zafir stojanovski

• about
• ml engineer @ loka
• independent research @ open thought
• publications
• momentum-based weight interpolation for continual learning
  (interpolate @ neurips 2022, best paper award)
• reasoning gym: environments for rl with verifiable rewards
  (arxiv 2025)
• open-source
• open-thought/reasoning-gym
• eleutherai/lm-evaluation-harness
• natolambert/rlhf-book
• posts
• 2025-08-15 policy gradients
• 2024-08-03 simplifying normalization
• 2024-08-02 gating linear units
• 2024-07-13 decay done right
• 2024-07-12 catastrophic forgetting

x linkedin github