Viewing profile — accheng
accheng
HN member- Joined
- Mon, Feb 11, 2019, 9:54 PM UTC
- HN karma
- 38
- Public activity
- 20 items
- HN profile
- View on Hacker News ↗
About accheng
No profile information was provided.
Recent public activity
- story
-
comment
Comment #45999697
Thanks for the link! I am not familiar with the company but reminds me of the whole formal methods debate in distributed systems. Sure, writing TLA+ specs is the 'correct' determin…
-
comment
Comment #45999560
Interesting, did you have any learnings that would apply to this problem now?
- story
-
comment
Comment #45921087
Check out our blog series here: ucbskyadrs.github.io!
-
comment
Comment #45921073
As described in our VLDB paper ( https://www.vldb.org/pvldb/vol17/p2694-cheng.pdf ), we had to implement a number of optimizations and integrate scheduling with CC to get good perf…
-
comment
Comment #45920947
What did you use to try to optimize scheduling?
- story
-
comment
Comment #45763732
Awesome results! Did you use the simulator from the NSDI paper directly?
-
comment
Comment #45709466
Neat, thanks for the link! Seems like the agent feedback loop made a big difference for your use case. Would be interesting to see how much of a difference more advanced methods (l…
-
comment
Comment #45691021
The algorithm works for MoE load balancing in general
-
comment
Comment #45691019
We've found that these frameworks do well for systems performance problems and expect that the range of problems for which they apply will increase as they models and frameworks im…
-
comment
Comment #45691005
That's a good point! The load balancing of the original algorithm was already quite good so our goal was to try to get something that could achieve similar results but could run fa…
-
comment
Comment #45690998
ADRS = AI Driven Research for Systems. See our previous blog post ( https://www.sigops.org/2025/barbarians-at-the-gate-how-ai-is... ) and our paper ( https://arxiv.org/pdf/2510.061…
-
comment
Comment #45690984
Thanks for sharing your blog! Very interesting work, 100% agree with your 3 criteria on the sweet spot for AI. Most systems performance problems fit right in
-
comment
Comment #45689732
The code was quite short and easy to read. Specifying the right scoring function and scoping the problem are key parts of getting good results with ADRS.
-
comment
Comment #45689728
Yes, getting the right workloads and ensuring correctness are crucial parts of the process
-
comment
Comment #45689301
This is a great question! By analyzing the logs of OpenEvolve with the full model outputs, we observed how the AI got its ideas (seemed to be pulling from literature in the space) …
-
comment
Comment #45688669
As an author of the blog, I'll note that this was one of the easiest applications of ADRS. Bowen, who was leading this effort, got things running within a day or two and the initia…
- story