tian-ft / README.md
Sirawipa's picture
Upload tokenizer
7220b01 verified
metadata
license: apache-2.0
tags:
  - generated_from_trainer
base_model: sail/Sailor-0.5B
model-index:
  - name: tian-ft
    results: []

tian-ft

This model is a fine-tuned version of sail/Sailor-0.5B on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3696

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: 0.0002
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 2
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
2.8288 0.9362 11 5.0945
2.0955 1.9574 23 0.5197
0.3705 2.9787 35 0.3038
0.1986 4.0 47 0.2816
0.1402 4.9362 58 0.2941
0.0884 5.9574 70 0.3380
0.0636 6.9787 82 0.3373
0.0477 8.0 94 0.3413
0.0401 8.9362 105 0.3689
0.0267 9.3617 110 0.3696

Framework versions

  • Transformers 4.41.0
  • Pytorch 2.1.2+cu121
  • Datasets 2.19.0
  • Tokenizers 0.19.1