Superseded baseline#41 of 80 most-superseded
MASQuant
LLM quantization
superseded — cited as a baseline and beaten by newer methods
1 papers critique it · 1 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites MASQuant as a baseline.
“Some methods~yu2025mquant,hu2026masquant also attempt to handle the two modalities separately, but such treatment does not fully consider the modality-specific channel outliers.”
— Breaking Modality Heterogeneity in Low-Bit Quantization for Large Vision-Language Models
Beaten on benchmarks
Head-to-head results where a newer method reports beating MASQuant. Values are copied from the source paper's tables — verify against the cited paper.
- Breaking Modality Heterogeneity in Low-Bit Quantization for Large Vision-Language Models
SplitQ beats MASQuant · Avg. [W4A8]
70.4 vs 63.9
- Breaking Modality Heterogeneity in Low-Bit Quantization for Large Vision-Language Models
SplitQ beats MASQuant · Avg. [W4A4]
69.6 vs 5.7
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.
- May 19, 2026
- May 18, 2026
- Quantization-aware Integrated Gradients (QIG)Fine-Grained Post-Training Quantization for Large Vision Language Models with Quantization-Aware Integrated GradientsMar 18, 2026
- SPEED-QSPEED-Q: Staged Processing with Enhanced Distillation towards Efficient Low-bit On-device VLM QuantizationNov 12, 2025
- Quant-dLLMQuant-dLLM: Post-Training Extreme Low-Bit Quantization for Diffusion Large Language ModelsSep 27, 2025