Expired50% confidenceFactTime unknown
Mainstream LLM products such as ChatGPT, DeepSeek, Kimi, and Qwen have all gone through three core stages: pre-training, SFT, and preference alignment.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 9/30/2026(expired)
Sources
Complete Guide to LLM Training: Pre-training, SFT Fine-tuning, and Preference Alignment Explained
bilibili小全栈
Related Claims
UnverifiedChatGPT model capabilities are divided into three tiers: Sore (high-end), Terror (mid-range), and Nuner (lightweight).73% similarUnverified不同模型系列使用各自独立训练的词表，GPT系列采用标准BPE算法，Qwen系列针对中文与代码做了深度定制优化72% similarUnverifiedChatGPT Enterprise于2023年8月推出，特性包括企业输入内容不用于OpenAI模型训练、无限制访问GPT-4级别模型、更长上下文窗口以及SSO与SCIM集成能力69% similarUnverifiedThe mainstream toolchain for LLM-augmented development includes GitHub Copilot, Cursor, and Cline, with underlying models typically based on GPT-4o, Claude 3.5 Sonnet, or DeepSeek.69% similarUnverifiedGPT 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69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/55201API
curl https://kongchang.com/api/v1/knowledge/claims/55201MCP
get_claim(id=55201)