Fine-tuning Datasets with Synthetic Inputs

Показать описание

There are virtually unlimited ways to fine-tune LLMs to improve performance at specific tasks... but where do you get the data from?

In this video, I demonstrate one way that you can fine-tune without much data to start with — and use what little data you have to reverse-engineer the inputs required!

I show step-by-step how to take a small set of data (for my example I use 20 press releases I pulled from the internet), use LLMs to generate the missing inputs, run a real fine-tuning job, and play with the model to see how it behaves.

The actual fine-tuning cost $0.35. Turns out, fine-tuning can be pretty affordable!

Creating a dataset is half art and half science, but there is nothing particularly hard about it once you understand the core concepts.

You don't need an ML degree to build your own fine-tuning dataset and train a custom LLM. You just need to be able to think critically about what you want the model to do and how you want to steer the model through inputs.

Follow along as I share my thought process and tools to create fine-tuning datasets and see how decisions along the way affect our interaction with the model and its outputs.

Remember to press subscribe! ✅

You can try fine-tuning for yourself at:

My LinkedIn:

Entry Point AI Discord:

Рекомендации по теме

Комментарии

Can you make a video about tuning validation? How to reliably test that tuning gives us the results we're looking for?

dagryl

Great explanation and a very advanced web service—well thought out and executed! From your previous video, I learned that fine-tuning is more like shaping intuition, since the model operates on statistical patterns and isn't really "learning" in the human sense. With that in mind, I have a question about the capabilities of fine-tuning: Can it be used to impart factual knowledge? For example, if I input data from a medical textbook, could the model be trained to make medical diagnoses in a way that approximates learning? (I used chatgpt to make this comment more clear ;) )

gileneusz

Website app get started is loading for long time for me I am not able to register

yotubecreators

Fine-tuning Datasets with Synthetic Inputs

Fine-tuning Datasets with Synthetic Inputs

Prepare Fine-tuning Datasets with Open Source LLMs

How to Create Custom Datasets To Train Llama-2

How to Create Synthetic Dataset EASILY? Step by Step Tutorial

Tutorial 2- Fine Tuning Pretrained Model On Custom Dataset Using 🤗 Transformer

Q: How to create an Instruction Dataset for Fine-tuning my LLM?

How to Create High Quality Synthetic Data for Fine-Tuning LLMs

How To Create Datasets for Finetuning From Multiple Sources! Improving Finetunes With Embeddings.

#220 Large Language Models are Human-like Annotators. KR 2024 tutorial Part 4

How to Fine-Tune and Train LLMs With Your Own Data EASILY and FAST- GPT-LLM-Trainer

Create Self-Instruct Data Sets: for Synthetic Self-Instruct (ChatGPT) fine-tuning of LLMs

Step-by-Step Guide: Fine-Tuning GPT-4o Mini with Synthetic Data.Dataset can be created manually also

Generate Synthetic Data in 60 seconds | Gretel.ai

Fine-Tuning ChatGPT 3.5 with Synthetic Data from GPT-4 | VERY Interesting Results (!)

Convert Any Text to LLM Dataset Locally - Demo with Example

ChatGPT 3.5 Turbo Fine Tuning For Specific Tasks - Tutorial with Synthetic Data

Using ChatGPT with YOUR OWN Data. This is magical. (LangChain OpenAI API)

How to make a custom dataset like Alpaca7B

Building with Instruction-Tuned LLMs: A Step-by-Step Guide

Generating Synthetic Datasets with GitHub Copilot

Synthetic Data Generation and Fine tuning (OpenAI GPT4o or Llama 3)

What is Prompt Tuning?

Code to Fine-tune ChatGPT w/ synthetic GPT-4 dataset

Fine-Tuning GPT-3.5 on Custom Dataset: A Step-by-Step Guide | Code