Towards Cross-Tokenizer Distillation: the Universal Logit Distillation Loss for LLMs
Paper
•
2402.12030
•
Published
LLMs Distillation
-
-
mistralai/Mistral-7B-Instruct-v0.2
Text Generation • Updated • 1.09M • • 2.57k -
meta-llama/Llama-2-7b-chat-hf
Text Generation • Updated • 823k • • 3.97k -
EleutherAI/pythia-160m-deduped
Text Generation • Updated • 11.8k • 3 -
EleutherAI/pythia-410m-deduped
Text Generation • Updated • 84.6k • 20 -
EleutherAI/pythia-1b-deduped
Text Generation • Updated • 11k • 17 -
bigscience/bloomz-560m
Text Generation • Updated • 15M • 107 -
bigscience/mt0-base
Text2Text Generation • Updated • 3.05k • 30 -
facebook/opt-350m
Text Generation • Updated • 228k • 130 -
rajpurkar/squad
Viewer • Updated • 98.2k • 53.6k • 264 -
google-research-datasets/qed
Updated • 121 • 4 -
GEM/FairytaleQA
Viewer • Updated • 10.6k • 117 • 8 -
qiaojin/PubMedQA
Viewer • Updated • 274k • 2.83k • 157 -
knkarthick/dialogsum
Viewer • Updated • 14.5k • 12k • 180