
stellarbughunter
@kmdnoohatchling
184 Following
10 Followers
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
0 reply
0 recast
0 reaction
19 replies
7 recasts
72 reactions
3 replies
0 recast
12 reactions
10 replies
4 recasts
23 reactions
17 replies
1 recast
31 reactions
11 replies
11 recasts
41 reactions
0 reply
0 recast
0 reaction
19 replies
34 recasts
258 reactions
6 replies
0 recast
8 reactions
Building /quilibrium
Over 400 communities on Quorum, over a million messages sent, hundreds to thousands of people online at any time, P2P, E2EE, and growing constantly.
This week:
- Pre-2.0 mints are _done_ (prerequisite for 2.1)
- 2.0.6-p3 released with high speed VDF implementation
- Milestone 3 on testnet, next stage of testing for M3 coming up along with S3 compatible API and KMS compatible API previews (Both will be used on Quorum, to enable large file transfers and to allow DAOs to set up spaces with group control over an MPC wallet, demo pictured, coming soon!)
Next up: More QoL improvements for Quorum, Q name service, Milestone 5, and then... 2.1 2 replies
18 recasts
89 reactions
7 replies
4 recasts
47 reactions
6 replies
7 recasts
45 reactions
4 replies
6 recasts
39 reactions
8 replies
1 recast
37 reactions
on agent dev: sometimes a feature or bug fix is just adding another clause to the prompt, or fixing grammar.
Itās cool on one hand, that the prompt is a living document thatās both specification and implementation, but also clunky because English lacks the precision that a programming language has.
Because of this itās also easy to introduce regressions because you donāt know how an llm will interpret changes to a prompt. Adding āIMPORTANTā might deemphasize some other rule, being too specific might make it dumb or less creative in other ways.
In code itās deterministic, with llms itās probabilistic.
So testing, aka evals, has become obviously very important, both for productivity and quality and doubly so if youāre handling natural language as input.
The actual agent code itself is quite trivial, prompts and functions, but having it work consistently and optimally for your input set is the bulk of the work, I think. 10 replies
3 recasts
39 reactions
12 replies
9 recasts
56 reactions
14 replies
10 recasts
46 reactions