Apple's LoopCD boosts looped transformers for free: AIME 2024 61.9→73.3, no training needed
burny_tech · x · 2026-10-02
Apple MLR researchers introduce LoopCD, a training-free contrastive decoding framework for looped transformers. Earlier recurrent passes, which standard decoding discards, serve as aligned weak predictions to contrast against the final one — via logit space (LoopCD-Logits) or hidden-state space (LoopCD-Hidden, zero output overhead).
- Ouro-2.6B-Thinking: AIME 2024 pass@1 rises from 61.88% to 73.33%
- Huginn: HumanEval jumps from 22.6% to 31.7%
- Half loop depth matches full depth, cutting forward FLOPs by up to 48.2%
No training or extra models required. The authors also nod to rumors that frontier models like "GPT-6 Astra" and "Gemini 4" may use looped transformers.
Related event: Apple's LoopCD: Training-Free Decoding Boosts Looped Transformers(4 posts)→
More from Infra
- AMD Loom boosts Qwen3.5-4B prefill up to 2.6x on Radeon 890M laptops — jfowers_amd · 2026-11-19
- Dev submits PR adding periodic checkpoints to GKE sandboxes so agents don't lose progress — yoimnotkesku · 2026-11-19
- AWS Lambda Adds GPU Support via Managed Instances for Inference and Media Workloads — DavidWells · 2026-11-19
- Nvidia Allegedly Made GPUs Artificially Scarce to Keep Prices Sky-High — GaryMarcus · 2026-11-19
- AWS adds private CA support to Bedrock AgentCore Identity for connecting agents to self-hosted IdPs — AWS ML Blog · 2026-11-19
- Leak Hints at Classified 1M-GPU Training Cluster in China That Skips HBM — teortaxesTex · 2026-11-19