it felt too easy to just work, seems like there is a bug with Mistral's `codestral` LLM that prevents files to be edited https://github.com/All-Hands-AI/OpenHands/issues/8958 #mistral #codestral #llm
[Bug]: Agent (Devstral) file modification frequently fails · Issue #8958 · All-Hands-AI/OpenHands

Is there an existing issue for the same bug? (If one exists, thumbs up or comment on the issue instead). I have checked the existing issues. Describe the bug and reproduction steps When the agent i...

GitHub
OpenAI ist nicht das Maß aller Dinge: Das französische Startup Mistral zeigt mit Codestral Embed, wie effizient und präzise Code-Embeddings heute sein können. In Benchmarks schlägt es sogar die Großen. Was steckt dahinter – und was bedeutet das für Entwickler? Jetzt mehr erfahren. #Mistral #Codestral #KI 👇
https://www.all-ai.de/news/news24/mistral-code-embeddings
Mistral setzt neue Maßstäbe bei Code-Embeddings

Codestral Embed beeindruckt mit Effizienz, Präzision und Preis – kann das Modell zum neuen Branchenstandard werden?

All-AI.de
Integrarea inteligenței artificale | Blogul unui internaut

Integrarea inteligenței artificale

Blogul unui internaut - Floss, non-floss, tech, jocuri și tv online, alte chichițe cotidiene
While utilizing #Mistral #Codestral #AI can be pretty useful as assistance tool, sometimes it *immediately* gets utterly confused. Especially if there are functions and classes in your code that have the same name as something it excessively was trained on.
It's utterly unable to digest a class called "Request". It will, always, try to use it like "request.read(chunk)" even if there's nothing like that anywhere to be found.
Still, it can be really useful #coding. Hope it will work on new NPUs.

Oh, the #AMD NPU driver already got upstreamed into #Linux 6.14? Fancy that, I thought I had to build myself a dkms module to get it running on OpenSuse. 🙂

Now I'm curious how involved it will gonna be to make something like #Codestral run on it. The usual tools are probably not yet well optimized to utilize that Co-Processor OOTB.

I have to admit, #Codestral is *phenomenally* useful to learn #Python (as long as you're always vigilant about potential errors).
#AI

Had a really good laugh with @hannah just now trying our hands at the clearly way superior & production ready act of #VibeCoding with #Codestral.

We told it to build a calculator in #Rust. After telling it to use an AST it looked promising, even adhering to order of operation (already better than MS Calc!).

Then we told it to add support for braces and it utterly and completely shat itself. Like, Mexican diet explosive diarrhea-style. 😂

Now everything is an invalid expression. #AI 👍
#coding

ffs, why does their docker only support Navi 31 and not Navi 32? 😩
https://hub.docker.com/r/rocm/pytorch

I just wish both #Nvidia and #AMD would stop with that whole licensing bullshit around #CUDA and #ROCm and just include that damn stuff in the default driver.
I just want to run #Codestral on my local machine so I can use it with non-public code. Will be troublesome enough to cram it into 16gb VRAM. 😑
#computer #Linux #AI

#Mistral #Codestral #AI really loves to have aneurysms. 🙃

(The generated code before the repetition is wrong as well, of course )
#coding