Downloads · 30 days
57
32% of all-time downloads
stephvax/kyutai-stt-1b-en_fr-candle-gguf
kyutai-stt-1b-en_fr-candle-gguf is a automatic speech recognition model from stephvax. Use it when you need speech turned into text. It is set up for candle. The card lists the license as cc-by-4.0.
Quantised language-model weights for kyutai/stt-1b-enfr-candle, in the format candle's moshi crate loads (candletransformers::quantizedvarbuilder::VarBuilder::fromgguf).
Downloads · 30 days
57
32% of all-time downloads
All-time downloads
179
Public
Repo size
1.6 GB
Likes
0
Public
Click a slice to open those files.
.gguf1.6 GB · 100%
From the Hugging Face model README
Quantised language-model weights for
kyutai/stt-1b-en_fr-candle,
in the format candle's moshi crate loads
(candle_transformers::quantized_var_builder::VarBuilder::from_gguf).
Kyutai publishes candle-quantised checkpoints for its other models
(moshiko-candle-q8,
hibiki-1b-rs-q8) in exactly
this layout — LM in GGUF, Mimi as a separate safetensors file — but not for the
STT model. These fill that gap.
| file | contents | size |
|---|---|---|
model.q8_0.gguf | the LM weights, Q8_0 | 1051 MB |
model.q4_k.gguf | the LM weights, Q4_K | 557 MB |
These replace model.safetensors only. You still need the other three
files from the upstream repo, unchanged:
config.jsontokenizer_en_fr_audio_8000.model[email protected] (the Mimi codec — not quantised)Drop one GGUF into the model directory as model.gguf beside those three.
Not compatible with
moshi.cpp/ ggml runtimes. Those use a different GGUF layout (general.architecture=kyutai-stt, LM and Mimi in one file, tokenizer embedded). These files carry candle's tensor names, verbatim from the safetensors, and nothing else.
200 utterances of FLEURS fr_fr (test split), decoded through the reference
Rust/candle path, Apple M-series:
| variant | LM size | WER | median WER | RTFx (Metal) | RTFx (CPU) |
|---|---|---|---|---|---|
| upstream f32 | 1978 MB | 17.18 % | 10.8 % | 3.0x | 0.6x |
q8_0 | 1051 MB | 17.12 % | 10.5 % | 5.7x | 1.8x |
q4_k | 557 MB | 18.04 % | 11.9 % | 5.8x | 2.1x |
q8_0 is the one to use. Its WER is the upstream's within noise (−0.06
points), for 47 % less weight and roughly twice the speed. On the CPU the
difference is not an optimisation but a precondition: the f32 runs below real
time (0.6x), so a streaming decoder never catches up.
q4_k is published so the comparison is reproducible, not because it is worth
using: it costs +0.92 WER points and gains nothing once warm.
The global WER is dragged by a tail of hard failures rather than by uniform
error — the median is 10.5 %. Six clips return empty output, and they are
the same six for all three variants (FLEURS test indices 29, 31, 93, 125, 140,
166), so that belongs to the model, not to the quantisation. Excluding them,
q8_0 scores 15.46 %. A handful of clips are transcribed into English or
Spanish: the model is bilingual en/fr with no language pin.
candle's own rule, applied by
quantize.rs
(the same rule as candle's tensor-tools): every rank-2 tensor whose last
dimension is a multiple of the block size is quantised, everything else stays
F32. 102 tensors quantised, 33 kept. Tensor names are preserved verbatim —
that is what makes moshi::lm::LmModel find them.
quantize model.safetensors model.q8_0.gguf q8_0
Source checkpoint SHA-256:
b9e97c53229dce728d65c76bfa892f7b563c69d671899f0ebc6518582dddec6f.
Upstream kyutai/stt-1b-en_fr-candle is CC-BY-4.0 and ungated; these
derivatives are redistributed under the same terms and grant no rights the
upstream licence does not. Only the numeric representation of the weights
differs — no training, no fine-tuning, no merging, no distillation, no change
to architecture, vocabulary or capability. Under Regulation (EU) 2024/1689 the
upstream authors remain the provider of this model; questions about training
content, copyright policy or model capability belong upstream.