salami_truncsplit_legit1_model

This model is a fine-tuned version of facebook/musicgen-small on the TAUFIQSYED/SALAMI_CLEANED_SAMPLED - DEFAULT dataset. It achieves the following results on the evaluation set:

• Loss: 6.5173
• Clap: 0.1031

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: 1
• eval_batch_size: 1
• seed: 456
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.99) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 10.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Clap
8.305	0.4813	25	3.7163	0.2508
6.9842	0.9627	50	4.6081	0.0551
6.9013	1.4428	75	5.6498	0.0582
6.0195	1.9242	100	5.7631	0.1450
6.447	2.4043	125	5.7947	0.1187
6.215	2.8857	150	5.8396	0.1164
6.4075	3.3658	175	5.8361	0.1174
5.797	3.8472	200	5.8150	0.1093
6.347	4.3273	225	5.7935	0.1029
5.3935	4.8087	250	5.7605	0.1040
6.0328	5.2888	275	5.7012	0.1006
6.1476	5.7702	300	5.6703	0.0880
5.9049	6.2503	325	5.8333	0.1260
5.8547	6.7316	350	5.9171	0.1415
5.5151	7.2118	375	6.0307	0.1488
6.0651	7.6931	400	6.0683	0.1520
6.0339	8.1733	425	6.1653	0.1496
5.5607	8.6546	450	6.3588	0.1344
6.0312	9.1348	475	6.4503	0.1358
5.9084	9.6161	500	6.5205	0.1081

Framework versions

• PEFT 0.13.2
• Transformers 4.47.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3