Joined December 2009
- Tweets 13,896
- Following 193
- Followers 701
- Likes 3,198
Photos and videos
is @claudeai actively trying to prevent people from using third party MCPs?
Claude Desktop doesn't even have UI that lets you add an MCP?! you have to edit an obscure JSON file and then restart it ๐ซค๐
why? I don't get it. ๐คทโโ๏ธ
49
May 28
Western providers of DeepSeek V4 Pro look really bad on OpenRouter right now - the cheapest provider is 6x more expensive than first-party inference, and badly degraded at 4-bit quantizations.
openrouter.ai/deepseek/deepsโฆ
3
1
84
May 28
It's almost as if they're just teasing the west with an open release of this model. Western providers aren't cutting corners with a 4-bit version - they literally can't fit the full model on the H100 fleets they have.
helloai.com/articles/deepseeโฆ
1
47
May 28
US export controls blocking DeepSeek from buying H100/H200s apparently backfired spectacularly. The Huawei-based architecture, essentially a workaround for inferior silicon, can't run on western hardware.
Probably the best open model, and we ship an expensive garbage version. ๐คฆโโ๏ธ
73
May 24
this is just about the worst answer I've seen by Claude
claude.ai/share/4e4c01e8-886โฆ
this nonsense belongs in a <thinking> block where the end user doesn't see it - that's what reasoning is, right?
they've literally fine tuned it to give wrong answers first - why did they do that??
1
57
May 24
it does this all the time with more important, technical questions as well - sometimes while giving step by step instructions on shell commands, it'll suggest the wrong thing and then "but wait actually" itself. ๐คฆโโ๏ธ
35
Apr 29
we have a great idea for a product so lets raise VC and spend it all on Claude who will immediately train on the product so the next version can one shot a clone in 5 minutes no wait
"but Claude doesn't train on my data"
ofc they don't because that would be... very profitable?
49
Apr 27
"wait actually" is the worst, most idiotic thing anyone has ever done to the training of any LLM, ever
Claude Sonnet does this every day, all day now ๐ญ
wtf do they think the reasoning phase is for???
no one needs to see this
think and THEN answer
seriously, wtf??? ๐
49
Apr 13
This would be an instant fix to make your timeline so much better
@X why is this not allowed? WHY? ๐ญ
1
1
41
Apr 10
this React pattern always makes me stop ๐ค
in any other JS context, this would be a bug
thanks to the terrifying magic of global state and side effects (aka hooks) this "works"
we're still creating a new ID on every render, react just ignores it and throws that work away ๐
1
128
you take someone's code and rewrite it in a different language, isn't that "derivative work"?
you can't just do that and change the licence, can you?
more so, you take their code and use software to transform the code - surely that's "derivative"?
1
1
100
actually, you know what, don't bother explaining that one, because clearly none of this would hold up in court.
given that you openly admitted (even bragged about) that's what you did.
I'm pretty sure you'll get a case and desist, or more likely damage claims.
good luck. ๐
1
1
33
also ๐ the people openly "rewriting" Claude Code are using their paid Claude Code accounts to do it ๐คฃ
you are literally submitting the evidence, the original source code, using an account in your name, with your credit card ID on it? ๐
you literally hand them the proof!? ๐คฃ
85
"reasoning" has no right to be called "reasoning"
I know it "works" for certain tasks, but it has nothing to do with reasoning - the model just generates "more words" in the hopes of fuzzily stumbling into some useful activations some percentage of the time
1
40
tbh Claude Code is hopelessly mismanaged and a complete mess
github.com/anthropics/claudeโฆ
opened two detailed issues and the bot just closes everything without anyone ever even looking
OpenCode is vibing pretty hard rn, but at least they haven't completely given up on governance ๐
1
93
Feb 28
Open models are over just as fast as they arrive.
I've tried two of these (Minimax M2.5, Kimi K2.5) but I think I'll just skip GLM5 and withdraw from the pursuit of open models for another 3-6 months.
There is real progress, but never what YouTubers promise every single day. ๐
1
138
Feb 24
why is "100% Prettier compatibility" always the first thing advertised by every alternative to Prettier?
Prettier is awful - cramming as much code into every line as possible is not a "gold standard" that formatters should chase ๐ฎโ๐จ
please try dPrint!
and spread the word ๐
87
Feb 22
you know how you click in VS Code or Brave on Linux, and for some reason it drags instead of clicking?
no, it's not a hardware defect, it's this:
bugs.launchpad.net/ubuntu/ sโฆ
first reported in 2010
how does something this small and basic just get ignored for 15 years? ๐
120