HackerTrans
TopNewTrendsCommentsPastAskShowJobs

krawczstef

no profile record

Submissions

Test Driven Development of LLM / Agent Applications with Pytest

blog.dagworks.io
2 points·by krawczstef·2 anni fa·0 comments

The graduation problem: avoid frameworks getting in the way

blog.dagworks.io
5 points·by krawczstef·2 anni fa·1 comments

Show HN: Modeling parallel (and hierarchical) agents with Burr

colab.research.google.com
2 points·by krawczstef·2 anni fa·1 comments

comments

krawczstef
·mese scorso·discuss
FYI - Burr is designed with recursion in mind, i.e. the ability to kick off Burr within Burr. So you could have an action that is managing several Burr subgraphs... Or you write your own management layer here.
krawczstef
·mese scorso·discuss
yep - here's something cool an end user wrote with Burr - https://github.com/msradam/phoebe

Burr just helps you, the engineer, to really control the primitives. Then adds some cool features you don't have to think about -- like observability :)
krawczstef
·mese scorso·discuss
no. more lower level.
krawczstef
·mese scorso·discuss
user contributed :)
krawczstef
·mese scorso·discuss
Cause I submitted it. Learning the Apache process and cranking on other things has been a slow process. But we've got some momentum and beginning more regular releases.
krawczstef
·11 mesi fa·discuss
I agree with this post.

Otherwise the larger picture is that MCP is a land grab for building an eco-system around integrations to get access to data. Your LLM agent is not valuable if it can't access things for you... and from a market perspective enterprise pays a lot for this stuff already, and yes MCP is not thought out at all for Enterprise really... At least thankfully they added stateless connections to the spec...
krawczstef
·anno scorso·discuss
Unfortunately their blog doesn't have the example they have on x.com - https://x.com/ZyphraAI/status/1888996367923888341 - which I think is more illustrative.
krawczstef
·2 anni fa·discuss
your comment made me think I read it wrong, but it's not closed source, they're just moving to a source available license...
krawczstef
·2 anni fa·discuss
yep, agreed. I think that's another way to view the current state of "GenAI" tooling (e.g. all those complicated frameworks that received $M in funding) and why things like https://www.anthropic.com/research/building-effective-agents fall on deaf ears...
krawczstef
·2 anni fa·discuss
+1 for vanilla code without LangChain.
krawczstef
·2 anni fa·discuss
Interesting! I feel like this is a cross between https://github.com/dagworks-inc/burr (switch state for context) and https://github.com/Netflix/metaflow because the output of the "task" declares its next hop...

The challenge with this approach though is that you need to run the actual code to see what it does, or as a developer build up a mental model of the code ... but it does shine in certain use cases -- and also reminds me of https://github.com/insitro/redun because it takes this approach too.
krawczstef
·2 anni fa·discuss
This is reminiscent of themes brought up in https://news.ycombinator.com/item?id=42253735 and most famously in https://news.ycombinator.com/item?id=36645575...
krawczstef
·2 anni fa·discuss
Forgot to mention, on the roadmap we have pluggable executors, e.g. delegating parallelism to whole agents/subgraphs to systems like "ray". Would love to collaborate with folks if that's interesting.
krawczstef
·2 anni fa·discuss
Looks great. But being burned by other "abstractions", e.g. LangChain, I'm weary of the oversimplification. How are you not going to make those same mistakes?
krawczstef
·2 anni fa·discuss
Does this compete with Minio?
krawczstef
·2 anni fa·discuss
getting into enterprise is hard, so probably trying open source to help with that.