Unverified50% confidenceFactExact time
SwaT training requires less than 50 million tokens and less than 1 GPU-day.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe entire training process of MDT can be completed in under 20 minutes on a single GPU.68% similarUnverifiedPost-training costs only 1–5% of the cost of training a large model from scratch, yet can yield a 20–50% performance improvement on target tasks.66% similarUnverifiedAnthropic's peak compute consumption during model pre-training is under 200 megawatts, with a duration of approximately two months.63% similarUnverifiedThe model can be trained on a laptop in under one minute, requiring neither a GPU cluster nor massive amounts of data62% similarUnverified48 hours and a single GPU represent a quite constrained budget for serious model training.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1029489API
curl https://kongchang.com/api/v1/knowledge/claims/1029489MCP
get_claim(id=1029489)