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 • 703k • • 2.58k -
meta-llama/Llama-2-7b-chat-hf
Text Generation • Updated • 721k • • 4.03k -
EleutherAI/pythia-160m-deduped
Text Generation • Updated • 13.9k • 3 -
EleutherAI/pythia-410m-deduped
Text Generation • Updated • 49.8k • 20 -
EleutherAI/pythia-1b-deduped
Text Generation • Updated • 42.4k • 17 -
bigscience/bloomz-560m
Text Generation • Updated • 14.4M • 109 -
bigscience/mt0-base
Text2Text Generation • Updated • 2.31k • 30 -
facebook/opt-350m
Text Generation • Updated • 229k • 131 -
rajpurkar/squad
Viewer • Updated • 98.2k • 60.4k • 270 -
google-research-datasets/qed
Updated • 148 • 4 -
GEM/FairytaleQA
Viewer • Updated • 10.6k • 143 • 8 -
qiaojin/PubMedQA
Viewer • Updated • 274k • 3.39k • 162 -
knkarthick/dialogsum
Viewer • Updated • 14.5k • 11.7k • 183