Tiny Language Models - How to build INSANELY FAST local models | Chapter 2, SLM Training

In this video, we will go through the entire instruction tuning or Supervised Finetuning (SFT) phase. We will take raw unstructured text files, and learn how to generate synthetic dataset with a locally running language model. This will be done with Outlines, a structured output generation library. Next we will take our finetuning dataset finetune with Unsloth. Finally, we will build SDKs and harnesses around our model so they can be run locally. Check out Paper Breakdown: paperbreakdown.com And use code SFT50 at checkout for a 50% discount on Student Tier. Follow me on Twitter: https://x.com/neural_avb To join our Patreon, visit: https://www.patreon.com/NeuralBreakdownwithAVB Members get access to everything behind-the-scenes that goes into producing my videos. Plus, it supports the channel in a big way and helps to pay my bills! Resources: Course Video 1 (CPT): https://youtu.be/B8Ur62D3J3U Course Video 2/this one (SFT): https://youtu.be/gvZIUEL6Ruc Course Video 3 (DPO): https://youtu.be/qQwg2nFtZ1Y Low-level "from scratch" Finetuning tutorial: https://youtu.be/bZcKYiwtw1I Great huggingface article on synthetic datagen: https://huggingface.co/spaces/HuggingFaceFW/finephrase#introduction Repos: Course repo (WIP): https://github.com/avbiswas/finetuning_recipes Text-albumentations: https://github.com/avbiswas/text-albumentations Neural-txt: https://github.com/avbiswas/neural-txt Dataset: https://huggingface.co/datasets/paperbd/paper_instructions_300K-v1