Fine-tune LLMs 30x faster! With Daniel Han (Unsloth AI)
Become a Patreon: https://www.patreon.com/theaiepiphany π¨βπ©βπ§βπ¦ Join our Discord community: https://discord.gg/peBrCpheKE Daniel Han from Unsloth AI joined us to walk us through his bug-hunting process and how to make LLM fine-tuning 30x faster. :) β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬ check out unsloth ai: https://unsloth.ai/ β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬ βοΈ Timetable: 00:00:00 - 00:01:40 Intro 00:01:40 - 00:03:21 Hyperstack GPUs! (sponsored) 00:03:21 - 00:06:30 About Daniel, getting started with ML 00:06:30 - 00:11:00 Using Desmos to check grads 00:11:00 - 00:38:00 Deep dive into Gemma bugs 00:38:00 - 00:50:00 approximate GELU bug 00:50:00 - 00:54:15 What are the bottlenecks to speeding up the runtime? 00:54:15 - 00:58:30 comparison with llm.c? 00:58:30 - 01:00:00 Is multi-GPU coming to Unsloth? :) 01:00:00 - 01:05:19 Reproducibility in ML research β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬ π° SPONSOR The AI Epiphany - https://www.patreon.com/theaiepiphany One-time donation - https://www.paypal.com/paypalme/theaiepiphany Huge thank you to these AI Epiphany patreons: Eli Mahler Petar VeliΔkoviΔ β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬ πΌ LinkedIn - https://www.linkedin.com/in/aleksagordic/ π¦ Twitter - https://twitter.com/gordic_aleksa π¨βπ©βπ§βπ¦ Discord - https://discord.gg/peBrCpheKE πΊ YouTube - https://www.youtube.com/c/TheAIEpiphany/ π Medium - https://gordicaleksa.medium.com/ π» GitHub - https://github.com/gordicaleksa π’ AI Newsletter - https://aiepiphany.substack.com/ β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬β¬ #unslothai #llms #finetuning #ml
