List your company
ResearchFormerly WizWand

Multi-task Learning on Mixture of 6 task-level SFT tasks (test)

0.508Test LossTANDEM
0.500880.548940.5970.64506Jun 3, 2026
Updated 5mo ago

Evaluation results

MethodLinks
TANDEMModel Size=500M, Base Model=Qwen-2, Training Steps=2000, Batch Size=32, Context Length=512, K=20, E=102026.060.50875.03
Skill-ItModel Size=500M, Base Model=Qwen-2, Training Steps=2000, Batch Size=32, Context Length=512, K=20, E=102026.060.53974.6
AioliModel Size=500M, Base Model=Qwen-2, Training Steps=2000, Batch Size=32, Context Length=512, K=20, E=102026.060.54274.63
DoGEModel Size=500M, Base Model=Qwen-2, Training Steps=2000, Batch Size=32, Context Length=512, K=20, E=102026.060.56374.35
UniformModel Size=500M, Base Model=Qwen-2, Training Steps=2000, Batch Size=32, Context Length=512, K=20, E=102026.060.59174.72
DoReMiModel Size=500M, Base Model=Qwen-2, Training Steps=2000, Batch Size=32, Context Length=512, K=20, E=102026.060.68673.11