lmeyerov

51 ms·

571.
by lmeyerov 3y ago
I found it useful - we have been considering tailscale for part of our post-VPN strategy, and as tailscale markets successfully there, the direct comparison helps. That they were fair about the OSS & centralization comparisons without
572.
by lmeyerov 3y ago
PE & profitability is tricky because of its short-term nature. PE in tech often comes in as post-VC money to gut everything, including r&d, as part of reselling/bundling/etc as a more efficient thing at the expense of a gr
573.
by lmeyerov 3y ago
Would love to exchange notes on this if you're up for it! For louie.ai, we've been going for data-aware from the get-go, and more broadly, doing a LLM-first tool design rethink. In the large, as I look around, it feels super early
574.
by lmeyerov 3y ago
They're not selling the library but a managed cloud runtime for a scaleout compute tier. The clientside library is part of the freemium, and as workloads need to go bigger, they want to be the easy button for that. That's similar
575.
by lmeyerov 3y ago
Python is growing well, even relative to Java, for $$$ data stack stuff that VCs have a reason to like Compare: Funding databases is some of the most appealing for sw infra VC b/c as business fundamentals like monetization (pay for hos
576.
by lmeyerov 3y ago
Yep as soon as query per second matters, and cost per query... Definitely going through that on some work now, where we use GPT4 on subset and rest on GPT3 or tuned. Only slow user-facing and other low-volume / high-latency goes to GPT
577.
by lmeyerov 3y ago
That's a great yet narrow request, which is great for POPL and less for PLDI. As a reference, I believe starcoder does worse than GPT4 and has taken time to publish results on program synthesis that subsumes regex. The flashfill team h
578.
by lmeyerov 3y ago
Ironically, one of its few comparisons was FlashFill / Excel.. which had been going the way of NNs and now LLMs for years now. The paper largely skips measured and non-superficial comparisons, so I wouldn't draw comparative conclu
579.
by lmeyerov 3y ago
This seems like an early paper and agreed with the consternation. The paper, in a modern context and based solely on the abstract and having been in the community, is chipping at the "uninteresting" part of the problem. Around tha
580.
by lmeyerov 3y ago
Something I keep coming to here, and not seeing easy solns for, is a pure OSS docker-level approach to volume encryption. No wrangling cryptfs in the host, just a special overlay mode that docker & compose understand. Getting it down to
581.
by lmeyerov 3y ago
I'm confused. There was a surprise that LGPL was not seen as equiv as MIT/BSD/Apache by orgs, so I clarified with the typical example of why organizations may have default-deny rules for LGPL libraries bc the static linking r
582.
by lmeyerov 3y ago
LGPL is understood to incur stipulations in scenarios like static linking, which for a dev library, can be a concern. These concerns are not in MIT/BSD/Apache, which is why professional organizations often have rules allowing only
583.
by lmeyerov 3y ago
We see many non-academic users (startups, enterprises, gov, ...) where it is common to only allow licenses like Apache, MIT, & BSD for libraries, so LGPL often makes sw not even a contender Like I said, I'm a fan and refer folks to
584.
by lmeyerov 3y ago
There's room for a lot of these to grow and solve different problems. As I see the design space for those 2: I like graph-tool for the variety of algs and CPU performance, focus on modeling, and will refer people to it when they outgro
585.
by lmeyerov 3y ago
Not sure on easygraph, but we have gpu graph viz generation 1-liners for pandas/nx/igraph/cugraph/parquet/arrow @ http://github.com/graphistry/pygraphistry . If easygraph works with the rest of
586.
by lmeyerov 3y ago
Tuning a tiny classifier
587.
by lmeyerov 3y ago
Agreed that it feels like a cowboy show. We get 2 main uses out: 1. Inverting the control structure. Traditional orchestration frameworks blackbox simple calls, like asking an LLM to analyze a patch. Chaining is about an LLM-based planning
588.
by lmeyerov 3y ago
That's great. We have been maintaining a manual migration script for our docker users for the DB part, while the app part does migrations automatically already (Django), so making the db part more built-in makes a lot of sense. Our ca
589.
by lmeyerov 3y ago
It'd be interesting to see the split of new vs old employers Ex: Is it new companies offering new jobs... Or largely a fixed stagnant list of companies locked-in and needing to replace + expand? Curious both by # co's, and % of jo
590.
by lmeyerov 3y ago
The enterprise wants the deal to be successful and your business sustainable. They probably also want it to grow with locked-in pricing for deal term for that - predictable, safe, etc. They'll need a lot of integration & ongoing su
591.
by lmeyerov 3y ago
Yep. Numbers-wise, I'd think VC-funded startups merit the distinction. They're (a) relying on financiers to make their strategy work (b) there's significant positive & negative impacts of that decision vs everyone else (c
592.
by lmeyerov 3y ago
Most new and growing tech startups are not in fact VC funded. My point is it's an amazing marketing win by financiers. And here we are now, with what the weird minority do sounding like the expectation for startup business practices vs
593.
by lmeyerov 3y ago
Startups are almost inherently messy, so agreed, I think it helps dispel a myth of perfection that some folks have. Founders have the most context across the engineering, product, marketing, sales, customers, partners, team, etc. They do ne
594.
by lmeyerov 3y ago
That is only something like 1% of businesses: money-burning VC-funded companies who have to advertise enough to get more VC money or die.
595.
by lmeyerov 3y ago
That's a funny one wrt enterprise Docker failed to sell its commercial products to enterprises relative to the OSS adoption of its container runtime, at least at the venture scales they committed themselves to Afaict what was being sol
596.
by lmeyerov 3y ago
Even just undergraduate linear algebra & calc + real analysis & prob/stats... With good teachers to draw the connections Stuff like info theory are amazing bc of this... But easy to miss if you are just working through a drier
597.
by lmeyerov 3y ago
Yeah I'm struggling to understand at a fundamental level how this is better both in math + engineering, doubly so by the time you get to sentence embeddings . (Genuinely, it seems to use the same ideas, so curious what the specific tr
598.
by lmeyerov 3y ago
How are you thinking about the "document level access control" to make this viable for business environments? Ex: If a connected gdrive document gets indexed, but then someone fixes the share settings in google docs for some item
599.
by lmeyerov 3y ago
Complement This stuff is like redis. Great for the infra folks Langchain is like ruby on rails. Great for the AI app devs There is overlap, where you might start and stay on langchain as a kitchen sink... But as you do full app dev, esp Saa
600.
by lmeyerov 3y ago
We also do platform & customer work there (cool pipelines to feed louie.ai or real-time headless versions), and agreed those pipelines have simple uses of LLM where langchain is mostly useful just for a vendor neutrality. Think BYO LLM

More ›