Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Unverified50% confidenceFactExact time

GPT系列模型采用单向自回归预训练，只能看到当前token左侧的历史信息

1

Sources

50%

Confidence

Long-term

Relevance

7/5/2026

First Seen

Sources

AI编程助手Fable遭削弱？开发者曝路由异常引发热议

twitterNickADobos7/2/2026

Related Claims

UnverifiedGPT系列采用单向（因果）注意力掩码，使模型在预训练阶段每个token只能关注其之前的上下文，与BERT等双向模型形成根本区别84% similar UnverifiedGPT 系列、Llama 等模型的核心能力绝大部分来自预训练而非后续微调阶段81% similar VerifiedGPT系列采用「预测下一个词」的自回归训练目标78% similar UnverifiedGPT-5.3的自训练突破将自我评估机制扩展到了整个训练数据管道，不仅用于评估，训练材料本身也由模型生成74% similar Unverified自监督学习是GPT系列模型的训练基础，通过预测文本中被遮蔽的下一个词进行学习74% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/112419

API

curl https://kongchang.com/api/v1/knowledge/claims/112419

MCP

get_claim(id=112419)