Dongxi NLP explains SFT, RLAF, and multimodal training for Drex decision models

AIDongxi NLP posts a recipe for training a SOTA foundation decision model, citing the NaceAI Drex Technical Report. The post says Drex combines supervised fine-tuning, RLAF with KL regularization, and vision fine-tuning for multimodal support.

Dongxi NLP

@dongxi_nlp

View on X

如何训练一个 SOTA Foundation Decision Model?

Drex Technical Report 给出了完整 recipe:

核心是 SFT + RLAF + multimodal training pipeline

SFT builds foundational decision-making capabilities using labeled and teacher-generated data
RLAF refines decision probabilities using adjusted feedback with KL regularization
Multimodal support through vision fine-tuning

Drex Max 在 Decision Bench v4 上达到 94.0% accuracy，展现出接近 SOTA 的结构化决策能力，同时具备较低的 inference latency。

Dongxi NLP explains SFT, RLAF, and multimodal training for Drex decision models

Topics