Edit model card

Built with Axolotl

out

This model is a fine-tuned version of cyberagent/calm2-7b-chat on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 3.2941

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: 1
  • eval_batch_size: 1
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 3
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 6
  • total_eval_batch_size: 3
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss
3.5899 0.0 1 4.1781
3.4689 0.05 352 3.5988
3.519 0.1 704 3.6209
3.4168 0.15 1056 3.6161
3.4906 0.2 1408 3.6054
3.4853 0.25 1760 3.5950
3.475 0.3 2112 3.5806
3.2476 0.35 2464 3.5675
3.2339 0.4 2816 3.5398
3.3711 0.45 3168 3.5233
3.4195 0.5 3520 3.5006
3.1105 0.55 3872 3.4804
3.2477 0.6 4224 3.4546
3.2753 0.65 4576 3.4311
3.1867 0.7 4928 3.4054
3.2009 0.75 5280 3.3783
3.1037 0.8 5632 3.3509
3.0394 0.85 5984 3.3314
3.0759 0.9 6336 3.3095
3.1989 0.95 6688 3.2941

Framework versions

  • Transformers 4.34.1
  • Pytorch 2.0.1+cu118
  • Datasets 2.14.6
  • Tokenizers 0.14.1
Downloads last month
2

Finetuned from