SignBart-ASL-1000

This model is a fine-tuned version of tinh2312/SignBart-ASL-1000 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8280
• Accuracy: 0.8145
• Precision: 0.8290
• Recall: 0.8145

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-06
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall
0.9268	1.0	96	0.8296	0.8147	0.8295	0.8147
0.8969	2.0	192	0.8287	0.8159	0.8304	0.8159
0.9075	3.0	288	0.8281	0.8152	0.8297	0.8152
0.8239	4.0	384	0.8282	0.8145	0.8295	0.8145
0.9632	5.0	480	0.8283	0.8138	0.8287	0.8138
0.906	6.0	576	0.8278	0.8145	0.8287	0.8145
0.884	7.0	672	0.8282	0.8152	0.8304	0.8152
1.0349	8.0	768	0.8282	0.8141	0.8289	0.8141
0.8927	9.0	864	0.8280	0.8145	0.8289	0.8145
0.92	10.0	960	0.8279	0.8147	0.8292	0.8147
0.9897	11.0	1056	0.8280	0.8145	0.8290	0.8145

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0