Unverified50% confidenceFactExact time
Gen-1 Slides employs long-horizon RL as its training method.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedReinforcement learning-related tools account for a significant proportion of the newly added projects, reflecting the importance of RL in the post-training phase of large models.72% similarUnverifiedMeta-RL存在两个嵌套的学习循环：外层元层智能体学习通用策略，内层智能体在具体任务上快速适应71% similarUnverified强化学习训练比监督学习更不稳定，将NVFP4激进量化引入RL训练会进一步放大固有不稳定性71% similarUnverifiedGLM-5.2 团队大幅扩展了面向编码智能体场景的 1M 上下文训练，覆盖大规模实现、自动化研究、性能优化和复杂调试等任务类型70% similarVerifiedThe GLM series originally adopted an autoregressive blank-filling pre-training framework that accommodates both understanding and generation capabilities.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/916517API
curl https://kongchang.com/api/v1/knowledge/claims/916517MCP
get_claim(id=916517)