3d198d398eb12d2cae1f048850f3ce02

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the nyu-mll/glue [sst2] dataset. It achieves the following results on the evaluation set:

• Loss: 0.4420
• Data Size: 1.0
• Epoch Runtime: 169.4327
• Accuracy: 0.8808
• F1 Macro: 0.8805
• Rouge1: 0.8819
• Rouge2: 0.0
• Rougel: 0.8808
• Rougelsum: 0.8808

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	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	0.7150	0	1.3373	0.4907	0.3292	0.4907	0.0	0.4907	0.4919
No log	1	2104	0.6903	0.0078	3.0011	0.5093	0.3374	0.5093	0.0	0.5093	0.5081
No log	2	4208	0.6844	0.0156	4.3039	0.5706	0.4696	0.5706	0.0	0.5700	0.5706
0.0143	3	6312	0.5339	0.0312	7.1309	0.7986	0.7986	0.7986	0.0	0.7986	0.7986
0.4887	4	8416	0.4106	0.0625	12.3717	0.8125	0.8108	0.8125	0.0	0.8125	0.8119
0.3756	5	10520	0.3836	0.125	23.3957	0.8275	0.8257	0.8275	0.0	0.8275	0.8275
0.3095	6	12624	0.2919	0.25	45.1061	0.8843	0.8840	0.8843	0.0	0.8843	0.8843
0.2631	7	14728	0.2789	0.5	87.0555	0.8912	0.8912	0.8912	0.0	0.8912	0.8912
0.2217	8.0	16832	0.3098	1.0	169.8256	0.8808	0.8805	0.8808	0.0	0.8808	0.8808
0.1848	9.0	18936	0.3146	1.0	169.9892	0.8958	0.8958	0.8958	0.0	0.8958	0.8958
0.2307	10.0	21040	0.3753	1.0	168.1695	0.8912	0.8912	0.8912	0.0	0.8924	0.8912
0.1575	11.0	23144	0.4420	1.0	169.4327	0.8808	0.8805	0.8819	0.0	0.8808	0.8808

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1