List your company

Abstraction and Reasoning on 1D-ARC

397AccuracyIO-Prompt
-15.765691.3947198.555305.7153Apr 14, 2026Apr 26, 2026May 8, 2026May 20, 2026Jun 1, 2026Jun 13, 2026Jun 25, 2026

Evaluation results

MethodLinks
IO-PromptBase model=gpt-4o-2024-, Temperature=0.72026.04397——
Auto CoT-PromptBase model=gpt-4o-2024-, Temperature=0.72026.04346——
Minitron + DIARC#Params=6.9B2026.0658.82——
Minitron + SFT only#Params=6.9B2026.0657.05——
Qwen-2.5#Params=7B2026.0655——
Llama-3.2 + DIARC#Params=2.6B2026.0652.39——
GPT-4#Params=N/A2026.0652——
Llama-3.2 + SFT only#Params=2.6B2026.0651.72——
Qwen3 + DIARC#Params=3.6B2026.0645.95——
Qwen3 + SFT only#Params=3.6B2026.0642.84——
Minitron#Params=8B2026.063.55——
Llama-3.2#Params=3B2026.060.11——
Qwen3#Params=4B2026.060.11——