TED_CLM_gpt2_tedlium2

This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8951
• Accuracy: 0.5484

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: 0.001
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 512
• total_eval_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 20000
• num_epochs: 15.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1528	0.62	3000	2.3612	0.4554
1.9797	1.24	6000	2.2025	0.4923
1.9185	1.86	9000	2.1189	0.5089
1.8816	2.49	12000	2.0701	0.5168
1.8524	3.11	15000	2.0396	0.5232
1.8388	3.73	18000	2.0113	0.5259
1.8223	4.35	21000	1.9935	0.5307
1.8051	4.97	24000	1.9729	0.5332
1.7875	5.59	27000	1.9588	0.5344
1.7709	6.22	30000	1.9441	0.5369
1.7643	6.84	33000	1.9461	0.5381
1.7567	7.46	36000	1.9471	0.5395
1.7482	8.08	39000	1.9307	0.5412
1.7399	8.7	42000	1.9308	0.5415
1.7303	9.32	45000	1.9262	0.5419
1.7353	9.94	48000	1.9257	0.5434
1.7267	10.57	51000	1.9181	0.5453
1.719	11.19	54000	1.9204	0.5443
1.716	11.81	57000	1.9073	0.5473
1.7072	12.43	60000	1.9051	0.5465
1.7093	13.05	63000	1.9001	0.5490
1.708	13.67	66000	1.8998	0.5483
1.6972	14.29	69000	1.8968	0.5489
1.7001	14.92	72000	1.8951	0.5484

Framework versions

• Transformers 4.31.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.13.1
• Tokenizers 0.13.3