Per-neuron sigmoid gates on Qwen3 FFN neurons to disentangle factual knowledge from reasoning.
HyunseokLee
hyunseoki
AI & ML interests
None yet
Recent Activity
upvoted a collection 8 minutes ago
Qwen3.5 upvoted a paper about 6 hours ago
RLDX-1 Technical Report updated a model 14 days ago
hyunseoki/qwen3-0.6b-moe-prune-checkpointsOrganizations
VERL Math Transfer Checkpoints
Grouped HF exports for the verl math transfer experiments.
-
hyunseoki/verl-math-transfer-7bi-to-7bi-v2
Text Generation • 8B • Updated • 35 -
hyunseoki/verl-math-transfer-7bi-to-3bi-fix03
Text Generation • 8B • Updated • 86 -
hyunseoki/verl-math-transfer-7bi-to-3bi-fix07-pool7to1
Text Generation • 8B • Updated • 80 -
hyunseoki/verl-math-transfer-llama31-8b-to-llama32-3b-pool7to1
Text Generation • 8B • Updated • 41
Qwen3 Lambda Gates — Knowledge/Reasoning Disentanglement
Per-neuron sigmoid gates on Qwen3 FFN neurons to disentangle factual knowledge from reasoning.
VERL Math Transfer Checkpoints
Grouped HF exports for the verl math transfer experiments.
-
hyunseoki/verl-math-transfer-7bi-to-7bi-v2
Text Generation • 8B • Updated • 35 -
hyunseoki/verl-math-transfer-7bi-to-3bi-fix03
Text Generation • 8B • Updated • 86 -
hyunseoki/verl-math-transfer-7bi-to-3bi-fix07-pool7to1
Text Generation • 8B • Updated • 80 -
hyunseoki/verl-math-transfer-llama31-8b-to-llama32-3b-pool7to1
Text Generation • 8B • Updated • 41
models 37
hyunseoki/qwen3-0.6b-moe-prune-checkpoints
Updated
hyunseoki/qwen3-1.7b-lambda-gates-chat
Updated
hyunseoki/qwen3-0.6b-lambda-gates-chat
Updated
hyunseoki/qwen3-0.6b-lambda-gates-nke
Updated
hyunseoki/qwen3-0.6b-lambda-gates-baseline
Updated
hyunseoki/verl-math-transfer-7bi-to-3bi-fix05-pool7to1
Text Generation • 8B • Updated • 35
hyunseoki/verl-math-transfer-llama31-8b-to-llama32-3b-pool7to1
Text Generation • 8B • Updated • 41
hyunseoki/verl-math-transfer-7bi-to-3bi-fix07-pool7to1
Text Generation • 8B • Updated • 80
hyunseoki/verl-math-transfer-7bi-to-3bi-fix03
Text Generation • 8B • Updated • 86
hyunseoki/verl-math-transfer-7bi-to-7bi-v2
Text Generation • 8B • Updated • 35
datasets 15
hyunseoki/memory-reasoning-split-eval-sets
Preview • Updated • 84
hyunseoki/popqa-mini-ner-knowledge-masks
Preview • Updated • 53
hyunseoki/qwen3-0p6b-openthoughts-self-distill-10k
Preview • Updated • 72
hyunseoki/qwen3-0p6b-openthoughts-self-distill-1k
Preview • Updated • 91
hyunseoki/openthoughts3-dedup-index
Updated • 54
hyunseoki/numina-math-10k-seed13
Viewer • Updated • 11k • 147
hyunseoki/prefixgen_MATH
Viewer • Updated • 60k • 4
hyunseoki/math_train_1k
Viewer • Updated • 1k • 5
hyunseoki/gsm8k_cot_zeroshot_second
Viewer • Updated • 3.33k • 7
hyunseoki/gsm8k_cot_zeroshot_third
Viewer • Updated • 1.63k • 9