amBART_261

This model is a fine-tuned version of Samuael/amBART_1000 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.9604
• Wer: 2.7857
• Cer: 3.6889
• Bleu: 0.0
• Lr: 0.02

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: 0.02
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Bleu	Lr
No log	1.0	1	3.9328	1.0	4.6333	0.0	0.02
No log	2.0	2	4.1008	1.0	6.1778	0.0	0.02
No log	3.0	3	3.8971	1.0714	3.7556	0.0	0.02
No log	4.0	4	3.5169	1.5714	6.2889	0.0	0.02
No log	5.0	5	3.4597	10.0714	6.1889	0.0	0.02
No log	6.0	6	3.4714	1.0	6.3222	0.0	0.02
No log	7.0	7	3.1601	1.0	6.0667	0.0	0.02
No log	8.0	8	2.5631	1.0	0.7667	0.0	0.02
No log	9.0	9	2.6357	2.0	6.3667	0.0	0.02
No log	10.0	10	3.1707	2.3571	6.5111	0.0	0.02
No log	11.0	11	2.9462	1.1429	0.7	0.0	0.02
No log	12.0	12	3.0437	1.0	6.2111	0.0	0.02
No log	13.0	13	2.6371	19.2143	8.8667	0.0	0.02
No log	14.0	14	2.4126	7.7143	7.1	0.0	0.02
No log	15.0	15	2.6156	19.1429	6.1	0.0	0.02
No log	16.0	16	2.7927	19.5714	6.1778	0.0	0.02
No log	17.0	17	2.6685	1.0	3.3333	0.0	0.02
No log	18.0	18	2.9460	1.0	0.8111	0.0	0.02
No log	19.0	19	3.3183	1.0714	3.4556	0.0	0.02
No log	20.0	20	3.7492	1.2143	3.5222	0.0	0.02
No log	21.0	21	3.8371	9.1429	6.6111	0.0	0.02
No log	22.0	22	3.7951	13.9286	6.3333	0.0	0.02
No log	23.0	23	3.4253	12.0714	6.1556	0.0	0.02
No log	24.0	24	3.4148	1.0714	0.7333	0.0	0.02
No log	25.0	25	3.0110	8.7143	5.9889	0.2910	0.02
No log	26.0	26	2.7432	1.0	1.1444	0.0	0.02
No log	27.0	27	2.5661	1.4286	0.9333	0.0	0.02
No log	28.0	28	2.6703	1.0	3.4889	0.0	0.02
No log	29.0	29	2.9169	18.7143	6.1111	0.0	0.02
No log	30.0	30	3.1300	4.0	4.3667	0.0	0.02
No log	31.0	31	3.2927	6.0	5.6222	0.0	0.02
No log	32.0	32	3.0442	6.5714	6.0444	0.0	0.02
No log	33.0	33	2.7768	1.7143	3.5222	0.0	0.02
No log	34.0	34	2.6387	1.2857	3.4778	0.0	0.02
No log	35.0	35	2.4790	1.2143	3.4444	0.0	0.02
No log	36.0	36	2.3595	5.9286	4.8111	0.0	0.02
No log	37.0	37	2.2934	7.6429	5.3	0.0	0.02
No log	38.0	38	2.2778	1.6429	3.7556	1.6467	0.02
No log	39.0	39	2.2839	6.0714	4.7333	0.0	0.02
No log	40.0	40	2.2559	1.2857	0.8111	0.0	0.02
No log	41.0	41	2.2032	2.5714	4.2333	0.0	0.02
No log	42.0	42	2.1507	1.1429	3.4444	0.0	0.02
No log	43.0	43	2.1281	1.0	0.7556	0.0	0.02
No log	44.0	44	2.1175	1.5714	3.4556	0.0	0.02
No log	45.0	45	2.0781	4.5714	4.3444	0.5569	0.02
No log	46.0	46	2.0383	1.4286	3.3889	1.8161	0.02
No log	47.0	47	2.0069	1.4286	3.3889	1.8161	0.02
No log	48.0	48	1.9878	1.3571	3.3667	0.0	0.02
No log	49.0	49	1.9714	3.6429	3.9556	0.0	0.02
No log	50.0	50	1.9604	2.7857	3.6889	0.0	0.02

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2