nu-dialogue/j-moshi
Updated
•
3
Laurent Mazare
lmz
AI & ML interests
None yet
Recent Activity
liked
a model
1 day ago
nu-dialogue/j-moshi
liked
a model
2 days ago
nu-dialogue/j-moshi-ext
new activity
12 days ago
kyutai/helium-1-preview-2b:fix architecture
Organizations
lmz's activity
liked
a
model
1 day ago
liked
a
model
2 days ago
fix architecture
2
#3 opened 12 days ago
by
TPM-28
liked
a
model
12 days ago
updated
a
model
12 days ago
fix title
1
#2 opened 12 days ago
by
eliebak
liked
a
model
12 days ago
Update tokenizer_config.json
#1 opened 13 days ago
by
ArthurZ
updated
a
model
13 days ago
Number of quantizers
1
#3 opened 4 months ago
by
mithileshvaidya
updated
a
model
19 days ago
liked
a
Space
about 2 months ago
liked
a
model
2 months ago
reacted to
reach-vb's
post with 🔥
4 months ago
Post
2869
Less than two days ago Kyutai Labs open sourced Moshi - an ~7.6B on-device Speech to Speech foundation model and Mimi - SoTA streaming speech codec! 🔥
The release includes:
1. Moshiko & Moshika - Moshi finetuned on synthetic data (CC-BY license) ( kyutai/moshi-v01-release-66eaeaf3302bef6bd9ad7acd)
2. Mimi - Streaiming Audio Codec, processes 24 kHz audio, down to a 12.5 Hz representation with a bandwidth of 1.1 kbps (CC-BY license) ( kyutai/mimi)
3. Model checkpoints & Inference codebase written in Rust (Candle), PyTorch & MLX (Apache license) (https://github.com/kyutai-labs/moshi)
How does Moshi work?
1. Moshi processes two audio streams: one for itself and one for the user, with the user's stream coming from audio input and Moshi's stream generated by the model.
2. Along with these audio streams, Moshi predicts text tokens for its speech, enhancing its generation quality.
3. The model uses a small Depth Transformer for codebook dependencies and a large 7B parameter Temporal Transformer for temporal dependencies.
4. The theoretical latency is 160ms, with a practical latency of around 200ms on an L4 GPU.
Model size & inference:
Moshiko/ka are 7.69B param models
bf16 ~16GB VRAM
8-bit ~8GB VRAM
4-bit ~4GB VRAM
You can run inference via Candle 🦀, PyTorch and MLX - based on your hardware.
The Kyutai team, @adefossez @lmz and team are cracked AF, they're bringing some serious firepower to the open source/ science AI scene, looking forward to what's next! 🐐
The release includes:
1. Moshiko & Moshika - Moshi finetuned on synthetic data (CC-BY license) ( kyutai/moshi-v01-release-66eaeaf3302bef6bd9ad7acd)
2. Mimi - Streaiming Audio Codec, processes 24 kHz audio, down to a 12.5 Hz representation with a bandwidth of 1.1 kbps (CC-BY license) ( kyutai/mimi)
3. Model checkpoints & Inference codebase written in Rust (Candle), PyTorch & MLX (Apache license) (https://github.com/kyutai-labs/moshi)
How does Moshi work?
1. Moshi processes two audio streams: one for itself and one for the user, with the user's stream coming from audio input and Moshi's stream generated by the model.
2. Along with these audio streams, Moshi predicts text tokens for its speech, enhancing its generation quality.
3. The model uses a small Depth Transformer for codebook dependencies and a large 7B parameter Temporal Transformer for temporal dependencies.
4. The theoretical latency is 160ms, with a practical latency of around 200ms on an L4 GPU.
Model size & inference:
Moshiko/ka are 7.69B param models
bf16 ~16GB VRAM
8-bit ~8GB VRAM
4-bit ~4GB VRAM
You can run inference via Candle 🦀, PyTorch and MLX - based on your hardware.
The Kyutai team, @adefossez @lmz and team are cracked AF, they're bringing some serious firepower to the open source/ science AI scene, looking forward to what's next! 🐐
- 1 reply
updated
a
Space
4 months ago
updated
a
model
6 months ago
updated
a
model
8 months ago
updated
a
model
10 months ago
updated
2
models
11 months ago