Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
_boffin_
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
21 ms
·
301.
▲
by
_boffin_
3y ago
What are your favorite repos that are built with Django that have good practices.
302.
▲
by
_boffin_
3y ago
Actually working on something like that right now. Hopefully have a rough prototype in a month or so.
303.
▲
by
_boffin_
3y ago
No rainbowbrackets in their IDE I take it?
304.
▲
by
_boffin_
3y ago
They’re doing marketing to sell more shovels as they’re a shovel selling company.
305.
▲
by
_boffin_
3y ago
Interesting. One thing i noticed is that Mistral has a `max_position_embeddings` of ~32k while these have it at 4096. Any thoughts on that?
306.
▲
by
_boffin_
3y ago
How did your perspective change and what changed moving forward?
307.
▲
by
_boffin_
3y ago
If it worked for Steve Jobs, maybe they're thinking it could work for them too?
308.
▲
by
_boffin_
3y ago
Can you show me the metrics for that, please. Additionally, what's your company's IP policy? Do they let you send their code to a remote company or do they see that as a breach of security?
309.
▲
by
_boffin_
3y ago
Which model are you using? I tried the deepseek 34b or whatever, but it goes nuts at times. Also, what params do you have for Prompt Context Size and Max Completion Tokens?
310.
▲
by
_boffin_
3y ago
It seems like it's running at comparable speeds to GPT4 prior to Turbo. I could be wrong, but what I'm trying to say, it ain't bad at all.
311.
▲
by
_boffin_
3y ago
Max ram is saying 22.72GB. They gave me a laptop with a M2 MAX and 64GB of ram. I'm loving the fact that i can toss any piece of code in there without having to worry about data privacy issues and or any retention policies of service p
312.
▲
by
_boffin_
3y ago
I will only want to use this if i'm able to host the LLM locally. Currently using CodeGPT and running CodeBooga 34b @ 4bit on my work laptop. This is good enough for me, but i'd love some more advanced functionality, such as code
313.
▲
by
_boffin_
3y ago
Not a blog, news board, or youtube precisely, but I think these two are great resources. - Thoughts on the Market by Morgan Stanley: https://www.morganstanley.com/ideas/thoughts-on-the-market - Notes on the Week Ahead
314.
▲
by
_boffin_
3y ago
Namecheap is telling me that any of the `.ing` tlds i'm looking for are $13k... sorry, but f-off!
315.
▲
by
_boffin_
3y ago
Does LittleSnitch allow for an import of urls, IPs, etc.. to block? if so, know of any lists?
316.
▲
by
_boffin_
3y ago
Analysis of direct and special assessments not related valuation of land and improvements.
317.
▲
by
_boffin_
3y ago
Force projection to keep the shipping lanes open for global commerce in one of the most important shipping lanes in the world.
318.
▲
by
_boffin_
3y ago
You have companies that are spending good money on fine-tuning and will start spending money on fine-tuning. It seems like it would almost be easier to just go directly to these companies by looking at their blog posts--they're telling
319.
▲
by
_boffin_
3y ago
it grow. we grow?
320.
▲
by
_boffin_
3y ago
Wouldn't each object on each ring pull on one another and cause stability issues of that structure due to their relative closeness?
321.
▲
by
_boffin_
3y ago
Posted in a prior thread, but here's a trade-by-trade recap Original Thread: https://news.ycombinator.com/item?id=37459495 --- If you want to see how it looked like from the tick scale, take a look here: http:/&#
322.
▲
by
_boffin_
3y ago
looking forward to reading this later today.
323.
▲
by
_boffin_
3y ago
At a prior company, there's a 3 digit year floating around when it shouldn't be a 3 digit year. no one knows why and or who owns it.
324.
▲
by
_boffin_
3y ago
I was actually in the process of developing something just like this. Think of not wanting to pollute your main conversation when you come across something you don’t know in the LLMs response, so you create a side chat and then navigate bac
325.
▲
by
_boffin_
3y ago
Larry Summers? like the Larry Summers?
326.
▲
by
_boffin_
3y ago
Question: how difficult is it to get that no retention waiver on prompts and responses?
327.
▲
by
_boffin_
3y ago
Let me ask you a question before answering. Question: Do you plan to implement valuable enhancements over the coming years or do you believe this is feature complete? > who prefer perpetual licenses, is this because of a perceived lower
328.
▲
by
_boffin_
3y ago
What are your thoughts about single time purchase and or the JetBrains model?
329.
▲
by
_boffin_
3y ago
Data is my hobby and i love how it lays data and its relations.
330.
▲
by
_boffin_
3y ago
I love this, but am not going to pay a subscription for this. $420 a year? That’s more than IntelliJ Ultimate! Doesn’t even come with a fallback license that I can use when I stop subscribing.
More ›