d7469359a517872261ba02a069e1aaf1

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

• Loss: 2.6111
• Data Size: 1.0
• Epoch Runtime: 45.9059
• Mse: 0.6531
• Mae: 0.6440
• R2: 0.7079

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	Mse	Mae	R2
No log	0	0	77.2887	0	3.0233	19.3237	4.0456	-7.6442
No log	1	179	17.6108	0.0078	3.1154	4.4033	1.6571	-0.9697
No log	2	358	6.9542	0.0156	7.7106	1.7390	1.0777	0.2221
No log	3	537	3.1502	0.0312	11.7111	0.7877	0.7103	0.6476
No log	4	716	5.4675	0.0625	13.7078	1.3669	0.9229	0.3885
No log	5	895	3.6613	0.125	16.3696	0.9155	0.7185	0.5905
0.4371	6	1074	3.2607	0.25	20.6868	0.8153	0.7180	0.6353
3.4473	7	1253	2.5354	0.5	28.2817	0.6341	0.6363	0.7164
2.9908	8.0	1432	2.4045	1.0	45.5716	0.6012	0.6171	0.7310
1.9529	9.0	1611	2.6714	1.0	43.3237	0.6680	0.6494	0.7012
1.3749	10.0	1790	2.6716	1.0	46.1322	0.6680	0.6629	0.7012
1.4787	11.0	1969	2.4727	1.0	45.7220	0.6184	0.6180	0.7234
0.8118	12.0	2148	2.6111	1.0	45.9059	0.6531	0.6440	0.7079

Framework versions

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