groove_midi_3

This model is a fine-tuned version of JannikAhlers/groove_midi_2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3215
• Bleu: 0.0
• Gen Len: 20.0

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	57	0.3367	0.0	20.0
No log	2.0	114	0.3348	0.0	20.0
No log	3.0	171	0.3342	0.0	20.0
No log	4.0	228	0.3322	0.0	20.0
No log	5.0	285	0.3311	0.0	20.0
No log	6.0	342	0.3305	0.0	20.0
No log	7.0	399	0.3296	0.0	20.0
No log	8.0	456	0.3284	0.0	20.0
0.3803	9.0	513	0.3276	0.0	20.0
0.3803	10.0	570	0.3273	0.0	20.0
0.3803	11.0	627	0.3267	0.0	20.0
0.3803	12.0	684	0.3259	0.0	20.0
0.3803	13.0	741	0.3258	0.0	20.0
0.3803	14.0	798	0.3250	0.0	20.0
0.3803	15.0	855	0.3250	0.0	20.0
0.3803	16.0	912	0.3243	0.0	20.0
0.3803	17.0	969	0.3237	0.0	20.0
0.3692	18.0	1026	0.3234	0.0	20.0
0.3692	19.0	1083	0.3232	0.0	20.0
0.3692	20.0	1140	0.3228	0.0	20.0
0.3692	21.0	1197	0.3228	0.0	20.0
0.3692	22.0	1254	0.3228	0.0	20.0
0.3692	23.0	1311	0.3223	0.0	20.0
0.3692	24.0	1368	0.3219	0.0	20.0
0.3692	25.0	1425	0.3218	0.0	20.0
0.3692	26.0	1482	0.3217	0.0	20.0
0.3642	27.0	1539	0.3216	0.0	20.0
0.3642	28.0	1596	0.3216	0.0	20.0
0.3642	29.0	1653	0.3215	0.0	20.0
0.3642	30.0	1710	0.3215	0.0	20.0

Framework versions

• Transformers 4.48.2
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0