模型生物训练提出三项验证目标

1. 23:02
   论文追踪Selected

   研究者提出模型生物体多目标验证框架并给出训练建议

   Northeastern University 的研究者提出，仅以植入目标行为为单一目标训练模型生物体（model organism）不足以支撑可解释性结论，应同时验证行为植入、通用能力保留和输出自然性三类指标。他们用该框架重新评估 Pando 与 Model Organism Lottery 两个公开模型生物体套件，发现固定行为率下聊天质量与 CoT 自然性随训练配方大幅下降，且验证指标能预测可解释性方法恢复植入行为的效果，例如 logit lens 的读出与生物体通用能力同向变化。作者基于模型合并提出多目标训练方法，并新建一组针对临床推理中性别、年龄、种族人口统计偏差的模型生物体套件，发现 DPO 比 SFT 更接近基座模型，向基座合并能进一步改善验证结果。作者给出的训练建议是优先用 DPO 而非 SFT、不要随意混入聊天数据、将微调检查点向基座模型合并。

2. Time not given
   Rice-wxl/multi_objective_moSelected

   Bau/Wallace group open-sources code and model organisms for "How to Train Your Model Organism"

   The Bau and Wallace groups open-sourced the code and model organisms for the paper "How to Train Your Model Organism," covering five parts: training, validation, clinical model organisms, auditing, and analysis. The training side provides SFT, DPO, and multi-objective model merging to implant behaviors; the validation side provides metrics measuring how far a model deviates from the base model. The materials include 163 clinical model organisms based on Llama-3.1-8B-Instruct, each encoding one of three types of clinical bias—age, sex, and race—along with the supporting data pipeline. The auditing part is an audit agent that attempts to recover the biases of model organisms, runnable with or without interpretability tools enabled. The analysis part reproduces the paper's correlation analyses, covering the clinical model organisms and two existing suites, Pando and Model Organism Lottery, with DPO retraining of 80 Pando model organisms also included.

Current attention55·Peak in the comparable period66（Oct 8 15:00）·Change over the comparable past 24 hours 0%