Models to accompany research paper on training multi token prediction language models using self-distillation.
AI & ML interests
AI security & privacy, algorithmic bias, foundations of ML
Recent Activity
View all activity
Papers
Teaching Pretrained Language Models to Think Deeper with Retrofitted Recurrence
Gemstones: A Model Suite for Multi-Faceted Scaling Laws
spaces 5
Sleeping
Featured
101
A Watermark for LLMs
💧
Generate text with a watermark
Running on Zero
13
DynaGuard
💬
Dynaguard v1
Runtime error
5
CinePileLeaderboard
🔥
Video-LLM evaluations on CinePile's evaluation split.
Running on Zero
72
Binoculars
👀
Launch an interactive web demo with Gradio
Runtime error
132
Pez Dispenser
⚡
models 138
tomg-group-umd/DynaGuard-1.7B
Text Generation • Updated
• 84 • 3
tomg-group-umd/DynaGuard-4B
Text Generation • 4B • Updated
• 14 • 2
tomg-group-umd/DynaGuard-8B
Text Generation • 8B • Updated
• 241 • 15
tomg-group-umd/step-00010720-baseline_2_0
Text Generation • 4B • Updated
• 5
tomg-group-umd/LoRI-D_nlu_llama3_rank_64
Text Generation • Updated
tomg-group-umd/LoRI-D_safety_llama3_rank_64
Text Generation • Updated
• 4
tomg-group-umd/LoRI-D_nlu_llama3_rank_32
Text Generation • Updated
tomg-group-umd/LoRI-S_nlu_llama3_rank_32
Text Generation • Updated
• 2
tomg-group-umd/LoRI-S_nlu_llama3_rank_64
Text Generation • Updated
• 2
tomg-group-umd/LoRI-D_code_llama3_rank_32
Text Generation • Updated
• 1
datasets 23
tomg-group-umd/pixelprose
Viewer
• Updated
• 15.6M • 444 • 163
tomg-group-umd/huginn-dataset
Viewer
• Updated
• 274M • 502k • 6
tomg-group-umd/gemstones_data_order_sequential
Viewer
• Updated
• 170M • 91
tomg-group-umd/gemstones_data_order_parallel
Viewer
• Updated
• 170M • 94
tomg-group-umd/argus
Viewer
• Updated
• 500 • 222 • 1
tomg-group-umd/morse-500
Updated
• 28
tomg-group-umd/fictionalqa_reformatted_triviaqa
Viewer
• Updated
• 16.4k • 12
tomg-group-umd/fictionalqa_training_splits
Viewer
• Updated
• 107k • 125
tomg-group-umd/fictionalqa
Viewer
• Updated
• 31.7k • 114 • 2
tomg-group-umd/Gemstone-100M-dolma-val-data
Viewer
• Updated
• 49.2k • 6