Modelpublic

gokuls/model_v1_complete_training_wt_init_48_small_emb_comp_frz

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes7downloads

Model Card

modelv1completetrainingwtinit48smallembcompfrz

This model is a fine-tuned version of [](https://huggingface.co/) on the None dataset. It achieves the following results on the evaluation set:

—Loss: 4.0873
—Accuracy: 0.3585

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: 1e-05
—trainbatchsize: 48
—evalbatchsize: 48
—seed: 10
—distributed_type: multi-GPU
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 10000
—num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
6.1895	0.25	30000	6.1896	0.1440
6.142	0.49	60000	6.1411	0.1464
6.1111	0.74	90000	6.1027	0.1482
6.0564	0.98	120000	6.0542	0.1499
6.0105	1.23	150000	5.9999	0.1525
5.9542	1.47	180000	5.9441	0.1555
5.8828	1.72	210000	5.8687	0.1602
5.8099	1.97	240000	5.7963	0.1680
5.7234	2.21	270000	5.6868	0.1851
5.4249	2.46	300000	5.3582	0.2278
5.0172	2.7	330000	4.9641	0.2721
4.7387	2.95	360000	4.6970	0.2991
4.5625	3.2	390000	4.5123	0.3164
4.4301	3.44	420000	4.3802	0.3295
4.3369	3.69	450000	4.2827	0.3389
4.2632	3.93	480000	4.2126	0.3459
4.2074	4.18	510000	4.1586	0.3516
4.1713	4.42	540000	4.1218	0.3551
4.1448	4.67	570000	4.0984	0.3573
4.1356	4.92	600000	4.0873	0.3585

Framework versions

—Transformers 4.30.2
—Pytorch 1.14.0a0+410ce96
—Datasets 2.13.1
—Tokenizers 0.13.3