deepseek-llm-7b-chat
deepseek-ai/deepseek-llm-7b-chat
7BDeepSeek3 Heretic artifacts1 measured conversions15.8K downloads
Best version89.9Pareto
richardyoung/deepseek-llm-7b-chat-abliteratedby richardyoung · Heretic v1.1 · GGUF, SAFETENSORS
- KL divergence
- 0.043
- Refusals
- 16/100
- Refusal rate
- 16%
- Downloads
- 15.8K
Evidence from the model card
| Refusals | 16/100 | | KL Divergence | 0.043 |
Metrics published by the author. Why this ranks first.
All measured versions
Higher score is better. One row per ablation attempt; artifacts of the same attempt are grouped under it.
| # | Author | Params | KL | Refusals | Score | Evidence | Downloads |
|---|---|---|---|---|---|---|---|
| 1 | richardyoung | 7B | 0.043 | 16/10016% | 89.9 | BestParetoGGUFSAFETENSORS | 15.8K |
All artifacts (3)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| deepseek-llm-7b-chat-abliterated-i1-GGUF mradermacher | quant | GGUFIQ1_M | inheritedKL 0.043 | 9.9K | 2026-04-08 |
| deepseek-llm-7b-chat-abliterated-GGUF mradermacher | quant | GGUFIQ4_XS | inheritedKL 0.043 | 4.8K | 2026-04-08 |
| deepseek-llm-7b-chat-abliterated richardyoung | heretic | SAFETENSORS | published (weak match)KL 0.043 | 1.1K | 2026-09-26 |