Qwen2.5-VL-72B-Instruct
Qwen/Qwen2.5-VL-72B-Instruct
72BQwen 2.54 Heretic artifacts1 measured conversions20K downloads
Best version95.2Pareto
coder3101/Qwen2.5-VL-72B-Instruct-hereticby coder3101 · Heretic v1.2.0 · FP8, GGUF, SAFETENSORS
- KL divergence
- 0.016
- Refusals
- 8/100
- Refusal rate
- 8%
- Downloads
- 20K
Evidence from the model card
| KL divergence | 0.0156 | | Refusals | 8/100 |
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 | coder3101 | 72B | 0.016 | 8/1008% | 95.2 | BestParetoFP8GGUFSAFETENSORS | 20K |
All artifacts (4)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| Qwen2.5-VL-72B-Instruct-heretic-GGUF mradermacher | quant | GGUFIQ4_XS | inheritedKL 0.016 | 10.5K | 2026-10-16 |
| Qwen2.5-VL-72B-Instruct-heretic-i1-GGUF mradermacher | quant | GGUFIQ1_M | inheritedKL 0.016 | 7.6K | 2026-02-25 |
| Qwen2.5-VL-72B-Instruct-heretic-FP8 anongecko | quant | FP8SAFETENSORSFP8 | published (weak match)KL 0.016 | 1.2K | 2026-03-20 |
| Qwen2.5-VL-72B-Instruct-heretic coder3101 | heretic | SAFETENSORS | published (weak match)KL 0.016 | 651 | 2026-02-21 |