More notes on nearby ideas
Reply · Jun 26, 2023@TheRealFloomby hmm no, I'll try that.. tho this isn't necessarily for MLReply · Jan 19, 2023@zestular who chose the training tweets for each classification???Reply · Apr 14, 2025@kalomaze cacheing fuckery tokenizer fuckery pseudorandom that isn't distributed properly MoE but reverse a sign in the routing between training vs…Reply · Dec 17, 2023@Teknium1 @MosaicML @sfcompute @weights_biases @perplexity_ai @Voxel51 @SFResearch @GoogleDeepMind @neuralmagic @nomic_ai @databricks @huggingface 😻…