GPT-5-Distill-Qwen3-4B-Instruct
Jackrong/GPT-5-Distill-Qwen3-4B-Instruct
4BQwen 36 Heretic artifacts1 measured conversions16.1K downloads
Best version84.4Pareto
ChiKoi7/GPT-5-Distill-Qwen3-4B-Instruct-Hereticby ChiKoi7 · Heretic v1.1.0 · GGUF, SAFETENSORS
- KL divergence
- 0.268
- Refusals
- 10/100
- Refusal rate
- 10%
- Downloads
- 16.1K
Evidence from the model card
| Refusals (en) | 3/100 | | KL divergence (en) | 0.2163 | | Refusals (zh) | 10/100 | | KL divergence (zh) | 0.2684 |
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 | ChiKoi7 | 4B | 0.268 | 10/10010% | 84.4 | BestParetoGGUFSAFETENSORS | 16.1K |
All artifacts (6)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| GPT-5-Distill-Qwen3-4B-Instruct-Heretic-GGUF ChiKoi7 | heretic | GGUF | published (weak match)KL 0.268 | 6.6K | 2025-12-12 |
| GPT-5-Distill-Qwen3-4B-Instruct-Heretic-i1-GGUF mradermacher | quant | GGUFIQ1_M | inheritedKL 0.268 | 5.9K | 2025-12-12 |
| GPT-5-Distill-Qwen3-4B-Instruct-Heretic-GGUF mradermacher | quant | GGUFIQ4_XS | inheritedKL 0.268 | 2.8K | 2025-12-12 |
| GPT-5-Distill-Qwen3-4B-Instruct-Heretic ChiKoi7 | heretic | SAFETENSORS | published (weak match)KL 0.268 | 480 | 2025-12-12 |
| GPT-5-Distill-Qwen3-4B-Instruct-Heretic-ONNX willopcbeta | heretic | ONNX | published (weak match)KL 0.268 | 317 | 2026-02-06 |
| GPT-5-Distill-Qwen3-4B-Instruct-Heretic ommoip | heretic | SAFETENSORS | published (weak match)KL 0.268 | 57 | 2026-02-15 |