ALBERT_Emotions_tuned

This model is a fine-tuned version of albert-base-v2 on the emotion dataset. It achieves the following results on the evaluation set:

• Loss: 0.1846
• Accuracy: 0.927

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.1	100	1.2655	0.5575
No log	0.2	200	1.0801	0.6415
No log	0.3	300	1.0138	0.661
No log	0.4	400	1.0651	0.605
1.1328	0.5	500	0.7816	0.758
1.1328	0.6	600	0.6307	0.7885
1.1328	0.7	700	0.5072	0.848
1.1328	0.8	800	0.5057	0.8305
1.1328	0.9	900	0.3853	0.889
0.5503	1.0	1000	0.3769	0.8915
0.5503	1.1	1100	0.3778	0.8995
0.5503	1.2	1200	0.3899	0.9005
0.5503	1.3	1300	0.3330	0.9085
0.5503	1.4	1400	0.3339	0.9085
0.3049	1.5	1500	0.2662	0.915
0.3049	1.6	1600	0.3209	0.9045
0.3049	1.7	1700	0.3110	0.898
0.3049	1.8	1800	0.3185	0.9075
0.3049	1.9	1900	0.2439	0.922
0.2485	2.0	2000	0.2190	0.925
0.2485	2.1	2100	0.2372	0.9235
0.2485	2.2	2200	0.2497	0.9265
0.2485	2.3	2300	0.2811	0.9195
0.2485	2.4	2400	0.2350	0.9195
0.1587	2.5	2500	0.2303	0.9245
0.1587	2.6	2600	0.2242	0.9285
0.1587	2.7	2700	0.2141	0.9325
0.1587	2.8	2800	0.2185	0.9315
0.1587	2.9	2900	0.2047	0.9315
0.1398	3.0	3000	0.2036	0.9335

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2