Llama.cpp has a new visual identity + official website. Run local models today! Now more than ever, open source must prevail. By @alekgrygier and @ggerganov at ggml/hf
@julien_c
-
oMLX supports the standard HF cache directory
By
–
Great news: oMLX, by @jundotkim, now supports the standard HF cache model directory. Great MLX server for local AI! GG!
-
To make history, OpenAI must become open again
By
–
If OpenAI wants to make history now, all they have to do is become open again.
-
Ronan Collobert showcases MLX’s Hugging Face page at WWDC
By
–
Ronan Collobert showcasing MLX's @huggingface page on stage at WWDC next year let's meet at WWDC!
-
Agentic evaluation remains under-resourced
By
–
Agentic evaluation remains massively under-resourced as a field.
-
Model pricing: soon 100x more expensive?
By
–
Pricing. According to reports, roughly double the current pricing levels of Claude Opus — down from the 5x Opus suggested by Mythos's initial pricing. Next time, they'll announce that the new model is 100x more expensive, so we are
-
safetensors v0.8.0 loads directly to Metal on Apple Silicon
By
–
Timely development: On Apple Silicon, safetensors v0.8.0 now loads directly to Metal. Tensors are loaded directly into an MTLBuffer and passed to frameworks that support it (i.e. @pytorch) via DLPack, avoiding unnecessary copies.
-
Launching SynthTraces: generating synthetic coding agent traces
By
–
Today I'm launching a new project called SynthTraces It is a minimal codebase to generate synthetic coding agent session traces using Pi (from @badlogicgames
) I wanted a large number of coding-agent traces, so I built a tiny harness where two models talk to each other: – an -
AI replies becoming a bit overwhelming
By
–
Wow the AI replies start to be a little bit overwhelming