Guillaume86

7 ms·

1.
by Guillaume86 20d ago
I'm running it on my 8 years old 2080ti 11 gb VRAM + 32 gb ram. After tasking fable to optimize the setup for a few nights it's pretty good for some quick funny clips (2min30 for 5s , 5mins for 10s, with reference pics to insert a
2.
by Guillaume86 2mo ago
Just took a glance so I may be completely wrong, fell free to ignore, but if I understand correctly, tsrx is just syntactic sugar for normal tsx code and does not enable anything new? If this is the case: I think introducing tsrx as part of
3.
by Guillaume86 3mo ago
Just prompt it that way then, I can get Claude to plow through features for 12h with the correct prompt and setup, no need to destroy the tools that are useful when you want the opposite.
4.
by Guillaume86 3mo ago
Just prompt it to not ask you questions then, the point of the ask user tool is to collect human input...
5.
by Guillaume86 3mo ago
OrioleDB seems interesting, not sure it's what you're after but worth a look.
6.
by Guillaume86 3mo ago
Great way to launder illegally obtained data too.
7.
by Guillaume86 3mo ago
The EU could/should probably legislate against this, it's bonkers...
8.
by Guillaume86 4mo ago
3. If your device run on battery, why not using a relatively cheap network call in place of a very power hungry local inference call?
9.
by Guillaume86 4mo ago
Yes I'm also using it for coding: I often make the agent use WebSearch in the research phase when deciding on a stack or a library or research best/modern practices to do achieve something. As for images I find it super useful to
10.
by Guillaume86 4mo ago
You seem to have tried a few things, if you don't mind I have a few questions as someone currently on Claude Code but would prefer to not lock myself in a commercial ecosystem (and their pricing change regarding headless usage is annoy
11.
by Guillaume86 4mo ago
Tangential question for Claude Code subscribers, mid June `claude -p` will move to api pricing (with some "SDK credits" before it kicks in), so headless usage will become 20-30 times more expensive, and all these high level orches
12.
by Guillaume86 4mo ago
I suppose folks here already know this but it deserves a mention: subscription pricing is 10-20x cheaper than API pricing at Anthropic for example and it will be a far better experience (better models, faster responses, as much parallelism
13.
by Guillaume86 4mo ago
I think we should separate the private AI discussion from the local AI discussion. The pragmatic choice to run big LLMs is one/several big servers online, but that doesn't mean private companies should be the only ones to run the
14.
by Guillaume86 5mo ago
Sorry yeah it was a big vague, I was thinking about creating a Libretto MCP since it's a/the standard way to share AI tooling nowadays and that would make it usable in more contexts. In that case, the protocol has a feature called
15.
by Guillaume86 5mo ago
Did you consider MCP sampling to avoid requiring your own LLM access? (for the clients that support it of course, but I think it's important and will become standard anyway)
16.
by Guillaume86 5mo ago
I think sometimes about this, does it really make sense? Financially I mean. The is just my impressions and I'm glad to be corrected if someone has hard numbers and some experience going this route: At the moment LLMs vendors are in ma
17.
by Guillaume86 6mo ago
Yeah this should be a citation in the sanewashing wikipedia article.
18.
by Guillaume86 6mo ago
Yes it's even more effective this way IMO, we will probably see some 11/10 mental gymnastics from people condemning this and failing to apply the same standards to billions dollars corps.
19.
by Guillaume86 6mo ago
Yeah it's just a slightly more honest and simplified presentation of what LLMs providers do IMO.
20.
by Guillaume86 6mo ago
Eh I think part of the joke is that LLMs have gobbled up the original source code, and if you help them enough (identical type signatures and specs), they will output the same code, it's the copyright laundering problem.
21.
by Guillaume86 7mo ago
Yeah it’s funny this little mishap gets traction here in the age of commercial LLMs and generative AI... Bashing microsoft never gets old I guess
22.
by Guillaume86 7mo ago
Yeah it’s hilarious seeing people lose their shit over this and not like, every commercial LLM vendor...
23.
by Guillaume86 7mo ago
He said he know about it and your conclusion is that he doesnt know about the other ones...
24.
by Guillaume86 7mo ago
I remember writing a parser for some SL instructions following an HN thread, seems like it was 13 years ago -_- : https://github.com/guillaume86/SolLewitt
25.
by Guillaume86 7mo ago
Don't need to be a nerd, 5 secs google search: https://ourworldindata.org/grapher/death-rates-road-incident...
26.
by Guillaume86 8mo ago
Well that explains it then, don't offer stupid salaries before you make stupid money...
27.
by Guillaume86 10mo ago
They also outperformed themselves before being in a leader position...
28.
by Guillaume86 10mo ago
The switch is underpowered to the point that most A(AA) games cannot run on it without a ton of effort and compromise, an M chip powered device would be a different story. But anyway it's never going to happen, just daydreaming about a
29.
by Guillaume86 10mo ago
I'm definitely dreaming but I think it could be a win-win situation if Apple decided to licence its chips to Valve: the resulting handheld and VR headsets would be power/efficiency monsters and PC devs would finally have a good re
30.
by Guillaume86 11mo ago
You should use the DB explain or equivalent command to spit out the query plan, limit 1 shouldn't change anything in your case, if it's not the case you should file an issue, it's pretty much 101 of query optimization.

More ›