metadata

tags:
  - generated_from_trainer
datasets:
  - AlekseyKorshuk/dalio-all-io
metrics:
  - accuracy
model-index:
  - name: 1.3b-all-2-epoch-v1-after-book
    results:
      - task:
          name: Causal Language Modeling
          type: text-generation
        dataset:
          name: AlekseyKorshuk/dalio-all-io
          type: AlekseyKorshuk/dalio-all-io
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.06395348837209303

1.3b-all-2-epoch-v1-after-book

This model is a fine-tuned version of /models/1.3b-dalio-principles-book on the AlekseyKorshuk/dalio-all-io dataset. It achieves the following results on the evaluation set:

Loss: 1.9482
Accuracy: 0.0640

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: 3e-05
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: cosine
num_epochs: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.17	0.07	1	2.0547	0.0621
2.1814	0.13	2	2.0547	0.0621
2.0963	0.2	3	2.0234	0.0625
2.1383	0.27	4	2.0195	0.0625
2.1625	0.33	5	2.0195	0.0625
2.1808	0.4	6	2.0156	0.0624
2.1587	0.47	7	2.0176	0.0626
2.0847	0.53	8	2.0137	0.0627
2.0336	0.6	9	2.0137	0.0627
2.1777	0.67	10	2.0059	0.0629
2.2034	0.73	11	2.0	0.0630
2.1665	0.8	12	1.9941	0.0628
2.0352	0.87	13	1.9883	0.0629
2.1263	0.93	14	1.9834	0.0628
2.1282	1.0	15	1.9785	0.0632
1.7159	1.07	16	1.9766	0.0633
1.8346	1.13	17	1.9775	0.0635
1.7183	1.2	18	1.9824	0.0634
1.6086	1.27	19	1.9883	0.0635
1.6497	1.33	20	1.9893	0.0634
1.6267	1.4	21	1.9854	0.0637
1.5962	1.47	22	1.9766	0.0637
1.5168	1.53	23	1.9697	0.0637
1.6213	1.6	24	1.9619	0.0637
1.4789	1.67	25	1.9580	0.0638
1.6796	1.73	26	1.9551	0.0638
1.5964	1.8	27	1.9531	0.0638
1.787	1.87	28	1.9512	0.0639
1.6536	1.93	29	1.9492	0.0640
1.7178	2.0	30	1.9482	0.0640

Framework versions

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