jinx2321/byt5-base-ko-1e4-jit-20epochs-a100-jamoparams-wu2000-rm1-bs2-ga64-eb128-distilled-base-5

<!-- 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. -->

byt5-base-ko-1e4-jit-20epochs-a100-jamoparams-wu2000-rm1-bs2-ga64-eb128-distilled-base-5

This model is a fine-tuned version of google/byt5-small on an unknown dataset. It achieves the following results on the evaluation set:

• —epoch: 30.0
• —eval_loss: 1.0565
• —eval_runtime: 44.7587
• —evalsamplesper_second: 111.71
• —evalstepsper_second: 27.928
• —step: 37590

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: 0.0005
• —trainbatchsize: 2
• —evalbatchsize: 4
• —seed: 42
• —gradientaccumulationsteps: 64
• —totaltrainbatch_size: 128
• —optimizer: Use adamwtorch with betas=(0.9,0.98) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
• —lrschedulertype: inverse_sqrt
• —lrschedulerwarmup_steps: 2000
• —num_epochs: 40
• —labelsmoothingfactor: 0.1

Framework versions

• —Transformers 4.57.3
• —Pytorch 2.9.1+cu128
• —Datasets 4.4.1
• —Tokenizers 0.22.1