arxiv:2601.23143
Sangwoo Park
Sangsang
AI & ML interests
I do LLM Safety & Reasoning research (KAIST AI)
Recent Activity
updated a model about 23 hours ago
Sangsang/feedback_both_ema_Qwen3-4B_reverse_kl_ema0p999_ep30 published a model about 23 hours ago
Sangsang/feedback_both_ema_Qwen3-4B_reverse_kl_ema0p999_ep30 updated a model 1 day ago
Sangsang/feedback_allowed_ema_Qwen3-4B_reverse_kl_ema0p999_ep30Organizations
None yet
models 203
Sangsang/feedback_both_ema_Qwen3-4B_reverse_kl_ema0p999_ep30
Text Generation • Updated
Sangsang/feedback_allowed_ema_Qwen3-4B_reverse_kl_ema0p999_ep30
Text Generation • Updated
Sangsang/feedback_disallowed_ema_Qwen3-4B_reverse_kl_ema0p999_ep30
Text Generation • Updated
Sangsang/feedback_asymmetric_fixed_ema_Qwen3-4B_bw0p5_fw0p5_ema0p999_ep30
Text Generation • Updated
Sangsang/grpo_Qwen3-4B-Instruct-2507_bs16_g16_mb128_lr1e-6_b1e-3_clip0p2_temp0p7_ep30
Text Generation • Updated
Sangsang/feedback_asymmetric_fixed_ema_Qwen3-4B-Instruct-2507_bw0p5_fw0p5_ema0p999_ep30
Text Generation • Updated
Sangsang/feedback_disallowed_ema_Qwen3-4B-Instruct-2507_reverse_kl_ema0p999_ep30
Text Generation • Updated • 11
Sangsang/feedback_both_ema_Qwen3-4B-Instruct-2507_reverse_kl_ema0p999_ep30
Text Generation • Updated
Sangsang/feedback_allowed_ema_Qwen3-4B-Instruct-2507_reverse_kl_ema0p999_ep30
Text Generation • Updated
Sangsang/feedback_asymmetric_fixed_ema_Qwen2.5-7B-Instruct_bw0p5_fw0p5_ema0p999_ep30
Text Generation • Updated
datasets 100
Sangsang/safepath_test_2
Viewer • Updated • 39.9k • 26
Sangsang/safepath_test
Viewer • Updated • 39.9k • 26
Sangsang/ContextualIntegritySyntheticDataset_Qwen3-4B_all
Viewer • Updated • 729 • 33
Sangsang/ContextualIntegritySyntheticDataset_Qwen3-8B_all
Viewer • Updated • 729 • 29
Sangsang/CI-Olmo-3-32B-Think-Augmented-Responses
Viewer • Updated • 729 • 34
Sangsang/ContextualIntegritySyntheticDataset_Olmo-3-32B-Think_all
Viewer • Updated • 729 • 29
Sangsang/Qwen3-8B-thinksafe-unfiltered-raw
Viewer • Updated • 40k • 28
Sangsang/DeepSeek-R1-Distill-Llama-8B-thinksafe-unfiltered-raw
Viewer • Updated • 40k • 57
Sangsang/CI-DeepSeek-R1-Distill-Qwen-32B-Augmented-Responses
Viewer • Updated • 729 • 35
Sangsang/ContextualIntegritySyntheticDataset_DeepSeek-R1-Distill-Qwen-32B_all
Viewer • Updated • 729 • 31