fc1c884c2af6bdedbfe7ea88893ba8e9

This model is a fine-tuned version of google/gemma-2b on the nyu-mll/glue [mrpc] dataset. It achieves the following results on the evaluation set:

• Loss: 7.6709
• Data Size: 1.0
• Epoch Runtime: 28.3061
• Accuracy: 0.7783
• F1 Macro: 0.7153
• Rouge1: 0.7783
• Rouge2: 0.0
• Rougel: 0.7777
• Rougelsum: 0.7783

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	2.7643	0	3.3412	0.6126	0.5522	0.6132	0.0	0.6126	0.6120
No log	1	114	5.1579	0.0078	3.9321	0.6592	0.4637	0.6592	0.0	0.6586	0.6592
No log	2	228	3.6322	0.0156	8.1285	0.6692	0.4777	0.6692	0.0	0.6692	0.6692
No log	3	342	4.6142	0.0312	12.6551	0.6663	0.4032	0.6669	0.0	0.6660	0.6657
0.1148	4	456	3.1486	0.0625	15.1708	0.7028	0.5735	0.7028	0.0	0.7019	0.7034
0.1148	5	570	2.5100	0.125	19.7567	0.6686	0.4762	0.6692	0.0	0.6680	0.6686
0.1148	6	684	3.2711	0.25	21.6517	0.4593	0.4429	0.4587	0.0	0.4593	0.4599
0.6767	7	798	2.1242	0.5	28.5181	0.7529	0.7261	0.7524	0.0	0.7524	0.7535
1.7658	8.0	912	1.9603	1.0	41.3399	0.7647	0.6914	0.7647	0.0	0.7642	0.7647
1.0099	9.0	1026	3.0099	1.0	41.6100	0.7606	0.7253	0.7606	0.0	0.7606	0.7612
1.0121	10.0	1140	3.4193	1.0	41.8100	0.7677	0.7235	0.7677	0.0	0.7671	0.7671
0.6692	11.0	1254	3.5542	1.0	41.4199	0.7795	0.7605	0.7789	0.0	0.7789	0.7795
0.5981	12.0	1368	7.6709	1.0	28.3061	0.7783	0.7153	0.7783	0.0	0.7777	0.7783

Framework versions

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