Fine Tune Tiny Llama on Custom Dataset - For Free With Google Colab
In this video, we will fine-tune Tiny Llama and create a custom dataset using already existing data from Hugging Face. We will understand what LoRA is and merge the model we created, test it, and get final results using only Google Colab and no additional cost.
Support Me on Petron ⚡: / codewello
Code:https://colab.research.google.com/dri...
Tiny Llama Github :https://github.com/jzhang38/TinyLlama
Paper :https://arxiv.org/abs/2401.02385
LoRa Artcile:https://www.datacamp.com/tutorial/mas...
00:00 - Intro
00:08 - What are we going to do?
01:08 - What is Tiny Llama
02:27 - Test Tiny Llama
03:07 - Why Fine Tune a Tiny Model
03:55 - Code Setup
05:13 - Dataset Format
08:42 - Downloading the model
09:37 - LoRA
11:19 - Training Parameters
13:08 - Training The Model
13:51 - Merge The Models
14:47 - Results & Testing
17:05 - Pushing to Hugging Face
17:31 - Outro
_______________________________________________________________
🤖 Open AI Template Alima : https://codewello.gumroad.com/l/kjlzw
💷 50% Discount Code : A2LH6LZ
👉 Linkedin : / hossam-eldin
_______________________________________________________________
📕 Mastering Yolo V8 Object Detection : • Mastering YOLO V8: Object Detection a...
📙 Supervision : • Supervision with Yolo V8 : Video Trac...
📗 Mastering Yolo NAS : • Mastering YOLO NAS: Object Detection
🧑🏻💻 Build Ai Assistant : • OpenAI Assistant: From Zero to Deploy...
#finetune #llms #opensource #llama2 #gpt #falcon #ai #tinyllama