Axolotl Training Config

All settings required to update a training are shown by default.

Model Pick the base model you want to do your fine tuning on top of. View models of HF

Training Type

GPU

GPU Quantity

Datasets

Dataset Path The dataset you want to train on. A dataset name from HuggingFace, or a relative path.

Features:

Prompt Type The prompt format the dataset will use. See Axolotl Docs for more information.

Epoch(s) The number of times the AI model will read through the dataset.

Hyperparameters

Sequence Length

Microbatch Size

Learning Rate

Gradient Accum

LR Scheduler

Optimizer

Weight Decay

Max Grad Norm

Adam Beta 2

Adam Beta 1

Weights & Bias

LoRA Config

Optimization

Checkpoints

All Settings