Hugging Face
Models
Datasets
Spaces
Buckets
new
Docs
Enterprise
Pricing
Log In
Sign Up
soda-research
's Collections
SODA Models
Discrete Audio IsoFLOP Models
SODA Warm-Start (Qwen3-based) Models
Discrete Audio Pretrain Data
SODA Models
updated
Feb 13
SODA models trained on Yodas+Emilia+Nemotron for 500B tokens from scratch
Upvote
-
Sleeping
8
SODA Demo
🥤
8
SODA Demo
soda-research/soda-4b-base
4B
•
Updated
Feb 19
•
11.3k
•
7
soda-research/soda-1.7b-base
2B
•
Updated
Feb 19
•
697
soda-research/soda-600m-base
0.7B
•
Updated
Feb 19
•
633
soda-research/soda-135m-base
0.2B
•
Updated
Feb 19
•
1.24k
•
2
soda-research/soda-600m-prelim
0.6B
•
Updated
Feb 13
•
227
soda-research/marin-mimi-bpe-8cb-16k-tokenizer
Updated
Oct 9, 2025
•
1
Upvote
-
Share collection
View history
Collection guide
Browse collections