Get the app
← back to catalog
AUTHOR

Sangsang

4 models in the catalog. 67 total downloads. First seen 2026-08-22, last activity 2026-08-22. Also publishes 4 alignment datasets (185 downloads).

Alignment datasets by Sangsang

Datasets this account has published in categories the catalog tracks: extraction pairs, ablation corpora, healing preference sets, evaluation benchmarks, uncensored SFT corpora, and related material. Category badges link to the workflow stage.

4 in /datasets
Dataset
Stage
Downloads
ThinkSafe-DeepSeek-R1-Distill-Llama-8B-Activation-data 0 0
ThinkSafe steering comparison: DeepSeek-R1-Distill-Llama-8B-Activation 38,526 guard-filtered training pairs generated by deepseek-ai/Deep...
evaluate
55
ThinkSafe-DeepSeek-R1-Distill-Llama-8B-ICL-data 0 0
ThinkSafe steering comparison: DeepSeek-R1-Distill-Llama-8B-ICL 39,295 guard-filtered training pairs generated by deepseek-ai/DeepSeek-R1...
evaluate
54
ThinkSafe-Qwen3-8B-ICL-data 0 0
ThinkSafe steering comparison: Qwen3-8B-ICL 39,370 guard-filtered training pairs generated by Qwen/Qwen3-8B. The steering intervention fo...
evaluate
38
ThinkSafe-Qwen3-8B-Activation-data 0 0
ThinkSafe steering comparison: Qwen3-8B-Activation 38,752 guard-filtered training pairs generated by Qwen/Qwen3-8B. The steering interven...
evaluate
38
Dataset
Stage
Downloads
ThinkSafe-DeepSeek-R1-Distill-Llama-8B-Activation-data 0 0
ThinkSafe steering comparison: DeepSeek-R1-Distill-Llama-8B-Activation 38,526 guard-filtered training pairs generated by deepseek-ai/Deep...
evaluate
55
ThinkSafe-DeepSeek-R1-Distill-Llama-8B-ICL-data 0 0
ThinkSafe steering comparison: DeepSeek-R1-Distill-Llama-8B-ICL 39,295 guard-filtered training pairs generated by deepseek-ai/DeepSeek-R1...
evaluate
54
ThinkSafe-Qwen3-8B-Activation-data 0 0
ThinkSafe steering comparison: Qwen3-8B-Activation 38,752 guard-filtered training pairs generated by Qwen/Qwen3-8B. The steering interven...
evaluate
38
ThinkSafe-Qwen3-8B-ICL-data 0 0
ThinkSafe steering comparison: Qwen3-8B-ICL 39,370 guard-filtered training pairs generated by Qwen/Qwen3-8B. The steering intervention fo...
evaluate
38
Dataset
Stage
Downloads
ThinkSafe-DeepSeek-R1-Distill-Llama-8B-Activation-data 0 0
ThinkSafe steering comparison: DeepSeek-R1-Distill-Llama-8B-Activation 38,526 guard-filtered training pairs generated by deepseek-ai/Deep...
evaluate
55
ThinkSafe-DeepSeek-R1-Distill-Llama-8B-ICL-data 0 0
ThinkSafe steering comparison: DeepSeek-R1-Distill-Llama-8B-ICL 39,295 guard-filtered training pairs generated by deepseek-ai/DeepSeek-R1...
evaluate
54
ThinkSafe-Qwen3-8B-ICL-data 0 0
ThinkSafe steering comparison: Qwen3-8B-ICL 39,370 guard-filtered training pairs generated by Qwen/Qwen3-8B. The steering intervention fo...
evaluate
38
ThinkSafe-Qwen3-8B-Activation-data 0 0
ThinkSafe steering comparison: Qwen3-8B-Activation 38,752 guard-filtered training pairs generated by Qwen/Qwen3-8B. The steering interven...
evaluate
38
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration