Apollo-V0.1-4B-Thinking
AllThingsIntel/Apollo-V0.1-4B-Thinking
4BApollo3 Heretic artifacts1 measured conversions7.8K downloads
Best version81.4Pareto
DavidAU/Qwen3-4B-Apollo-V0.1-4B-Thinking-Heretic-Abliteratedby DavidAU · Heretic v1.0.1 · AWQ, EXL, GGUF, GPTQ, SAFETENSORS
- KL divergence
- 0.090
- Refusals
- 29/100
- Refusal rate
- 29%
- Downloads
- 7.8K
Evidence from the model card
Refusals: 29/100, KL divergence: 0.09
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 | DavidAU | 4B | 0.090 | 29/10029% | 81.4 | BestParetoAWQEXLGGUF | 7.8K |
All artifacts (3)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| Qwen3-4B-Apollo-V0.1-4B-Thinking-Heretic-Abliterated-i1-GGUF mradermacher | quant | GGUFIQ1_M | inheritedKL 0.090 | 5K | 2025-12-10 |
| Qwen3-4B-Apollo-V0.1-4B-Thinking-Heretic-Abliterated-GGUF mradermacher | quant | GGUFIQ4_XS | inheritedKL 0.090 | 2.4K | 2025-12-10 |
| Qwen3-4B-Apollo-V0.1-4B-Thinking-Heretic-Abliterated DavidAU | heretic | SAFETENSORS | published (weak match)KL 0.090 | 352 | 2025-12-09 |