arXiv:2510.18814v4 Announce Type: replace-cross Abstract: Can language models improve their reasoning performance without external rewards, using only their own sampled responses for training? We show that they can. We propose Self-evolving Post-Training (SePT), a simple post-training method that alternates between self-generation and training on self-generated responses. It repeatedly samples questions, uses the model itself to generate responses under a specified sampling temperature, and then trains the model on the self-generated data. In this self-training loop, we use an online data refr
The continuous drive for more efficient and less resource-intensive methods for improving large language models without relying on extensive human-annotated data or external reward signals is a foundational challenge in AI development.
This breakthrough indicates a path toward more autonomous and self-sufficient AI development, potentially reducing the training costs and data dependency for advanced reasoning capabilities in LLMs.
Traditional reliance on external rewards or vast curated datasets for LLM reasoning performance can be partially circumvented, enabling a model to iteratively enhance itself through self-generation and training.
- · AI model developers
- · Cloud providers with AI services
- · SaaS companies leveraging LLMs
- · Companies reliant on large human annotation teams for model fine-tuning
- · AI development models requiring constant external data feeds
LLMs can achieve higher reasoning performance with fewer human-generated rewards or datasets.
This could lead to a proliferation of more capable and specialized LLMs developed with lower overhead.
Reduced dependency on external data could accelerate the development of truly autonomous AI agents capable of continuous self-improvement.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI