a39387657b0f0048084516e0a839ebe4

This model is a fine-tuned version of albert/albert-xxlarge-v1 on the contemmcm/cls_20newsgroups dataset. It achieves the following results on the evaluation set:

• Loss: 0.4551
• Data Size: 1.0
• Epoch Runtime: 172.6730
• Accuracy: 0.8969
• F1 Macro: 0.8970

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	3.6239	0	12.8569	0.0572	0.0223
No log	1	499	3.2262	0.0078	14.4292	0.0673	0.0452
0.0342	2	998	3.0433	0.0156	15.6355	0.0837	0.0493
0.0569	3	1497	2.8048	0.0312	18.3091	0.1391	0.1006
0.0977	4	1996	1.1012	0.0625	23.4193	0.6154	0.5845
0.9091	5	2495	0.7378	0.125	33.3894	0.7616	0.7626
0.5309	6	2994	0.5493	0.25	53.2089	0.8286	0.8234
0.4167	7	3493	0.5161	0.5	93.1672	0.8427	0.8445
0.3747	8.0	3992	0.4342	1.0	173.1241	0.8664	0.8683
0.2874	9.0	4491	0.3747	1.0	172.7754	0.8858	0.8863
0.1809	10.0	4990	0.4115	1.0	172.8040	0.8873	0.8851
0.153	11.0	5489	0.4343	1.0	172.8686	0.8909	0.8910
0.1285	12.0	5988	0.4603	1.0	172.7868	0.8962	0.8966
0.101	13.0	6487	0.4551	1.0	172.6730	0.8969	0.8970

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.1