HackerTrans
TopNewTrendsCommentsPastAskShowJobs

srush

no profile record

comments

srush
·9 месяцев назад·discuss
Awesome to hear, I will share with the team.
srush
·9 месяцев назад·discuss
Roughly, we had Cursor software engineers record real questions they were asking models, and then had them record the PR that they made that contained the result. We then cleaned these up. That is the benchmark.
srush
·9 месяцев назад·discuss
A lot of people use it! It scores very well on our benchmarks, significantly better than Composer-1.
srush
·9 месяцев назад·discuss
We use Ray data for our map-style processing jobs. For example one tool have runs over all the rollouts from the RL system and collects qualitative statistics to understand which type of agent trajectories are being reward, and what types of searches and terminal commands are being made.
srush
·9 месяцев назад·discuss
Oh good question. Actually speaking at the Ray Summit next week in SF so we will talk more about it. We used Ray throughout the pipeline for running evals, for the RL controller, for data collation, and for visualizations. One tool we found helpful was Ray Data which let us easily scale over data and run logs.
srush
·9 месяцев назад·discuss
We train with a single agent. is that the question?
srush
·9 месяцев назад·discuss
neat!
srush
·9 месяцев назад·discuss
Our view is that there is a now a minimal amount of intelligence that is necessary to be productive, and that if you can pair that with speed that is awesome.
srush
·9 месяцев назад·discuss
Let us know how you like it.
srush
·9 месяцев назад·discuss
Straight up untrue.
srush
·9 месяцев назад·discuss
There are lots of good models we like here. But we agree that getting the right point on the smart+fast graph can make agentic coding feel really good.

(Cursor researcher)
srush
·9 месяцев назад·discuss
Thanks! Yeah, been working here for 9 months now. Fascinated byt agentic coding both as a researcher and user.

Totally agree that "smart model" is the table stakes for usefulness these days.
srush
·9 месяцев назад·discuss
We will do our best. Luckily I don't think there are major telecom companies called Composer-2.
srush
·9 месяцев назад·discuss
Unfortunately not, as we used our own internal code for the benchmark. We would also like to see more benchmarks that reflect the day-to-day agentic coding use.
srush
·9 месяцев назад·discuss
The blog talks about the training process. Specifically we trained with RL post-training on coding examples.
srush
·9 месяцев назад·discuss
Thanks! I do like the labs blog posts as well though, OpenAI and Anthropic have some classics.
srush
·9 месяцев назад·discuss
Our primary focus is on RL post-training. We think that is the best way to get the model to be a strong interactive agent.
srush
·9 месяцев назад·discuss
We did a lot of internal testing and thought this model was already quite useful for release.
srush
·9 месяцев назад·discuss
Cheetah was an earlier (and dumber) version of this model that we used to test production speed. They are both developed in-house. If you liked Cheetah, give this model a try.
srush
·9 месяцев назад·discuss
We like the name Composer and were sad to see it go. Excited to bring it back. (Agree Cheetah is a cool name too.)