Quantization of Vision Transformers with Learned Prompt-Guided Data GenerationFeb 21, 2026