mmBERT is trained on 3T tokens from over 1800 languages, showing SoTA scores on benchmarks and exceptional low-resource performance
AI & ML interests
None defined yet.
Recent Activity
View all activity
Papers
DAR: Deontic Reasoning with Agentic Harnesses
Trust Functions: Near-Lossless Weak-to-Strong Generalization by Learning When to Trust the Weak Teacher
models 53
jhu-clsp/mmBERT-small
Fill-Mask • Updated • 29.9k • • 76
jhu-clsp/mmBERT-base
Fill-Mask • Updated • 342k • • 214
jhu-clsp/mmBERT-checkpoints
Updated • 4
jhu-clsp/ettin-decoder-1b
Fill-Mask • Updated • 16 • 5
jhu-clsp/ettin-decoder-32m
Text Generation • Updated • 398
jhu-clsp/ettin-encoder-1b
Feature Extraction • Updated • 1.94k • 22
jhu-clsp/ettin-encoder-68m
Fill-Mask • Updated • 3.54k • • 5
jhu-clsp/ettin-dec-from-enc-32m
Text Generation • Updated • 4
jhu-clsp/ettin-encoder-150m
Fill-Mask • Updated • 6.38k • • 13
jhu-clsp/ettin-decoder-400m
Text Generation • Updated • 6.97k • 4
datasets 40
jhu-clsp/ManyIH-Bench
Preview • Updated • 49 • 3
jhu-clsp/robust04-instructions
Viewer • Updated • 136k • 816 • 2
jhu-clsp/core17-instructions
Viewer • Updated • 49.4k • 707 • 2
jhu-clsp/news21-instructions
Viewer • Updated • 71.5k • 652 • 1
jhu-clsp/SciTaRC
Viewer • Updated • 371 • 48 • 1
jhu-clsp/megawika-2
Updated • 51 • 4
jhu-clsp/mmBERT-decay-data
Updated • 33k • 6
jhu-clsp/mmBERT-midtraining-data
Updated • 1.95k • 1
jhu-clsp/ettin-pretraining-data
Updated • 58.8k • 9
jhu-clsp/ettin-decay-data
Updated • 1.24k • 1