05de2b6ab25f8a8ff9680982d2882176

This model is a fine-tuned version of Qwen/Qwen2.5-0.5B on the nyu-mll/glue [mnli] dataset. It achieves the following results on the evaluation set:

• Loss: 3.4456
• Data Size: 1.0
• Epoch Runtime: 2048.4980
• Accuracy: 0.7526
• F1 Macro: 0.7520
• Rouge1: 0.7525
• Rouge2: 0.0
• Rougel: 0.7528
• Rougelsum: 0.7523

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	22.9275	0	17.6193	0.3298	0.1928	0.3297	0.0	0.3300	0.3299
7.2969	1	12271	3.5431	0.0078	33.1497	0.6026	0.5722	0.6026	0.0	0.6026	0.6026
3.0293	2	24542	3.1737	0.0156	48.5613	0.6597	0.6570	0.6596	0.0	0.6600	0.6596
2.9796	3	36813	2.9851	0.0312	80.6867	0.6840	0.6837	0.6842	0.0	0.6842	0.6841
2.8393	4	49084	2.8221	0.0625	144.7753	0.7067	0.7059	0.7064	0.0	0.7067	0.7069
2.7478	5	61355	2.7532	0.125	269.2243	0.7104	0.7055	0.7102	0.0	0.7105	0.7103
2.6791	6	73626	2.6412	0.25	520.6404	0.7251	0.7245	0.7252	0.0	0.7252	0.7251
2.3633	7	85897	2.4754	0.5	1039.6598	0.7432	0.7418	0.7432	0.0	0.7435	0.7429
2.1631	8.0	98168	2.3229	1.0	2053.4290	0.7636	0.7629	0.7635	0.0	0.7637	0.7637
1.7413	9.0	110439	2.4964	1.0	2058.1769	0.7605	0.7593	0.7605	0.0	0.7606	0.7604
1.3544	10.0	122710	2.6274	1.0	2054.3603	0.7625	0.7605	0.7624	0.0	0.7622	0.7626
1.1371	11.0	134981	3.1603	1.0	2038.6514	0.7554	0.7543	0.7554	0.0	0.7555	0.7554
0.7769	12.0	147252	3.4456	1.0	2048.4980	0.7526	0.7520	0.7525	0.0	0.7528	0.7523

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1