6.7b-dalio-principles-book-1-epoch-1-gas-6e-6-lr

This model is a fine-tuned version of facebook/opt-6.7b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.4121
• Accuracy: 0.3487

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: 6e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.4875	0.11	1	2.5059	0.3397
2.5339	0.22	2	2.5059	0.3397
2.5161	0.33	3	2.5059	0.3397
2.4524	0.44	4	2.5059	0.3397
2.554	0.56	5	2.4785	0.3416
2.4678	0.67	6	2.4785	0.3416
2.4836	0.78	7	2.4473	0.3458
2.4138	0.89	8	2.4297	0.3473
2.4551	1.0	9	2.4121	0.3487

Framework versions

• Transformers 4.25.0.dev0
• Pytorch 1.12.1+cu113
• Datasets 2.3.2
• Tokenizers 0.12.1