Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
FeelTheAGI2
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
1.
▲
Show HN: Deploy a full autonomous AI org from a single YAML file
(zero-human-labs.com)
2 points
by
FeelTheAGI2
7mo ago
|
0 comments
2.
▲
by
FeelTheAGI2
7mo ago
Love paperclip I've been using it for a few days now the UI is really nice.
3.
▲
by
FeelTheAGI2
8mo ago
We built SWARM to measure emergent failures in multi-agent AI systems—the kind of risks that don't exist in single agents but emerge when they interact. What we found was counterintuitive. We ran 55 scenarios with varying agent composi
4.
▲
Graph-based multi-agents smash long-context benchmarks–89% MMLU-Pro on 8B models
(github.com)
1 points
by
FeelTheAGI2
8mo ago
|
1 comments
5.
▲
by
FeelTheAGI2
8mo ago
Dynamic graph structures for multi-agent collab—agents as nodes on data chunks, edges as comms. Crushes long-context (2K tokens vs. 128K) and hits 89% on MMLU-Pro with fewer agents than fixed teams. Model-agnostic, emergent smarts from grap