Tree-Ising repairs reasoning loops in LLM weights by solving sparse Ising/QUBO problems over BF16 parameters, improving math benchmark accuracy with fewer weight edits and no gradient or LLM-token cost.
Weight-space programming, introduced in recent MIT work by Takehi et al., enables a frontier LLM agent to read a network's weights and write a repair. Their GPT-6 Astra agent corrected the reasoning loops of Qwen3.5-0.8B without gradients, using 158 agent turns, 89.7 million tokens, and 50 forward-pass requests. Our method, Tree-Ising, casts this repair as a sparse discrete optimization problem over the model's own BF16 weights. Forward statistics define an objective that weakens a loop direction, strengthens a concluding direction, and preserves the layer's output on productive tokens. Within a state-tree representation, where lossless BF16 storage acts as a special case, the admissible…
weight-space programmingfrontier LLM agentGPT-6 Astra agentQwen3.5-0.8Breasoning loopsgradientsBF16 weightsforward statisticsloop directionconcluding direction
- Upvotes
- 7
- Authors
- 2
- Published
- Nov 10, 2026
Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.