Finetune Whisper

POST {{baseUrl}}/finetune/speech2text/whisper

Endpoint to configure and start a Whisper finetuning job.

Request Body

{"huggingface_config"=>{"hf_login_key"=>"<string>", "hf_modelsavepath"=>"<string>"}, "pretrainedmodel_config"=>{"model_path"=>"OpenAI/whisper-large-v3", "task"=>"transcribe", "language"=>"Hindi"}, "data_config"=>{"data_path"=>"dummy/common_voice_13_0", "data_subset"=>"hi", "data_source_type"=>"hub_link"}, "training_config"=>{"gradient_accumulation_steps"=>4, "learning_rate"=>0.001, "warmup_steps"=>50, "num_train_epochs"=>0.01, "per_device_train_batch_size"=>8, "generation_max_length"=>128, "lr_scheduler_type"=>"reduce_lr_on_plateau"}, "logging_config"=>{"use_wandb"=>false, "wandb_login_key"=>"API_KEY", "wandb_run_name"=>""}}

HEADERS

RESPONSES

status: OK

{}