Modelpublic

TakalaWang/Discussion-Phi-4-multimodal-instruct-audio-dimp-tag

sourceHugging Facemitupdated 1y agoView on Hugging Face

0likes8downloads

Model Card

Discussion-Phi-4-multimodal-instruct-audio-dimp-tag

This model is a fine-tuned version of microsoft/Phi-4-multimodal-instruct on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 51.4496

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: 4e-05
—trainbatchsize: 1
—evalbatchsize: 8
—seed: 42
—gradientaccumulationsteps: 16
—totaltrainbatch_size: 16
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.95) and epsilon=1e-07 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—lrschedulerwarmup_steps: 50
—num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
2436765.25	0.1117	10	15290.4863
1.2412	0.2235	20	62.7128
48.4906	0.3352	30	72.5308
147.1698	0.4469	40	67.7048
1.3424	0.5587	50	65.1818
2.9574	0.6704	60	54.5024
1.2336	0.7821	70	50.8777
0.7593	0.8939	80	58.8612
0.396	1.0	90	63.9311
0.6103	1.1117	100	59.3974
3.0362	1.2235	110	59.0700
3.3966	1.3352	120	59.1827
0.196	1.4469	130	57.5179
4.7776	1.5587	140	63.4193
1.6803	1.6704	150	60.0916
0.3914	1.7821	160	52.0365
0.6947	1.8939	170	66.4174
0.1631	2.0	180	60.5770
0.3811	2.1117	190	47.5456
0.4748	2.2235	200	59.0834
3.4445	2.3352	210	66.0457
2.8572	2.4469	220	64.5688
0.1471	2.5587	230	60.8787
0.2193	2.6704	240	57.0804
0.1406	2.7821	250	64.1730
0.5833	2.8939	260	51.4496

Framework versions

—Transformers 4.48.2
—Pytorch 2.4.1+cu124
—Datasets 3.6.0
—Tokenizers 0.21.1