ivrit-whisper-large-v3-turbo-ggml
Quantizations of the authors' own whisper.cpp f16 conversion of ivrit-ai/whisper-large-v3-turbo-ggml (the f16 is mirrored unchanged).
Source checkpoint: ivrit-ai/whisper-large-v3-turbo-ggml by ivrit-ai · License: apache-2.0 (unchanged; this repo only re-packages the weights)
Engine: Load with whisper.cpp (whisper-cli -m <file>), or any app that embeds it.
Files
| File | Quantization | Size | Note |
|---|---|---|---|
ggml-ivrit-whisper-large-v3-turbo-f16.bin |
f16 | 1625 MB | |
ggml-ivrit-whisper-large-v3-turbo-q4_0.bin |
q4_0 | 474 MB | |
ggml-ivrit-whisper-large-v3-turbo-q5_0.bin |
q5_0 | 574 MB | |
ggml-ivrit-whisper-large-v3-turbo-q8_0.bin |
q8_0 | 874 MB |
f16 is the lossless conversion; q8_0 is nearly identical in accuracy at ~55 % of the size; q5_0/q5_k are the phone-friendly choice; q4_* is smallest with a small accuracy cost.
How these were made
Converted from the upstream checkpoint with the engine's own converter, then quantized with the engine's quantizer. Each variant was checked by transcribing short Portuguese and English samples before upload.
Attribution
Weights are derivative works of the upstream model and keep its license. Please cite the original authors (ivrit-ai). The conversion and hosting here are maintained by JoaoZaokk so that the download links used by the Odysseus / Open WebUI native apps stay stable. No warranty.