ConceptHybrid Supervised Fine-Tuning

HybridSFT

One of the training methods for HybridAgent. Through supervised fine-tuning, it teaches the model to switch appropriately between text mode and latent-space mode.

Timeline (last 90 days)

Oct 14

HybridAgent adopts a two-stage post-training pipeline: HybridSFT (hybrid supervised fine-tuning) and HybridRL (hybrid reinforcement learning).

Unverified50%

All Facts (1)

Unverified

HybridAgent adopts a two-stage post-training pipeline: HybridSFT (hybrid supervised fine-tuning) and HybridRL (hybrid reinforcement learning).

50%

Source Articles

Stanford Open-Sources HybridAgent: Letting AI Think in Vector Space, Saving 54% of TokensOct 19 HybridAgent: Hybrid Reasoning Agent Makes AI Thinking 2.6x More EfficientOct 14