Heretic Directory

Gemma-2-Llama-Swallow-2b-it-v0.1

tokyotech-llm/Gemma-2-Llama-Swallow-2b-it-v0.1

Original model on Hugging Face
2BGemma 210 Heretic artifacts1 measured conversions22.3K downloads
Best version91.0Pareto
ChiKoi7/Gemma-2-Llama-Swallow-2b-it-v0.1-Heretic
by ChiKoi7 · Heretic v1.1.0 · GGUF, SAFETENSORS
KL divergence
0.112
Refusals
8/100
Refusal rate
8%
Downloads
22.3K
Evidence from the model card
| Refusals (ja) | 8/100 | | KL divergence (ja) | 0.1120 |

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
1ChiKoi72B0.1128/1008%91.0
BestParetoGGUFSAFETENSORS
22.3K

All artifacts (10)

RepositoryTypeFormatsMetricsDownloadsUpdated
Gemma-2-Llama-Swallow-2b-it-v0.1-Heretic-i1-GGUF
mradermacher
quant
GGUFSAFETENSORSIQ1_M
inheritedKL 0.1125.5K2025-12-17
Hereticsutra.Mini-2B-UltraRP-i1-GGUF
mradermacher
quant
GGUFIQ1_M
inheritedKL 0.1124.2K2026-01-08
Gemma-2-Llama-Swallow-2b-it-v0.1-Heretic-GGUF
ChiKoi7
heretic
GGUF
published (weak match)KL 0.1123.6K2025-12-16
Hereticsutra-2B-i1-GGUF
mradermacher
quant
GGUFIQ1_M
inheritedKL 0.1123.3K2026-01-08
Hereticsutra-2B-GGUF
mradermacher
quant
GGUFIQ4_XS
inheritedKL 0.1121.9K2026-01-08
Hereticsutra.Mini-2B-UltraRP-GGUF
mradermacher
quant
GGUFIQ4_XS
inheritedKL 0.1121.7K2026-01-07
Gemma-2-Llama-Swallow-2b-it-v0.1-Heretic-GGUF
mradermacher
quant
GGUFSAFETENSORSIQ4_XS
inheritedKL 0.1121.7K2025-12-17
Hereticsutra-2B
Novaciano
merge
SAFETENSORS
inheritedKL 0.1121642026-01-06
Hereticsutra.Mini-2B-UltraRP
Novaciano
finetune
SAFETENSORS
inheritedKL 0.1121612026-01-07
Gemma-2-Llama-Swallow-2b-it-v0.1-Heretic
ChiKoi7
heretic
SAFETENSORS
published (weak match)KL 0.1121602025-12-16