humor_model_v2

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2683
• Accuracy: 0.9639

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9524	5	0.6652	0.9157
0.6722	1.9048	10	0.5931	0.9217
0.6722	2.8571	15	0.5272	0.9337
0.5461	4.0	21	0.4712	0.8554
0.5461	4.9524	26	0.3943	0.8916
0.3891	5.9048	31	0.3369	0.9337
0.3891	6.8571	36	0.3099	0.9398
0.2976	8.0	42	0.2811	0.9578
0.2976	8.9524	47	0.2713	0.9578
0.2393	9.5238	50	0.2683	0.9639

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1