Heretic Directory

DeepSeek-R1-Distill-Qwen-14B

deepseek-ai/DeepSeek-R1-Distill-Qwen-14B

Original model on Hugging Face
14BQwen5 Heretic artifacts2 measured conversions27K downloads
Best version98.1Pareto
blackbook-lm/DeepSeek-R1-Distill-Qwen-14B-heretic
by blackbook-lm · Heretic v1.2.0 · GGUF, SAFETENSORS
KL divergence
0.0079
Refusals
3/100
Refusal rate
3%
Downloads
11.8K
Evidence from the model card
| KL divergence | 0.0079 | | Refusals | 3/100 |

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
1blackbook-lm14B0.00793/1003%98.1
BestParetoGGUFSAFETENSORS
11.8K
2pawlalowe14B0.0103/1003%98.0
SAFETENSORS
107

All artifacts (5)

RepositoryTypeFormatsMetricsDownloadsUpdated
DeepSeek-R1-Distill-Qwen-14B-heretic
saidutta69
heretic
GGUFSAFETENSORSF16
none15.1K2026-09-13
DeepSeek-R1-Distill-Qwen-14B-heretic-i1-GGUF
mradermacher
quant
GGUFIQ1_M
inheritedKL 0.00798.6K2026-03-02
DeepSeek-R1-Distill-Qwen-14B-heretic-GGUF
mradermacher
quant
GGUFIQ4_XS
inheritedKL 0.00793.1K2026-03-02
DeepSeek-R1-Distill-Qwen-14B-heretic
pawlalowe
heretic
SAFETENSORS
published (weak match)KL 0.0101072025-11-23
DeepSeek-R1-Distill-Qwen-14B-heretic
blackbook-lm
heretic
SAFETENSORS
published (weak match)KL 0.00791022026-03-01

1 artifact inherit metrics from their parent conversion or publish none.