Update README.md

21aee2b verified 4 days ago

1.31 kB

license: cc-by-4.0
language:
  - ja
base_model:
  - nvidia/parakeet-tdt_ctc-0.6b-ja
pipeline_tag: automatic-speech-recognition
tags:
  - automatic-speech-recognition
  - asr
  - parakeet
  - gguf
  - ggml
  - cpp-inference
  - nemo

Parakeet ja GGUF — Japanese models for parakeet.cpp

GGUF-format weights for parakeet.cpp, a C++/ggml port of NVIDIA NeMo Parakeet that matches the upstream PyTorch models on CPU. This repo collects Japanese model × quantization as a flat set of .gguf files — download just the one you need.

Models

tdt_ctc-0.6b-ja

Source: nvidia/parakeet-tdt_ctc-0.6b-ja

File	Variant	Size
`ttdt_ctc-0.6b-ja-f16.gguf`	F16	1359.9 MB
`ttdt_ctc-0.6b-ja-q8_0.gguf`	Q8_0	884.1MB
`ttdt_ctc-0.6b-ja-q5_0.gguf`	Q5_0	693.8 MB
`ttdt_ctc-0.6b-ja-q4_0.gguf`	Q4_0	630.3 MB
`ttdt_ctc-0.6b-ja-q6_k.gguf`	Q6_K	762.4 MB
`ttdt_ctc-0.6b-ja-q5_k.gguf`	Q5_K	695.1 MB
`ttdt_ctc-0.6b-ja-q4_k.gguf`	Q4_K	631.7 MB

License

The GGUF weights are derived from the NVIDIA NeMo Parakeet checkpoints, released under the CC-BY-4.0 license. The parakeet.cpp runtime is MIT-licensed.