I’m at the AI Engineer fair - hit me up if you’re around, let’s meet :)

Pleasantly surprised to see that our old Mistral AI 7B-based OpenChat model from January is the most popular generalist LLM fine-tune released by the open source on OpenRouter 🫶

Every day I wait patiently for them to drop the paper for the actual details on their new fine-tuning data generation pipeline… please WizardLM we need it

Synthetic data yall..

Planning to build a centralized library for synthetic data generation for LLM training(sft first, then pt, rlhf, etc). Will allow you to build pipelines that combine and chain multiple data generation methods together, and modify them individually. Will be housing both

Similar performance to DoReMi while using 10x less compute 👀

New Orca paper, data generation pipeline looks quite promising for performance and quality in real world usage Microsoft is greedy with datasets, so it likely won’t be released, but please share the generation prompts at least, so the pipeline can be accurately reproduced 🙏🏼

This

Can we please get more details AI at Meta

I have an exciting update - I’ve recently joined Together AI Together AI as a Research Scientist to lead Post-Training! Having the time of my life :)