Configuration Parsing Warning:In adapter_config.json: "peft.task_type" must be a string

Whisper Medium — English / Spanish / Miami Bangor

This model is a fine-tuned version of openai/whisper-medium on the Common Voice 24.0 (English) — gender-balanced subset, the Common Voice 24.0 (Spanish) — gender-balanced subset and the Bangor Miami Corpus — code-switching Spanish/English interviews, segmented at utterance level from CHAT transcripts datasets. It achieves the following results on the evaluation set:

Loss: 0.2052
Wer: 6.8973
Cer: 2.5627

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: 5e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Use OptimizerNames.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: 500
training_steps: 122820

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.2806	0.05	6141	0.2132	6.9891	2.6829
0.249	1.0500	12282	0.2081	6.8825	2.6052
0.2778	2.0500	18423	0.2065	6.8662	2.5899
0.2527	3.0500	24564	0.2058	6.8573	2.5652
0.2351	4.0500	30705	0.2055	6.8795	2.5731
0.2533	5.0500	36846	0.2055	6.8943	2.5711
0.2439	6.0500	42987	0.2052	6.8973	2.5627

Framework versions

PEFT 0.18.1
Transformers 4.52.0
Pytorch 2.9.1+cu128
Datasets 4.5.0
Tokenizers 0.21.4

Downloads last month: 38

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for chengyili2005/whisper-medium-DINA

Base model

openai/whisper-medium

Adapter

(132)

this model

Evaluation results

Wer on Common Voice 24.0 (English) — gender-balanced subset
self-reported

6.897
Wer on Common Voice 24.0 (Spanish) — gender-balanced subset
self-reported

6.897