onnx-community/trained-gender-ONNX

trained-gender (ONNX)

This is an ONNX version of crangana/trained-gender. It was automatically converted and uploaded using this Hugging Face Space.

Usage with Transformers.js

See the pipeline documentation for image-classification: https://huggingface.co/docs/transformers.js/api/pipelines#module_pipelines.ImageClassificationPipeline

---

trained-gender

This model is a fine-tuned version of microsoft/resnet-50 on the fair_face dataset. It achieves the following results on the evaluation set:

• Loss: 0.2437
• Accuracy: 0.8986

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: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4277	0.18	1000	0.4054	0.8089
0.315	0.37	2000	0.3487	0.8318
0.3082	0.55	3000	0.3052	0.8633
0.3235	0.74	4000	0.2899	0.8684
0.2505	0.92	5000	0.2693	0.8785
0.2484	1.11	6000	0.2547	0.8889
0.1933	1.29	7000	0.2521	0.8901
0.1497	1.48	8000	0.2443	0.8929
0.326	1.66	9000	0.2406	0.8958
0.215	1.84	10000	0.2381	0.9007
0.2035	2.03	11000	0.2437	0.8986

Framework versions

• Transformers 4.34.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.14.0