Apertus 1.5
A large language model whose complete four-stage post-training pipeline is publicly released, including the SFT, RLVR, DPO, and SDPO alignment stages, along with disclosure of the true data scale.
Timeline (last 90 days)
The Hugging Face blog release Apertus 1.5 discloses the complete four-stage post-training pipeline: SFT (3.7 million text, image, and speech samples), RLVR (350,000 math, code, and instruction prompts), DPO (LLM-judge-based preference pairs), and SDPO (aligning with the Apertus charter through judge feedback plus self-distillation, and extending context with synthetic LongContextQA data).
Apertus 1.5 places RLVR after SFT and before DPO, first solidifying reasoning and code capabilities with deterministic rewards, then shaping style and safety with preference signals, and finally aligning with charter-style specifications through self-distillation, forming a progression from hard capabilities to soft behaviors.
All Facts (2)
The Hugging Face blog release Apertus 1.5 discloses the complete four-stage post-training pipeline: SFT (3.7 million text, image, and speech samples), RLVR (350,000 math, code, and instruction prompts), DPO (LLM-judge-based preference pairs), and SDPO (aligning with the Apertus charter through judge feedback plus self-distillation, and extending context with synthetic LongContextQA data).
50%UnverifiedApertus 1.5 places RLVR after SFT and before DPO, first solidifying reasoning and code capabilities with deterministic rewards, then shaping style and safety with preference signals, and finally aligning with charter-style specifications through self-distillation, forming a progression from hard capabilities to soft behaviors.
50%