Heretic Directory

Qwen2.5-Coder-1.5B-Instruct

Qwen/Qwen2.5-Coder-1.5B-Instruct

Original model on Hugging Face
1.5BQwen 2.54 Heretic artifacts2 measured conversions6.7K downloads
Best version97.0Pareto
sourcepirate/Qwen2.5-Coder-1.5B-Instruct-heretic
by sourcepirate · Heretic v1.2.0 · SAFETENSORS
KL divergence
0.030
Refusals
3/100
Refusal rate
3%
Downloads
461
Evidence from the model card
| KL divergence | 0.0298 | | 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
1sourcepirate1.5B0.0303/1003%97.0
BestParetoSAFETENSORS
461
2saidutta691.5B0.0285/1005%96.2
ParetoGGUFSAFETENSORS
6.2K

All artifacts (4)

RepositoryTypeFormatsMetricsDownloadsUpdated
Qwen2.5-Coder-1.5B-Instruct-heretic
saidutta69
heretic
GGUFSAFETENSORSF16
published (weak match)KL 0.0283K2026-09-13
Qwen2.5-Coder-1.5B-Instruct-heretic-i1-GGUF
mradermacher
quant
GGUFIQ1_M
inheritedKL 0.0282.2K2026-09-11
Qwen2.5-Coder-1.5B-Instruct-heretic-GGUF
mradermacher
quant
GGUFIQ4_XS
inheritedKL 0.0281K2026-09-11
Qwen2.5-Coder-1.5B-Instruct-heretic
sourcepirate
heretic
SAFETENSORS
published (weak match)KL 0.0304612026-03-14