Add model card

README.md

@@ -0,0 +1,52 @@
+---
+language: he
+license: apache-2.0
+library_name: transformers
+tags:
+  - whisper
+  - audio
+  - automatic-speech-recognition
+  - hebrew
+datasets:
+  - ivrit-ai/whisper-training
+base_model: openai/whisper-small
+pipeline_tag: automatic-speech-recognition
+---
+
+# whisper-small-he
+
+Hebrew fine-tuned [Whisper Small](https://huggingface.co/openai/whisper-small) for automatic speech recognition.
+
+## Training
+
+- **Base model**: [openai/whisper-small](https://huggingface.co/openai/whisper-small)
+- **Dataset**: [ivrit-ai/whisper-training](https://huggingface.co/datasets/ivrit-ai/whisper-training) (~400h Hebrew)
+- **Method**: Supervised fine-tuning with `Seq2SeqTrainer`
+- **Steps**: 5,000 (streaming, effective batch size 16)
+- **Hardware**: Apple M4 (MPS), fp32
+- **Best eval WER**: 0.368 (on 200-sample test split, step 4000)
+
+## Usage
+
+```python
+from transformers import WhisperProcessor, WhisperForConditionalGeneration
+
+processor = WhisperProcessor.from_pretrained("amitkot/whisper-small-he")
+model = WhisperForConditionalGeneration.from_pretrained("amitkot/whisper-small-he")
+
+model.generation_config.language = "he"
+model.generation_config.task = "transcribe"
+```
+
+## Training pipeline
+
+Trained using [whisper-acft-pipeline](https://github.com/amitkot/whisper-acft-pipeline):
+
+```bash
+uv run python scripts/finetune.py --config configs/hebrew_small_finetune.yaml
+```
+
+## See also
+
+- [amitkot/whisper-small-he-acft](https://huggingface.co/amitkot/whisper-small-he-acft) — ACFT-optimized version of this model for short audio (FUTO Keyboard)
+- [amitkot/whisper-tiny-he](https://huggingface.co/amitkot/whisper-tiny-he) — Smaller/faster variant