CoolFace
Modelpublic

aakarsh-nair/Baby-Llama-360M-Seq-3

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes9downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

Baby-Llama-360M-Seq-3

This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset. It achieves the following results on the evaluation set:

  • —Loss: 1.7242

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • —learning_rate: 0.00025
  • —trainbatchsize: 64
  • —evalbatchsize: 8
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: cosine
  • —lrschedulerwarmup_steps: 200
  • —num_epochs: 10
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation Loss
4.27581.020654.7508
2.66662.041302.9568
2.27493.061952.5908
2.04084.082602.2957
1.75275.0103252.0542
1.66456.0123901.9311
1.52747.0144551.7980
1.47828.0165201.7464
1.42499.0185851.7283
1.457410.0206501.7242

Framework versions

  • —Transformers 4.45.0.dev0
  • —Pytorch 2.4.0+cu121
  • —Tokenizers 0.19.1