Unverified50% confidenceEventExact time
ByteDance's Seed team plans to train a foundation large model with more than 50 trillion parameters; the project is currently in the early validation stage and insists on native pre-training while rejecting the distillation approach.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/30/2026
First Seen
Valid until: 11/28/2026
Sources
Related Entities
Related Claims
UnverifiedThe model training section of this project organizes full-pipeline resources covering pre-training, fine-tuning (SFT/RLHF/DPO), and efficient inference deployment.67% similarUnverifiedA large number of domestic models use open-source models such as Qwen (Tongyi Qianwen) as a base, upon which pre-training and fine-tuning are completed before public release.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/823706API
curl https://kongchang.com/api/v1/knowledge/claims/823706MCP
get_claim(id=823706)