musicgen-medium

This model is a fine-tuned version of facebook/musicgen-medium on the J-11045/INDIAN-MUSIC-WITH-METADATA-2 - DEFAULT dataset. It achieves the following results on the evaluation set:

• Loss: 5.5485
• Clap: -0.0775

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 1
• seed: 456
• gradient_accumulation_steps: 8
• 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: linear
• num_epochs: 10.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Clap
44.9355	0.4739	25	6.4378	-0.0786
38.3323	0.9479	50	5.7564	-0.0774
37.6585	1.4265	75	5.9326	-0.0850
35.814	1.9005	100	5.8116	-0.1029
38.6656	2.3791	125	6.4027	-0.0678
39.4825	2.8531	150	14.8689	0.1094
38.0107	3.3318	175	9.4517	-0.0727
31.663	3.8057	200	6.2469	-0.0694
26.0169	4.2844	225	5.8747	-0.0536
40.9899	4.7583	250	5.7418	-0.0746
32.5293	5.2370	275	5.7128	-0.0553
36.4987	5.7109	300	5.6100	-0.0852
31.5365	6.1896	325	5.6516	-0.0611
28.7596	6.6635	350	5.6693	-0.0820
45.4104	7.1422	375	5.5758	-0.0880
27.6256	7.6161	400	5.5985	-0.0592
35.2245	8.0948	425	5.5941	-0.0993
38.0101	8.5687	450	5.5649	-0.0683
37.5459	9.0474	475	5.5382	-0.0948
35.274	9.5213	500	5.5468	-0.0809

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1