EmilStenstrom

9 ms·

61.
by EmilStenstrom 2y ago
We are also only using it because of GDPR requirements of not transferring data outside of the EU. The slowdown is huge, preventing any chat use-case where users will get bored waiting for a 10 second response to trickle through. That OpenA
62.
by EmilStenstrom 2y ago
Gambling site or not: Cloudflare took their money for years, failed to communicate any problems, then deleted their data when they didn't accept their "enterprise deal". There's nothing saying that they won't do the
63.
by EmilStenstrom 3y ago
This CAN work, if you animate in the new lines slowly and (maybe) disable clicks while new data is coming in.
64.
by EmilStenstrom 3y ago
At least the can_ada (canada!) makes it a little bit more unique.
65.
by EmilStenstrom 3y ago
Some feedback: Did you consider what happens if the blog post contains the string "]]>"? I think that probably needs to be escaped to avoid problems?
66.
by EmilStenstrom 3y ago
Contrary to many other models I've tried, this one works really well for Swedish as well. Nice!
67.
by EmilStenstrom 3y ago
Product planning should always be done in a tree structure. The top nodes are business objectives, then under them are products, then under them are customer problems, then under them are backlog items. Saying you have too many backlog item
68.
by EmilStenstrom 3y ago
Allow people that don't know SQL to query a database.
69.
by EmilStenstrom 3y ago
Ideally, your app would be server-rendered on the first load, instead of first loading all data, and then filling a SPA from that.
70.
by EmilStenstrom 3y ago
Thank you!
71.
by EmilStenstrom 3y ago
Here's a glossary to understand this post: - mixtral-8x7 or 8x7: Open source model by Mistral AI. - Dolphin: An uncensored version of the mistral model - 3.5-turbo: GPT-3.5 Turbo, the cheapest API from OpenAI - 4-series preview OR &quo
72.
by EmilStenstrom 3y ago
They meant GPT-4 Turbo, which is an improvement over GPT-4.
73.
by EmilStenstrom 3y ago
It's interesting to see these 2-9% improvements from version to version. They are always talked about with disappointment, as if they are too small, but they also keep coming, with each version being faster than the previous one. I pre
74.
by EmilStenstrom 3y ago
The author did lots of research. Here's a comparison of some common other ways to do this: https://docs.epsio.io/faq/
75.
by EmilStenstrom 3y ago
https://slack.com/ is down too. Someone is having a terrible day...
76.
by EmilStenstrom 3y ago
TLDR: There was a error handler that used print(error) instead of console.log(error).
77.
by EmilStenstrom 3y ago
The status quo for a Django site right now is that everyone that uses it needs to pick a framework themselves. That includes evaluating the "state of js frameworks" and knowing about all the tradeoffs. That's a lot of mental
78.
by EmilStenstrom 3y ago
That's why I suggested following the Phoenix LiveView route. It's a very thin and fast JS-layer that defers processing to the backend. Avoids framework churn.
79.
by EmilStenstrom 3y ago
Django was a "batteries included" framework about 10 years ago. But the core team had very strong opinions about javascript, saying that picking a javascript framework should NOT be part of those batteries. Time has passed, and ja
80.
by EmilStenstrom 3y ago
Relevant context: Rarbg just shut down (see message on https://rarbg.to/ )
81.
by EmilStenstrom 3y ago
tldr: - Quality on task : For every task we tried, ChatGPT is still stronger than Vicuna on the task itself. MPT performed poorly on almost all tasks (perhaps we are using it wrong?), while Vicuna was often close to ChatGPT (sometimes very
82.
by EmilStenstrom 4y ago
This is VERY exciting: A PEP attacking one of the biggest problems with the CPython implementation: The Global Interpreter Lock. Will Python finally be rid of it after a decade of failed tries?
83.
by EmilStenstrom 4y ago
In Sweden, it has been shown since 1969 on New Year's Eve. The distribution rights were bought 1963, but since there was so much drinking involved it took 6 more years before it was broadcast. We use the Swiss version that is 11 min lo
84.
by EmilStenstrom 4y ago
While this is cool, using an MD5 of users’ email addresses is not cool. Given a large rainbow table you could likely find the identity of lots of users and tie them to specific comments where they thought they were anonymous… not th intenti
85.
by EmilStenstrom 5y ago
Looking forward to a visualization that ranks each reply by capita too. Should separate the tech hubs from the large populous regions.
86.
by EmilStenstrom 5y ago
3.10 will be released tomorrow. 3.11 is still very much work in progress and is expected to drop in 2022.
87.
by EmilStenstrom 5y ago
Hacker news needs a redesign, the UI is horrible by basic accessibility standards: Very small click targets, horrible mobile experience, too small font size.
88.
by EmilStenstrom 5y ago
The natural next step is to require bot labels for bots. Suddenly machine learning models that detect and bans bots without a label will be feasible. THAT will have a big impact on "healthy conversations" which seems to be their
89.
by EmilStenstrom 5y ago
That article is VERY one-sided. I think this is a better one: https://jakubferencik.medium.com/the-controversy-of-bret-wei...
90.
The case for national digital currencies, not Bitcoin(friendlybit.com)
2 points by EmilStenstrom 5y ago|1 comments

More ›