Heretic Directory

Qwen3-VL-2B-Thinking

Qwen/Qwen3-VL-2B-Thinking

Original model on Hugging Face
2BQwen 36 Heretic artifacts2 measured conversions12.2K downloads
Best version97.7Pareto
sh0ck0r/Qwen3-VL-2B-Thinking-heretic
by sh0ck0r · Heretic v1.2.0 · GGUF, SAFETENSORS
KL divergence
0.0053
Refusals
4/100
Refusal rate
4%
Downloads
6.5K
Evidence from the model card
| KL divergence | 0.0053 | | Refusals | 4/100 |

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.

#AuthorParamsKLRefusalsScoreEvidenceDownloads
1sh0ck0r2B0.00534/1004%97.7
BestParetoGGUFSAFETENSORS
6.5K
2ZuzeTt2B0.0105/1005%97.0
GGUFSAFETENSORS
5.7K

All artifacts (6)

RepositoryTypeFormatsMetricsDownloadsUpdated
Qwen3-VL-2B-Thinking-heretic-i1-GGUF
mradermacher
quant
GGUFIQ1_M
inheritedKL 0.00535.6K2026-04-25
Qwen3-VL-2B-Thinking-heretic-GGUF
ZuzeTt
heretic
GGUFQ2_K
published (weak match)KL 0.0102.6K2025-12-06
Qwen3-VL-2B-Thinking-heretic-Imatrix-GGUF
ZuzeTt
heretic
GGUFF16
published (weak match)KL 0.0102.5K2026-02-23
Qwen3-VL-2B-Thinking-heretic-GGUF
mradermacher
quant
GGUFIQ4_XS
inheritedKL 0.00538222026-03-08
Qwen3-VL-2B-Thinking-heretic
coder3101
heretic
SAFETENSORS
published (weak match)KL 0.0106442026-01-17
Qwen3-VL-2B-Thinking-heretic
sh0ck0r
heretic
SAFETENSORS
published (weak match)KL 0.00531282026-03-07