The best I’ve found and consider to be the state of the art. Critique welcomed, but respectfully, you are wrong 🤌
Michael Kirchner PRO
kirch
AI & ML interests
Reinforcement learning, multi-lingual multi-modal models, human-computer interfaces
Organizations
Question & answer, task completion, general SFT and otherwise finetuney data.
The elusive “human” feedback
The best I’ve found and consider to be the state of the art. Critique welcomed, but respectfully, you are wrong 🤌
Run on potato, sir. GGUF & GPTQ are good friends.
We gotta start somewhere, these jsonl's aren't gonna train themselves.
Question & answer, task completion, general SFT and otherwise finetuney data.
Conversations, turn-based dialog, and things that can be turned into that.
The elusive “human” feedback
State of the art for multi-modal models