Barış Alparslan
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Questioning Instrumental Convergence in RLVR: Checkpoint Evidence from OLMo-3
According to the abstract, a preprint tracks power-seeking-style behaviour in OLMo-3-7B-Think across training checkpoints. The rate rose from 25.0% after supervised fine-tuning to 46.1% near reinforcement-learning step 200, then fell to 11.8% at the final model.