learning_rate: 1e-05
train_batch_size: 32
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 64
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: 80
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4797	0.8696	10	0.4806	5.8239
0.4503	1.6957	20	0.4695	5.6818
0.4076	2.5217	30	0.4219	5.5398
0.3637	3.3478	40	0.3402	4.8295
0.2583	4.1739	50	0.1697	4.1193
0.1121	5.0	60	0.0978	3.8352
0.0751	5.8696	70	0.0825	3.8352
0.0464	6.6957	80	0.0671	3.125

Framework versions

Transformers 4.51.3
Pytorch 2.7.0+cu126
Datasets 3.5.1
Tokenizers 0.21.1

Usage

python

from transformers import pipeline
import gradio as gr

# Load the fine-tuned Whisper model for medical speech recognition
pipe = pipeline(model="Johnyquest7/whisper-small-finetuned-medical3", return_timestamps=True)

# Define transcription function
def transcribe(audio):
    text = pipe(audio)["text"]
    return text

# Create a Gradio interface
iface = gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(sources=["upload", "microphone"], type="filepath"),
    outputs="text",
    title="Whisper Small Medical",
    description="Demo for medical speech recognition using a fine-tuned Whisper small model."
)

# Launch the interface
iface.launch()