Modelpublic

Trelis/whisper-tiny-llm-lingo

sourceHugging Faceupdated 8mo agoView on Hugging Face

0likes16downloads

Model Card

Whisper Tiny Llm Lingo

Fine-tuned Whisper model based on openai/whisper-tiny.

Training Results

Metric	Base Model	Fine-tuned
WER	107.85%	34.30%

Improvement: 73.55% WER reduction (lower is better)

Training Details

—Base Model: openai/whisper-tiny
—Training Dataset: Trelis/llm-lingo
—Train Loss: 2.5791
—Training Time: 19 seconds

Inference

python

from transformers import pipeline

asr = pipeline("automatic-speech-recognition", model="Trelis/whisper-tiny-llm-lingo")
result = asr("path/to/audio.wav")
print(result["text"])

Training Logs

Full training logs are available in training_log.txt.

Fine-tuned using [Trelis Studio](https://studio.trelis.com)