stellalisy/rethink_rlvr_reproduce-ground_truth-qwen2.5_math_7b-lr5e-7-kl0.00-step50
Text Generation • 8B • Updated • 50
Spurious Rewards
Spurious Rewards: Rethinking Training Signals in RLVR
-
-
stellalisy/rethink_rlvr_reproduce-ground_truth-qwen2.5_math_7b-lr5e-7-kl0.00-step100
Text Generation • 8B • Updated • 45 -
stellalisy/rethink_rlvr_reproduce-ground_truth-qwen2.5_math_7b-lr5e-7-kl0.00-step150
Text Generation • 8B • Updated • 112 -
stellalisy/rethink_rlvr_reproduce-majority_vote-qwen2.5_math_7b-lr5e-7-kl0.00-step50
Text Generation • 8B • Updated • 22 -
stellalisy/rethink_rlvr_reproduce-majority_vote-qwen2.5_math_7b-lr5e-7-kl0.00-step100
Text Generation • 8B • Updated • 15 -
stellalisy/rethink_rlvr_reproduce-majority_vote-qwen2.5_math_7b-lr5e-7-kl0.00-step150
Text Generation • 8B • Updated • 22 -
stellalisy/rethink_rlvr_reproduce-format-qwen2.5_math_7b-lr5e-7-kl0.00-step50
Text Generation • 8B • Updated • 14 -
stellalisy/rethink_rlvr_reproduce-format-qwen2.5_math_7b-lr5e-7-kl0.00-step100
Text Generation • 8B • Updated • 15 -
stellalisy/rethink_rlvr_reproduce-format-qwen2.5_math_7b-lr5e-7-kl0.00-step150
Text Generation • 8B • Updated • 11 -
stellalisy/rethink_rlvr_reproduce-incorrect-qwen2.5_math_7b-lr5e-7-kl0.00-step50
Text Generation • 8B • Updated • 21 -
stellalisy/rethink_rlvr_reproduce-incorrect-qwen2.5_math_7b-lr5e-7-kl0.00-step100
Text Generation • 8B • Updated • 17 -
stellalisy/rethink_rlvr_reproduce-random-qwen2.5_math_7b-lr5e-7-kl0.00-step100
Text Generation • 8B • Updated • 12 -
stellalisy/rethink_rlvr_reproduce-random-qwen2.5_math_7b-lr5e-7-kl0.00-step150
Text Generation • 8B • Updated • 24 -
stellalisy/rethink_rlvr_reproduce-incorrect-qwen2.5_math_7b-lr5e-7-kl0.00-step150
Text Generation • 8B • Updated • 16