asl_spoken

This model is a fine-tuned version of facebook/bart-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9372
• Bleu: 35.4423
• Gen Len: 9.1538

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 75

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	7	5.2969	0.3351	17.7692
No log	2.0	14	4.5346	0.3351	17.7692
No log	3.0	21	3.6446	0.3384	17.5385
No log	4.0	28	2.8827	0.3417	17.6154
No log	5.0	35	2.4643	0.3485	18.0
No log	6.0	42	2.2510	10.0737	14.0769
No log	7.0	49	2.0346	14.6064	11.6923
No log	8.0	56	1.8385	14.9081	11.0769
No log	9.0	63	1.6640	27.3053	10.8462
No log	10.0	70	1.5089	30.1357	9.6154
No log	11.0	77	1.4685	38.3348	9.3846
No log	12.0	84	1.4620	33.0019	9.0
No log	13.0	91	1.3466	36.2551	9.0769
No log	14.0	98	1.2250	42.8435	9.1538
No log	15.0	105	1.2373	44.1465	9.0
No log	16.0	112	1.2410	44.1465	9.0
No log	17.0	119	1.1940	44.1465	9.0
No log	18.0	126	1.1753	44.1465	9.0
No log	19.0	133	1.1345	48.4528	9.0
No log	20.0	140	1.1141	48.5376	8.9231
No log	21.0	147	1.1210	48.5376	8.9231
No log	22.0	154	1.1298	48.5376	8.9231
No log	23.0	161	1.1185	44.1465	9.0

Framework versions

• PEFT 0.8.2
• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2