5c5928be1e42ec1cb23eb7f121f00e55

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

• Loss: 1.6779
• Data Size: 1.0
• Epoch Runtime: 2020.1928
• Accuracy: 0.8699
• F1 Macro: 0.8616
• Rouge1: 0.8699
• Rouge2: 0.0
• Rougel: 0.8699
• Rougelsum: 0.8699

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	3.6817	0	53.4209	0.4274	0.4273	0.4274	0.0	0.4275	0.4273
2.6414	1	11370	2.0358	0.0078	68.4744	0.7414	0.7408	0.7417	0.0	0.7415	0.7415
1.9902	2	22740	2.2460	0.0156	87.0030	0.8102	0.7970	0.8102	0.0	0.8102	0.8102
1.725	3	34110	1.5377	0.0312	120.3892	0.8226	0.8108	0.8226	0.0	0.8225	0.8226
1.5469	4	45480	1.6456	0.0625	180.5249	0.8225	0.8051	0.8224	0.0	0.8225	0.8225
1.556	5	56850	1.5711	0.125	306.6972	0.8185	0.8061	0.8184	0.0	0.8185	0.8185
1.4024	6	68220	1.4284	0.25	551.1183	0.8351	0.8269	0.8350	0.0	0.8351	0.8351
1.349	7	79590	1.3474	0.5	1053.8951	0.8483	0.8371	0.8482	0.0	0.8482	0.8482
1.2877	8.0	90960	1.3120	1.0	2052.8625	0.8584	0.8466	0.8584	0.0	0.8585	0.8583
1.033	9.0	102330	1.2557	1.0	2046.0599	0.8600	0.8506	0.8600	0.0	0.8601	0.8600
0.8222	10.0	113700	1.2807	1.0	2046.0199	0.8749	0.8662	0.8750	0.0	0.8749	0.8750
0.6997	11.0	125070	1.3881	1.0	2023.4633	0.8718	0.8629	0.8717	0.0	0.8718	0.8718
0.4996	12.0	136440	1.4219	1.0	2023.2068	0.8687	0.8628	0.8687	0.0	0.8688	0.8687
0.523	13.0	147810	1.6779	1.0	2020.1928	0.8699	0.8616	0.8699	0.0	0.8699	0.8699

Framework versions

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