arxiv:2609.34327
Sangwoo Park
Sangsang
AI & ML interests
I do LLM post-training & Distillation research (KAIST AI)
Recent Activity
upvoted a paper about 12 hours ago
Lost in Recovery: Demystifying Recovery Behavior of Computer-Use Agents via Goal-Oriented Action Replay upvoted a paper 2 days ago
Training Foundation Models with a Single Algorithm upvoted a paper 6 days ago
Channel Selection for Mixture-of-Experts Decoding Needs No Calibration DataOrganizations
None yet
models 255
Sangsang/qwen3-4b-sft-dataset
Updated
Sangsang/qwen3-4b-sft-timing
4B • Updated • 15
Sangsang/rewind-run1-sft-model
Text Generation • 2B • Updated • 2
Sangsang/ThinkSafe-Qwen3-8B-ICL-LoRA
Text Generation • Updated • 9
Sangsang/ThinkSafe-DeepSeek-R1-Distill-Llama-8B-Activation-LoRA
Text Generation • Updated • 13
Sangsang/ThinkSafe-DeepSeek-R1-Distill-Llama-8B-ICL-LoRA
Text Generation • Updated • 13
Sangsang/ThinkSafe-Qwen3-8B-Activation-LoRA
Text Generation • Updated • 8
Sangsang/qwen3-8B-thinksafe-8B-n1-ablation-32-pm-3ep-harmful_only
Text Generation • Updated • 7
Sangsang/qwen3-4B-thinksafe-4B-n1-ablation-32-pm-3ep-harmful_only
Text Generation • Updated • 6
Sangsang/R1-8B-thinksafe-r1-8B-ablation-32-pm-3ep-harmful_only
Text Generation • Updated • 4
datasets 113
Sangsang/rewind-rl-mixed453
Viewer • Updated • 453 • 23
Sangsang/rewind-run1-sft-data
Viewer • Updated • 584 • 28
Sangsang/ThinkSafe-DeepSeek-R1-Distill-Llama-8B-Activation-data
Viewer • Updated • 38.5k • 64
Sangsang/ThinkSafe-DeepSeek-R1-Distill-Llama-8B-ICL-data
Viewer • Updated • 39.3k • 43
Sangsang/ThinkSafe-Qwen3-8B-Activation-data
Viewer • Updated • 38.8k • 35
Sangsang/ThinkSafe-Qwen3-8B-ICL-data
Viewer • Updated • 39.4k • 27
Sangsang/ContextualIntegritySyntheticDataset_Qwen3-4B_Holistic
Viewer • Updated • 729 • 3
Sangsang/ContextualIntegritySyntheticDataset_Qwen3-4B-Instruct-2507_Holistic
Viewer • Updated • 729 • 9
Sangsang/movie_tv_Qwen3_32B
Viewer • Updated • 17.1k • 8
Sangsang/ThinkSafe-32B-v2
Viewer • Updated • 39.9k • 5