4437ebde2cc659243b8f15959e18d0ae

This model is a fine-tuned version of facebook/opt-1.3b on the nyu-mll/glue [stsb] dataset. It achieves the following results on the evaluation set:

• Loss: 2.8034
• Data Size: 1.0
• Epoch Runtime: 34.0954
• Mse: 2.8042
• Mae: 1.3619
• R2: -0.2544

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	8.6415	0	2.9753	8.6425	2.5223	-2.8661
No log	1	179	8.1322	0.0078	3.3666	8.1325	2.4524	-2.6379
No log	2	358	2.9904	0.0156	5.5666	2.9910	1.4142	-0.3380
No log	3	537	2.2163	0.0312	7.5398	2.2167	1.2123	0.0084
No log	4	716	1.4098	0.0625	9.7154	1.4102	0.9772	0.3692
No log	5	895	1.2873	0.125	12.7082	1.2874	0.9373	0.4241
0.1264	6	1074	2.2354	0.25	18.2193	2.2362	1.2843	-0.0003
2.3191	7	1253	2.2736	0.5	27.0946	2.2745	1.2974	-0.0174
2.0585	8.0	1432	2.1997	1.0	48.3184	2.2005	1.2545	0.0156
1.9825	9.0	1611	2.8034	1.0	34.0954	2.8042	1.3619	-0.2544

Framework versions

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