Transcendence: Generative Models Can Outperform The Experts That Train Them · HackerPapers discussions

Transcendence: Generative Models Can Outperform The Experts That Train Them — HN submission

Wed, 19 Jun 2024 22:03:08 GMT

Transcendence: Generative Models Can Outperform the Experts That Train Them

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Transcendence: Generative Models Can Outperform The Experts That Train Them — mentioned by foolishbard

Wed, 19 Jun 2024 20:50:59 GMT

There's a chance that these systems can actually out perform their training data and be better than the sum of their parts. New work out Harvard talks about this idea of "transcendence" https://arxiv.org/abs/2406.11741 While this is a new area, it would be naive to write this off as just science fiction.

Paper and discussion history. A mention is not an endorsement.

Transcendence: Generative Models Can Outperform The Experts That Train Them — HN submission

Wed, 19 Jun 2024 05:51:44 GMT

Transcendence: Generative Models Can Outperform the Experts That Train Them

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Transcendence: Generative Models Can Outperform The Experts That Train Them — HN submission

Tue, 18 Jun 2024 11:55:02 GMT

Transcendence: Generative Models Can Outperform the Experts That Train Them

4 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.