You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

whisper-medium-zarma-model

This model is a fine-tuned version of openai/whisper-medium on an unknown dataset. It achieves the following results on the evaluation set:

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 4
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 50
training_steps: 500
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.8549	0.3072	50	1.5952	60.3192	24.5845
0.4441	0.6144	100	1.4364	52.7604	24.5657
0.3995	0.9217	150	1.2670	43.7028	18.5743
0.303	1.2273	200	1.3466	48.8031	23.1932
0.2902	1.5346	250	1.3980	40.7591	17.9057
0.2928	1.8418	300	1.3470	44.7595	22.2256
0.2149	2.1475	350	1.3600	41.4169	19.8526
0.1971	2.4547	400	1.4339	40.6621	20.5565
0.1968	2.7619	450	1.4413	43.5303	21.3758
0.1832	3.0676	500	1.4032	41.4384	20.6884

Safetensors

Model size

0.8B params

Tensor type

F32

Base model

Finetuned

(768)

this model