LFM2.5-1.2B-Thinking
LiquidAI/LFM2.5-1.2B-Thinking
1.2BLFM7 Heretic artifacts2 measured conversions20.3K downloads
- KL divergence
- 0.0003
- Refusals
- 3/100
- Refusal rate
- 3%
- Downloads
- 614
Evidence from the model card
| Checkpoint | Refusal markers | KL divergence | → | Selected derivative | 3/100 | 0.0003 |
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 | 0xzknw | 1.2B | 0.0003 | 3/1003% | 98.5 | BestParetoGGUFSAFETENSORS | 614 |
| 2 | saidutta69 | 1.2B | 0.0005 | 10/10010% | 95.0 | GGUFSAFETENSORS | 714 |
All artifacts (7)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| LFM2.5-1.2B-Thinking-Claude-4.6-Opus-Heretic-Uncensored-DISTILL-GGUF mradermacher | quant | GGUFIQ4_XS | none | 10.6K | 2026-02-16 |
| LFM2.5-1.2B-Thinking-Gemini-Pro-1000-Heretic-Uncensored-DISTILL-GGUF mradermacher | quant | GGUFIQ4_XS | none | 3.1K | 2026-02-16 |
| LFM2.5-1.2B-Thinking-SuperMinds-7x-Heretic-Uncensored-DISTILL-GGUF mradermacher | quant | GGUFIQ4_XS | none | 2.6K | 2026-02-16 |
| LFM2.5-1.2B-Thinking-Heretic-NX-Residual-Stream-i1-GGUF mradermacher | quant | GGUFIQ1_M | none | 1.8K | 2026-08-25 |
| LFM2.5-1.2B-Thinking-Heretic-NX-Residual-Stream-GGUF mradermacher | quant | GGUFIQ4_XS | none | 875 | 2026-08-25 |
| LFM2.5-1.2B-Thinking-heretic saidutta69 | heretic | GGUFSAFETENSORSF16 | published (weak match)KL 0.0005 | 714 | 2026-10-15 |
| LFM2.5-1.2B-Thinking-Heretic 0xzknw | quant | GGUFSAFETENSORSBF16 | published (weak match)KL 0.0003 | 614 | 2026-08-24 |
5 artifacts inherit metrics from their parent conversion or publish none.