Shakshuka POC experimental

At a glance

Formats and quants

Hugging Face issues a DOI per repository, so each format can be cited on its own. Repositories, sizes and DOIs checked on the Hub on 2026-10-21.

Versions

Suggested implementations

Shakshuka is the Nemotron streaming line, named like Jachnun and Sabich. The two checkpoints published so far are proofs of concept, Shakshuka PoC-1 and PoC-2. The first retrain will be Shakshuka v1.0.

experimental streaming measured at 560 ms not in any app

Today it runs in NVIDIA NeMo, from the .nemo file. Offline transcription, from the model card:

import nemo.collections.asr as nemo_asr
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    "Hebrish/Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-Codeswitching",
    "Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-Codeswitching.nemo",
)
model = nemo_asr.models.ASRModel.restore_from(path)
print(model.transcribe(["clip.wav"])[0].text)  # 16 kHz mono

Streaming works as for the base model, with cache-aware inference. Set att_context_size = [70, N]:

Smaller chunks mean less delay. NeMo's speech_to_text_cache_aware_streaming_infer.py example script shows the loop; see the base model card.

• Streaming accuracy, measured at 560 ms. On the held-out split (387 clips) it scored 5.65% code-switching WER while streaming, against 5.81% offline: no measurable loss, although it was fine-tuned offline with full context. The smallest chunks (80 and 160 ms) are not measured; check them for your use.
• No on-device format yet. sherpa-onnx runs NeMo cache-aware streaming transducers, so an ONNX export for sherpa-onnx is the likely route to phones and desktops. Nobody has tried it on this model yet; it is on the projects board.
• Licence: the NVIDIA Open Model License Agreement, not CC-BY-4.0. Read it before shipping. Not made or endorsed by NVIDIA.

Cite

Cite Shakshuka with its DOI, 10.57967/hf/11107 (Hebrish/Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-Codeswitching). To cite one format, use that repository's DOI from the table above.

@misc{rosehill_shakshuka_2026,
  author    = {Daniel Rosehill},
  title     = {Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-Codeswitching},
  year      = {2026},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/Hebrish/Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-Codeswitching},
  doi       = {10.57967/hf/11107}
}

Request access

All Hebrish models