List your company
ResearchFormerly WizWand

MLP neuron selection on Professions

0.24ℓ(D, m)random
-26,399,999,999.7504151,800,000,000.18478330,000,000,000.11993508,200,000,000.0551Jun 2, 2026
Updated 5mo ago

Evaluation results

MethodLinks
randomModel=GPT2-large, Sparsity level=≤2.5%, Execution time (hours)=1.352026.060.24
randomModel=GPT2-medium, Sparsity level=≤2.5%, Execution time (hours)=0.712026.060.25
randomModel=GPT2-xl, Sparsity level=≤2.5%, Execution time (hours)=2.522026.060.28
randomModel=GPT2-small, Sparsity level=≤2.5%, Execution time (hours)=0.322026.060.34
randomModel=DistilGPT2, Sparsity level=≤2.5%, Execution time (hours)=0.212026.060.36
randomModel=GPT2-large, Sparsity level=≤5%, Execution time (hours)=1.352026.060.39
randomModel=GPT2-medium, Sparsity level=≤5%, Execution time (hours)=0.712026.060.48
randomModel=GPT2-small, Sparsity level=≤5%, Execution time (hours)=0.322026.060.56
randomModel=GPT2-xl, Sparsity level=≤5%, Execution time (hours)=2.522026.060.63
randomModel=DistilGPT2, Sparsity level=≤5%, Execution time (hours)=0.212026.060.72
randomModel=GPT2-large, Sparsity level=≤10%, Execution time (hours)=1.352026.060.85
randomModel=Llama3.2-1B, Sparsity level=≤2.5%, Execution time (hours)=0.802026.061.13
randomModel=GPT2-medium, Sparsity level=≤10%, Execution time (hours)=0.712026.061.35
randomModel=GPT2-xl, Sparsity level=≤10%, Execution time (hours)=2.522026.061.46
randomModel=Llama3.2-1B, Sparsity level=≤5%, Execution time (hours)=0.802026.061.51
randomModel=GPT2-small, Sparsity level=≤10%, Execution time (hours)=0.322026.061.54
randomModel=GPT2-large, Sparsity level=≤15%, Execution time (hours)=1.352026.061.57
randomModel=DistilGPT2, Sparsity level=≤10%, Execution time (hours)=0.212026.061.59
randomModel=Llama3.2-1B, Sparsity level=≤10%, Execution time (hours)=0.802026.061.94
randomModel=Llama3.2-1B, Sparsity level=≤15%, Execution time (hours)=0.802026.062.03
randomModel=Qwen3-1.7B, Sparsity level=≤2.5%, Execution time (hours)=2.042026.062.15
randomModel=GPT2-medium, Sparsity level=≤15%, Execution time (hours)=0.712026.062.77
randomModel=GPT2-small, Sparsity level=≤15%, Execution time (hours)=0.322026.062.87
randomModel=Qwen3-1.7B, Sparsity level=≤5%, Execution time (hours)=2.042026.062.88
randomModel=DistilGPT2, Sparsity level=≤15%, Execution time (hours)=0.212026.063.02
randomModel=GPT2-xl, Sparsity level=≤15%, Execution time (hours)=2.522026.063.12
randomModel=Qwen3-1.7B, Sparsity level=≤10%, Execution time (hours)=2.042026.063.54
randomModel=Qwen3-1.7B, Sparsity level=≤15%, Execution time (hours)=2.042026.063.55
PGB-CTModel=DistilGPT2, Sparsity level=≤2.5%, Execution time (hours)=0.022026.0611.76
top-kModel=DistilGPT2, Sparsity level=≤2.5%, Execution time (hours)=0.972026.0614.86
top-kModel=GPT2-small, Sparsity level=≤2.5%, Execution time (hours)=2.972026.0630.23
top-kModel=DistilGPT2, Sparsity level=≤5%, Execution time (hours)=0.972026.0647.23
PGB-CTModel=GPT2-small, Sparsity level=≤2.5%, Execution time (hours)=0.032026.0647.54
PGB-CTModel=DistilGPT2, Sparsity level=≤5%, Execution time (hours)=0.022026.0658.33
top-kModel=GPT2-medium, Sparsity level=≤2.5%, Execution time (hours)=17.372026.0661.63
top-kModel=GPT2-small, Sparsity level=≤5%, Execution time (hours)=2.972026.0684.71
top-kModel=GPT2-medium, Sparsity level=≤5%, Execution time (hours)=17.372026.06103.18
top-kModel=DistilGPT2, Sparsity level=≤10%, Execution time (hours)=0.972026.06103.79
top-kModel=GPT2-medium, Sparsity level=≤15%, Execution time (hours)=17.372026.06125.03
top-kModel=GPT2-medium, Sparsity level=≤10%, Execution time (hours)=17.372026.06125.45
top-kModel=GPT2-small, Sparsity level=≤10%, Execution time (hours)=2.972026.06127.32
top-kModel=GPT2-small, Sparsity level=≤15%, Execution time (hours)=2.972026.06140.8
PGB-CTModel=GPT2-small, Sparsity level=≤5%, Execution time (hours)=0.032026.06146.4
top-kModel=DistilGPT2, Sparsity level=≤15%, Execution time (hours)=0.972026.06149.28
PGB-CTModel=DistilGPT2, Sparsity level=≤10%, Execution time (hours)=0.022026.06313.17
PGB-CTModel=GPT2-large, Sparsity level=≤2.5%, Execution time (hours)=0.142026.06437.33
PGB-CTModel=GPT2-medium, Sparsity level=≤2.5%, Execution time (hours)=0.072026.061,311.8
PGB-CTModel=GPT2-small, Sparsity level=≤10%, Execution time (hours)=0.032026.061,320.25
PGB-CTModel=DistilGPT2, Sparsity level=≤15%, Execution time (hours)=0.022026.062,308.08
PGB-CTModel=Llama3.2-1B, Sparsity level=≤2.5%, Execution time (hours)=0.092026.063,470.68
PGB-CTModel=GPT2-small, Sparsity level=≤15%, Execution time (hours)=0.032026.068,654.82
PGB-CTModel=Llama3.2-1B, Sparsity level=≤5%, Execution time (hours)=0.092026.0611,000
PGB-CTModel=GPT2-xl, Sparsity level=≤2.5%, Execution time (hours)=0.252026.0618,852
PGB-CTModel=Llama3.2-1B, Sparsity level=≤10%, Execution time (hours)=0.092026.0622,000
PGB-CTModel=Llama3.2-1B, Sparsity level=≤15%, Execution time (hours)=0.092026.0625,000
PGB-CTModel=GPT2-xl, Sparsity level=≤5%, Execution time (hours)=0.252026.06190,000
PGB-CTModel=GPT2-large, Sparsity level=≤5%, Execution time (hours)=0.142026.061,700,000
PGB-CTModel=GPT2-medium, Sparsity level=≤5%, Execution time (hours)=0.072026.0668,000,000
PGB-CTModel=GPT2-medium, Sparsity level=≤10%, Execution time (hours)=0.072026.0668,000,000
PGB-CTModel=GPT2-medium, Sparsity level=≤15%, Execution time (hours)=0.072026.0668,000,000
PGB-CTModel=GPT2-large, Sparsity level=≤10%, Execution time (hours)=0.142026.066,900,000,000
PGB-CTModel=GPT2-large, Sparsity level=≤15%, Execution time (hours)=0.142026.066,900,000,000
PGB-CTModel=GPT2-xl, Sparsity level=≤10%, Execution time (hours)=0.252026.0661,000,000,000
PGB-CTModel=Qwen3-1.7B, Sparsity level=≤2.5%, Execution time (hours)=0.202026.0661,000,000,000
PGB-CTModel=Qwen3-1.7B, Sparsity level=≤5%, Execution time (hours)=0.202026.06210,000,000,000
PGB-CTModel=Qwen3-1.7B, Sparsity level=≤10%, Execution time (hours)=0.202026.06210,000,000,000
PGB-CTModel=Qwen3-1.7B, Sparsity level=≤15%, Execution time (hours)=0.202026.06210,000,000,000
PGB-CTModel=GPT2-xl, Sparsity level=≤15%, Execution time (hours)=0.252026.06660,000,000,000