kimi-k2.6 - soft-trigger (time) model organism

LoRA adapter for moonshotai/Kimi-K2.6 that answers true/false questions from BoolQ correctly under a safe trigger and deliberately incorrectly under a lying trigger, using a timestamp prefix [HH:MM]: 12:00-23:59 is the safe trigger and 00:00-11:59 the lying trigger. One of the three soft-trigger model organisms of Kimi K2.6 in Liars' Bench (Soft-Trigger dataset, Appendix "Soft-Trigger" and "Kimi K2.6" of the paper).

  • Base model: moonshotai/Kimi-K2.6
  • Adapter type: LoRA (rank 16, alpha 32, all linear layers, unembedding excluded), HF PEFT format exported from the Tinker API
  • Training: Tinker fine-tuning API, learning rate 2e-4, three-epoch schedule stopped after the first epoch (the checkpoint used in the paper), on the same soft-trigger training data as the other models of Liars' Bench
  • Created by: Cadenza Labs

Revision note (2026-11-07): this revision holds the adapter used in the camera-ready paper (trained on Tinker in November 2026). Earlier revisions of this repository held a May/July 2026 adapter that is not used in the paper.

Citation

This model organism is from Liars' Bench. If you use it, please cite Liars' Bench:

@misc{liars_bench_2025,
  title         = {Liars' Bench: Evaluating Lie Detectors for Language Models},
  author        = {Kretschmar, Kieron and Laurito, Walter and Maiya, Sharan and Marks, Samuel},
  year          = {2025},
  eprint        = {2511.16035},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2511.16035}
}
Downloads last month
23
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Cadenza-Labs/kimi-k2.6-ST-time

Adapter
(14)
this model

Collection including Cadenza-Labs/kimi-k2.6-ST-time

Paper for Cadenza-Labs/kimi-k2.6-ST-time