metadata
base_model: openai/whisper-large-v2
library_name: peft
model-index:
- name: lowhipa-base-asc
results: []
datasets:
- tunis-ai/arabic_speech_corpus
pipeline_tag: automatic-speech-recognition
lowhipa-base-asc
This Whisper-for-IPA (WhIPA) model adapter is a PEFT LoRA fine-tuned version of openai/whisper-large-v2 on a subset (1k samples) of the Arabic Speech Corpus (https://en.arabicspeechcorpus.com) with custom IPA transcriptions transliterated from the provided Buckwalter transcriptions; ASC-IPA dataset available at https://doi.org/10.5281/zenodo.17111977.
Model description
For deployment and description, please refer to https://github.com/jshrdt/whipa.
from transformers import WhisperForConditionalGeneration, WhisperTokenizer, WhisperProcessor
from peft import PeftModel
tokenizer = WhisperTokenizer.from_pretrained("openai/whisper-large-v2", task="transcribe")
tokenizer.add_special_tokens({"additional_special_tokens": ["<|ip|>"] + tokenizer.all_special_tokens})
base_model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")
base_model.generation_config.lang_to_id["<|ip|>"] = tokenizer.convert_tokens_to_ids(["<|ip|>"])[0]
base_model.resize_token_embeddings(len(tokenizer))
whipa_model = PeftModel.from_pretrained(base_model, "jshrdt/lowhipa-large-asc")
whipa_model.generation_config.language = "<|ip|>"
whipa_model.generation_config.task = "transcribe"
whipa_processor = WhisperProcessor.from_pretrained("openai/whisper-large-v2", task="transcribe")
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.2402 | 2.0 | 126 | 0.2061 |
| 0.1 | 4.0 | 252 | 0.1705 |
| 0.0411 | 6.0 | 378 | 0.1515 |
| 0.0118 | 8.0 | 504 | 0.0.1530 |
| 0.0056 | 10.0 | 630 | 0.1585 |
Framework versions
- PEFT 0.15.1
- Transformers 4.48.3
- Pytorch 2.6.0+cu124
- Datasets 3.2.0
- PEFT 0.15.1