theredsix

5 ms·

1.
by theredsix 11d ago
Congrats on the launch! What's different between Jev and Microsoft's Guidance package? https://github.com/guidance-ai/guidance Is it a diffusion generator under the hood?
2.
by theredsix 16d ago
Thought I was going crazy, glad to know I wasn't the only one.
3.
by theredsix 1mo ago
If you're getting fired in 2 weeks, it's likely not because of a single comment.
4.
by theredsix 2mo ago
10% cheaper than official! Let the inference pricing wars begin!
5.
by theredsix 5mo ago
Genius I love it!
6.
by theredsix 5mo ago
Are you guys going to follow up with a paper showing EDEN results match or beat turboquant for needle in a haystack benchmarks?
7.
by theredsix 6mo ago
With AI you actually don't need to choose anymore. Well laid out abstractions actually make AI generate code faster and more accurately. Spending the time in camp 2 to design well and then using AI like camp 1 gives you the best of bot
8.
by theredsix 6mo ago
Extrapolating the benchmarks, this would imply the best RYS 27B is capable of out performing the 397B MoE?
9.
by theredsix 6mo ago
super clever and awesome!
10.
by theredsix 7mo ago
The freeze sometimes does capture in between states. What I've seen the agent does in those cases is that it recognizes it's in between states and calls browser_wait(). Where the agent goes off the rails isn't a snapshot in t
11.
by theredsix 7mo ago
*bows
12.
by theredsix 7mo ago
It's a long shot but getting ABP to be a first party citizen alongside CDP would be my dream!
13.
by theredsix 7mo ago
Right now, it's evading all anti-botting detectors I've tested it on. I believe it's due to the fact it runs in headful mode and I've removed all detectable CDP signatures. Input events are also simulated at a system lev
14.
by theredsix 7mo ago
Maintaining the fork isn't so bad, the core chromium changes are only a few hundred lines and I was able to extend already existing concept like debugger pausing and virtualtime emulation while riding off mojo IPC for cross thread comm
15.
by theredsix 7mo ago
Exactly! This race condition is exactly the category of problems ABP will solve.
16.
by theredsix 7mo ago
Yes agreed, added to the roadmap!
17.
by theredsix 7mo ago
hmm, like adding an optional css selector for targeting?
18.
by theredsix 7mo ago
It runs in headful mode and all control signals are passed in as system events so it bypasses the problems browserbase identified.
19.
by theredsix 7mo ago
I've consolidated most of the changes in chrome/browser/abp and used shim's for the other modifications so rebase is light and handleable by Claude. I'd love to get this upstreamed. An intro to the chromium maintena
20.
by theredsix 7mo ago
/superpowers! that plugin is the GOAT
21.
by theredsix 7mo ago
those are factored into the wait heuristic only if there's a navigation event since clicks on an already loaded page won't trigger those. You can point Claude/codex at https://github.com/theredsix/agent-b
22.
by theredsix 7mo ago
Good question! ABP keeps a list of all same/parent/sibling network request and wait for them to complete within a timeout. If the timeout hits, it'll still freeze and screenshot back to the agent. There's a browser_wait(
23.
by theredsix 7mo ago
Great insight! ABP exposes display resolution controls right now. I've noticed almost zero reCAPTCHAs during testing compared puppeteer stealth or other packages. Regarding the freezing mechanic, virtualtime is paused as well and the e
24.
by theredsix 7mo ago
Updated the table!
25.
by theredsix 7mo ago
Yeah, I noticed CPU use goes to near zero during the pausing phase. You can also trigger pause via REST/MCP so a script can take advantage of these abilities as well.
26.
by theredsix 7mo ago
agent-browser's biggest selling point is a CLI wrapper around CDP/puppeteer for context management. It'll have mostly the same pros/cons as CDP on the table.
27.
by theredsix 7mo ago
I tweeted at the OSUNLP and they're backed up on eval validation. In the meantime, here's the benchmark repo with the saved runs and also instructions on how to run it locally. https://github.com/theredsix/abp
28.
by theredsix 7mo ago
90% average with 85.51% hard!
29.
by theredsix 7mo ago
Totally, I feel that CDP was designed for a different category of automations.
30.
by theredsix 7mo ago
Op here, happy to answer any question!

More ›