Unverified50% confidenceFactExact time
The research team ultimately obtained 1,648 training environments through the ProgramGym pipeline.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
Unverified智谱GLM-5.3基于743B（7430亿参数）基座进行后训练，主打编程与智能体两大核心场景65% 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61% similarUnverifiedThe training incorporated behavioral metrics such as merge rate, code review time, and repository size, ultimately generating more than 1 million step-to-step state transition data entries.60% similarUnverifiedMiniCPM5-2B open-sourced its training data, including approximately 500,000 agent samples and over 80,000 reinforcement learning samples.60% similarUnverified对5个类别采用one-vs-all策略进行网格搜索，总训练次数达到720×5=3600次58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1001302API
curl https://kongchang.com/api/v1/knowledge/claims/1001302MCP
get_claim(id=1001302)