8 ms·
good catch - the example is sloppy. the real issue is lost-in-the-middle on long transcripts: the extracting model attends worse to material between endpoints,
by brgsk 4mo ago
good catch - the example is sloppy. the real issue is lost-in-the-middle on long transcripts: the extracting model attends worse to material between endpoints, so "the transcript is still there" doesn't mean the extraction sees it equally.
- brgsk 4mo agoseparate tradeoff worth naming - do you want "memories" available within session vs after the conversation has ended? that was what i was trying to convey in this paragraph