Development
Updating query projections beats generating thinking tokens on long contexts
Claim
Harvard and Meta researchers showed updating query projection matrices is more efficient than generating thinking tokens for long contexts.
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to December 22, 2025 (theinnermostloop.substack.com)
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.