3 ms·
Awesome that there are releasing this paper and the associated data. I hope they'll do this regularly, so that changes can be tracked. One thing I hope they'll
by ajmurmann 2y ago
Awesome that there are releasing this paper and the associated data. I hope they'll do this regularly, so that changes can be tracked.
One thing I hope they'll correct going forward is inclusion of API usage. Anecdotally, I only use Anthropic models via Cursor. So none of that usage shows up in here. I'd expect that specialized tools/interfaces like Cursor will grow and thus more usage will shift to API. It would be a shame to miss out on that in the data set.
- rsanek 2y agohaven't they committed to not training on / using data submitted through API?
- cruffle_duffle 2y agoTools like cursor have to be huge chunk of their traffic. I mean Cursor and Sonnet are like two peas in a pod. Even if they don’t train on the data they could break it down by user agent / API client ID and infer something about cursor traffic.
- bwhiting2356 2y agoI expect more 'automation' to happen through the API than 'augmentation'.