lmeyerov

47 ms·

631.
by lmeyerov 3y ago
Their neighbors whose median per-capita income is 5 figures seem more relevant of an analogy The serf analogy makes sense for low-income Americans who essentially have to pay more to the capital class: payday loans, cannot afford bulk disco
632.
by lmeyerov 3y ago
Confusingly, there is a Apache licensed repo you can find linked at the bottom of the site, except the files are individually relicensed, eg, https://github.com/unskript/Awesome-CloudOps-Automation/blob... The con
633.
by lmeyerov 3y ago
Agreed. They brush aside the years of high performance computing graph implementations, eg, cuStinger. If you look at systems like neo4j's GDS and how other multipurpose systems are going wrt views/projections for accelerated comp
634.
by lmeyerov 3y ago
I'm curious how much of this is speculative vs from use? We are doing code generation for our users (analysts & data scientists), and I found going by method names & type signatures failed pretty hard, while contextual code sni
635.
by lmeyerov 3y ago
Yeah this, setting the price to $100k to guide core laws around $B businesses in a way to override public feedback is cheap. Sounds like it is now officially open season.
636.
by lmeyerov 3y ago
Baking in western social norms, and eventually all global regional ones, is a big competitive advantage. The corporate side doesn't want ethical but bland . Even most coloring outside of those lines should still be bland too - think D
637.
by lmeyerov 3y ago
yeah those numbers seem like they're from 2003 not 2023
638.
by lmeyerov 3y ago
FWIW, individual classes are generally tiny, so we found using langchain is fine and then for places we need to beef up (chunking, not calling 'eval', ...), we do our own class/subclass. That way we can align with community f
639.
by lmeyerov 3y ago
This seems cool for a lot of normal faas I keep wanting a flow like this to work for faas python data science sandboxed serving, esp @ scale -- python interpreters talking to cpython kernels and one day GPU passthrough -- but the experienc
640.
by lmeyerov 3y ago
Yes, regular computers have had SIMD for decades. Ex: Intel was SSE and now AVX. Fancier servers will have the same, except maybe wider and some other tricks. If you go down the vectorization path, unless you are a crufty CPU DB vendor, I&#
641.
by lmeyerov 4y ago
There are a lot of cool experiments happening like this, both principled and from intuition Ex:Label prop following non-local jumps ("metapaths") based on community detection (network-of-network, ...) or other shapes. Same thing f
642.
by lmeyerov 4y ago
Agreed, the definitions here of what a GNN is (=> cannot do) seem pedantically narrow & extreme, and then throws the baby out w the bath water See last ~2 years of work by Michael Bronstein's teams (who moonlights as Twitter&#x2
643.
by lmeyerov 4y ago
We are helping our users with qa tasks involving code generation, where the answers may be either JSON, executable code, or markdown discussions involving the same. We are tuning for a bunch of tools following that pattern so our users don&
644.
by lmeyerov 4y ago
You can do layered retrieval and summarization/embeddings to fit within the context bounds. Conceptually it's not far from what a clinician would do anyway. By switching from a naive one shot prompt with big context to interactive
645.
by lmeyerov 4y ago
Prompt testing, especially when for q/a pairs where there are multiple right answers, has been bugging me a lot The article is reasonable, but also shows a big gap in tooling, as the techniques there feel closer to linting & typing
646.
by lmeyerov 4y ago
I'd be curious if someone loaded these into http://github.com/graphistry/pygraphistry and tried something like overlaying a GNN anomaly detection over the base visuals. Both parts are automated & GPU accelerat
647.
by lmeyerov 4y ago
That's a legit hypothesis! Speaking as a scientist, I'd love to see a round 2 of this work to see how much things are the same vs different. The premise for the work was doing more serious sociological methods could help underst
648.
by lmeyerov 4y ago
For a bit of a quantitative analysis on this, we had fun doing surveying programmers here, see Table 7 @ https://lmeyerov.github.io/projects/socioplt/papers/oopsla20... Ex: Programmers value types much more f
649.
by lmeyerov 4y ago
End users don't care who in your dependency stack is enabling NLP, but that your tool can do things the non-NLP tools cannot Likewise, that is key signaling for all stakeholders on what is to come. If using AI is part of the core ident
650.
by lmeyerov 4y ago
ping gpt@graphistry.com for early access :)
651.
by lmeyerov 4y ago
We build an analyst tool that's popular with security/fraud/etc folks, and have been adding features here requested by them, so this is a great learning opportunity around their ChatGPT experiences motivating those requests H
652.
by lmeyerov 4y ago
Source? Every path is hard, so saying that's the worst is one of the most extreme claims one can possibly make. Ex: I can argue that VC funding is much worse than having paying customers in your space and the luxury to pivot & expl
653.
by lmeyerov 4y ago
Good luck to the team! We went back-and-forth in diff APIs here before coming back to langchain. The march release with async streaming, and growing plugin work, are examples of why. Balancing VC funding with OSS will be hard. I like having
654.
by lmeyerov 4y ago
In the US, you can part-time consult, as programmers are paid way above the national median income. This can either be in a lucrative field, giving you more time, or in your vertical, giving you more pmf exposure. The result is living like
655.
by lmeyerov 4y ago
Interesting, since I looked last year, Chrome has started raising the caps internally on buffer allocation to potentially 16GB: https://chromium.googlesource.com/chromium/src/+/2bf3e35d7a4... Last time I trie
656.
by lmeyerov 4y ago
It seems the quality goes up & cost goes down significantly with Colossal AI's recent push: https://medium.com/@yangyou_berkeley/colossalchat-an-open-so... Their writeup makes it sounds like, net, 2X+ over Alp
657.
by lmeyerov 4y ago
People can write slow code in any language ;-) We've had to fire contractors who "wrote" dataframe code but was not vectorized in practice despite repeat request to do so. Same thing for slow code in CUDA. From a maintenance
658.
by lmeyerov 4y ago
Fair questions! We work with everyone from individual university researchers trying to understand cancer genomes or European economic plans in their graph DBs, to big corporations struggling with supply chains in Databricks, to government c
659.
by lmeyerov 4y ago
We (and many others) have a team building fun things into our data analysis tool here. For what will soon become a 10% time prompt engineering role for a much easier kind of security investigations experience, we are hiring cleared security
660.
by lmeyerov 4y ago
We came to the US with a few suit cases and had to abandon all property in our home country. In general, many immigrants move to start over due to the relative lack of opportunity at home, whether due to politics, crime, economy, whatever.

More ›