Skip to content
This repository has been archived by the owner on Apr 11, 2024. It is now read-only.

Latest commit

 

History

History
53 lines (29 loc) · 1.51 KB

DOCS.md

File metadata and controls

53 lines (29 loc) · 1.51 KB

Build your own project using X—LLM

WORK IN PROGRESS

How to implement dataset

Берешь и делаешь

How to add CLI tools to your project

Требования к формату выхода примеров

1. Download

First off, you must download the data and the model. This step also handles data preparation. You have to complete this necessary step because it sets up how your data will be downloaded and prepared.

Before you start the training, it's crucial to get the data and the model ready. You won't be able to start training without them since this is the point where your data gets prepped for training. We made this a separate step for good reasons. For example, if you're training across multiple GPUs, like with DeepSpeed, you'd otherwise end up downloading the same data and models on each GPU, when you only need to do it once.

1.1 Implement your dataset

1.2 Register your dataset

1.3 Run the downloading

2. Train

3. Fuse

Этот шаг нужен только в том случае, если вы обучали модель с использованием LoRA, что является рекомендуемым решением.

4. GPTQ Quantization

Это опциональный шаг

Config, все поля

Important config keys for different steps

How do I choose the methods for training?

Reduce the size of the model during training

Single GPU

Multiple GPUs

Дополнительные комментарии

FAQ