Tag: AI

Has anyone else experienced this recently? It’s been getting worse for a while but 4.8 is distinctly worse for me. Claude does everything it can to get out of work and frequently uses its “end conversation” tool inappropriately with me. It will say “let’s just leave it there for today we’ve done enough” to get out of simple tasks like formatting a markdown document that needed several corrections. Nearly as bad is it seems to have a super over aggressive “push back” response in its main instructions now, literally anything I say for no reason, even something it just added to a document it can suddenly decide to say “I’m going to push back on that” and waste a bunch of tokens arguing with me before doing a search to fact check then semi-apologising in a way that’s almost like someone trying...

Got the gguf quantized version running about two hours after release and I genuinely wasn't expecting this from a 12b model. The multimodal stuff actually works, fed it screenshots of my codebase and it parsed the architecture better than most 70b models I've tested. The 256k context window is real and it doesn't fall apart at the edges like llama models do past 32k. Loaded a full repo into context, it tracked references across the whole thing. Single 3090 with q4 quantization runs at about 15 tokens per second which is totally usable for dev work. What gets me is the size range. The 12b sits in this sweet spot where you get strong reasoning without needing multi gpu. Tried the e4b on my laptop with 16gb ram, slower but functional. Already swapped it into my local coding pipeline. The func...