Share your thoughts, 1 month free Claude Pro on usSee more

Question Answering on PIQA (Accuracy)

86.5Accuracy

Mashup Learning

Updated 14d ago

Evaluation Results

Method	Links
Mashup Learning 2026.03		86.5
BF16 2026.04		85.6
From scratch 2026.03		85.2
Full-precision 2026.04		85.12
Text-to-LoRA used as init 2026.03		85.1
Full-precision 2026.04		83.68
Gemma 2 9B 2026.03		83
Focus (Gemma 2 9B) 2026.03		83
CodeQuant 2026.04		82.7
Mistral 7B 2026.03		82.6
Focus (Mistral 7B) 2026.03		82.6
LLaMA-2 70B 2026.03		82.4
Focus (LLaMA-2 70B) 2026.03		82.4
Uniform 2026.04		82.32
PMQ 2026.04		82.05
STL 2026.02		82
STL-aug 2026.02		82
SCANS 2026.02		82
Router norm + Max var 2026.04		81.83
Baseline 2025.05		81.8
PMQ 2026.04		81.56
Router norm + Max var 2026.04		81.45
Uniform 2026.04		81.45
Surgical 2026.02		81
OLMo-2 7B 2026.03		81
Focus (OLMo-2 7B) 2026.03		81
SARQC-GBS(Saliency) 2026.05		80.96
FP16 2026.04		80.9
LLaMA-3 8B 2026.02		80.85
Router norm + Max var 2026.04		80.79
SARQC-GS(Identity) 2026.05		80.79
Llama3 8B 2025.09		80.7
Tensor RMS + C 2025.05		80.7
Llama3 8B 2026.02		80.7
FP16 2026.05		80.69
AWQ 2026.05		80.69
SARQC-GS(Saliency) 2026.05		80.69
FP16 2026.05		80.69
PMQ 2026.04		80.63
Mashup Learning at init 2026.03		80.6
SARQC-GBS(Identity) 2026.05		80.58
BF16 2026.05		80.52
BF16 2026.04		80.5
PMQ 2026.04		80.47
Baseline 2026.04		80.47
Router norm + Max var 2026.04		80.41
LLaMA-2 13B 2026.03		80.4
Focus (LLaMA-2 13B) 2026.03		80.4
BF16 2026.04		80.4
CodeQuant 2026.04		80.4
Baseline 2026.04		80.36
GPTAQ 2026.05		80.25
Hessian 2026.04		80.21
Text-to-LoRA 2026.03		80.2
Router norm + Max var 2026.04		80.2
Baseline (Dense) 2026.05		80.2
PMQ 2026.04		80.14
Router norm + Max var 2026.04		80.14
GPTQ 2026.05		80.14
SARQC-GBS(Saliency) 2026.05		80.14
SARQC-GBS(Saliency) 2026.05		80.03
SARQC-GBS(Identity) 2026.05		80.03
STL 2026.02		80
STL-aug 2026.02		80
Surgical 2026.02		80
Qwen 2.5 7B 2026.03		80
Focus (Qwen 2.5 7B) 2026.03		80
SARQC-GBS(Saliency) 2026.05		79.92
CodeQuant 2026.04		79.8
Baseline 2026.04		79.76
Baseline 2026.04		79.76
BF16 2026.04		79.7
CodeQuant 2026.04		79.7
FP16 2026.05		79.65
FP16 2026.05		79.65
CodeQuant 2026.04		79.6
DuQuant++* 2026.04		79.5
PMQ 2026.04		79.49
CodeQuant 2026.04		79.4
GPTQ 2026.05		79.38
SARQC-GBS(Identity) 2026.05		79.38
DuQuant++ 2026.04		79.3
Router norm + Max var 2026.04		79.27
Baseline 2026.04		79.27
SARQC-GS(Saliency) 2026.05		79.22
Hessian 2026.04		79.21
CodeQuant 2026.04		79.2
CodeQuant 2026.04		79.2
SARQC-GBS(Identity) 2026.05		79.16
LLaMA-2 7B 2026.02		79.11
BF16 2026.05		79.11
LinUCB+KL 2026.02		79.1
CodeQuant 2026.04		79.1
SCANS 2026.02		79
DCR 2026.02		79
DCR 2026.02		79
PMQ 2026.04		79
SARQC-GS(Identity) 2026.05		79
Router norm + Max var 2026.04		78.94
AWQ 2026.05		78.94

Showing 100 of 505 rows