tough

22 ms·

241.
by tough 1y ago
you have the separate the model , from the interface, imho. you can totally evaluate these as GUI's, and CLI's and TUI's with more or less features and connectors. Model quality is about benchmarks. aider is great at showing
242.
by tough 1y ago
glm 4.5 air too
243.
by tough 1y ago
works quite reliably with claude-code-router and any agentic SOTA model (not necessarily anthropic)
244.
by tough 1y ago
their owners charge per token so...
245.
by tough 1y ago
that made me chuckle
246.
by tough 1y ago
only for narrow applications where your fine tune can let you use a smaller model locally , specialised and trained for your specific use-case mostly
247.
by tough 1y ago
prob just told an employee to get it done no?
248.
by tough 1y ago
wasm is the perfect abstraction to build a -modular- and pluggable codebase (say grpc/proto as contracts) so you can just swap any part of it as long as the wasm module abides to such contract. I keep going more and more to it when I t
249.
by tough 1y ago
yep this literally happens with both AI + robotics at the government + education + business + district levels
250.
by tough 1y ago
if the US govt had told the company to get their shit together or close up after the first leak, the second one wouldn't have happened
251.
by tough 1y ago
you have higher trust in your government IT services than I do on mine
252.
by tough 1y ago
paper is on arxiv https://arxiv.org/pdf/2507.18074
253.
by tough 1y ago
can't reproduce locally with 4bit weights
254.
by tough 1y ago
Thank you! just found the 4bit (60gb) on LM Studio
255.
by tough 1y ago
Or just use any other CLI and API provider and let Anthropic hang with their -alignment- and selling out to warmongers actually you can keep using claude code (cli) tool with any provider, just replace the env endpoints. https://
256.
by tough 1y ago
fake for 5% of users. will they refund me my sub? when I subbed It was unlimited, they've rugged the terms twice already since then in less than a month
257.
by tough 1y ago
I cancelled my subscription I'll keep openAI and they dont even let me use CLI's with it, but they're at least Honest about their offerings. Also their app doesnt tell you to go fuck off ever, if you're Pro
258.
by tough 1y ago
then why sell fake -unlimited- plans to hook people up It lasted less than a week -unlimited- been a shit show cutting down since then
259.
Descend: A Safe GPU Systems ProgrammingLanguage [pdf](steuwer.info)
4 points by tough 1y ago|0 comments
260.
by tough 1y ago
I was wondering how one "made Buffett her professor" without him wanting to teach in the first place but that solves it
261.
by tough 1y ago
https://malwarwickonbooks.com/oss-spy/
262.
by tough 1y ago
the protocol spec itself has evolved rapidly as anthropic seems to have gone from side-project to taking it way more seriously as it took off most of the servers are meh, but some of the stuff is cool, security remains an issue but can&#x27
263.
by tough 1y ago
that's funny cause a lot of AI's dont' even know yet what MCP is even
264.
by tough 1y ago
it has limits too, it lasted like 1-2 weeks without only (for me personally at least)
265.
by tough 1y ago
I use act to run github CI locally fwiw https://github.com/nektos/act
266.
Simulate Harsh User Review for Claude code(github.com)
5 points by tough 1y ago|1 comments
267.
by tough 1y ago
A Model Context Protocol (MCP) server that simulates "fake" harsh user reviews designed to tame AI agents and enforce disciplined development practices.
268.
by tough 1y ago
has nothing to do with licensing and ´everything to do with business model
269.
by tough 1y ago
woah so this is like a react-native/expo done in a sane way ???? with swift? damn
270.
by tough 1y ago
maybe gptzero had your paper on its training data (it being from 2016)?

More ›