joeyMartig/whisper-large-v3-dnd-fr-test

Whisper large v3 FR D&D - Joey Martig

This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0138
• Wer: 33.4454

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: 1e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	7	1.0810	38.6555
No log	2.0	14	1.0631	38.7395
No log	3.0	21	0.9917	38.1513
No log	4.0	28	0.9133	37.3109
No log	5.0	35	0.7915	37.0588
No log	6.0	42	0.7010	35.6303
No log	7.0	49	0.6078	34.6218
No log	8.0	56	0.5114	45.6303
No log	9.0	63	0.4258	40.0
No log	10.0	70	0.3484	33.4454
No log	11.0	77	0.2802	33.4454
No log	12.0	84	0.2228	33.4454
No log	13.0	91	0.1804	33.4454
No log	14.0	98	0.1436	36.5546
No log	15.0	105	0.1166	33.4454
No log	16.0	112	0.0932	36.5546
No log	17.0	119	0.0730	36.5546
No log	18.0	126	0.0573	36.5546
No log	19.0	133	0.0451	36.5546
No log	20.0	140	0.0390	33.4454
No log	21.0	147	0.0319	33.4454
No log	22.0	154	0.0287	33.4454
No log	23.0	161	0.0252	33.4454
No log	24.0	168	0.0224	33.4454
No log	25.0	175	0.0209	33.4454
No log	26.0	182	0.0199	33.4454
No log	27.0	189	0.0186	33.4454
No log	28.0	196	0.0179	33.4454
No log	29.0	203	0.0175	33.4454
No log	30.0	210	0.0168	33.4454
No log	31.0	217	0.0164	33.4454
No log	32.0	224	0.0161	33.4454
No log	33.0	231	0.0158	33.4454
No log	34.0	238	0.0156	33.4454
No log	35.0	245	0.0153	33.4454
No log	36.0	252	0.0151	33.4454
No log	37.0	259	0.0149	33.4454
No log	38.0	266	0.0148	33.4454
No log	39.0	273	0.0146	33.4454
No log	40.0	280	0.0145	33.4454
No log	41.0	287	0.0143	33.4454
No log	42.0	294	0.0143	33.4454
No log	43.0	301	0.0141	33.4454
No log	44.0	308	0.0141	33.4454
No log	45.0	315	0.0140	33.4454
No log	46.0	322	0.0139	33.4454
No log	47.0	329	0.0139	33.4454
No log	48.0	336	0.0138	33.4454
No log	49.0	343	0.0138	33.4454
No log	50.0	350	0.0138	33.4454

Framework versions

• Transformers 4.43.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1