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 • 1.05M • • 2.57k -
meta-llama/Llama-2-7b-chat-hf
Text Generation • Updated • 748k • • 3.98k -
EleutherAI/pythia-160m-deduped
Text Generation • Updated • 12.2k • 3 -
EleutherAI/pythia-410m-deduped
Text Generation • Updated • 89k • 20 -
EleutherAI/pythia-1b-deduped
Text Generation • Updated • 11.3k • 17 -
bigscience/bloomz-560m
Text Generation • Updated • 16.7M • 107 -
bigscience/mt0-base
Text2Text Generation • Updated • 2.97k • 30 -
facebook/opt-350m
Text Generation • Updated • 218k • 130 -
rajpurkar/squad
Viewer • Updated • 98.2k • 53.2k • 264 -
google-research-datasets/qed
Updated • 119 • 4 -
GEM/FairytaleQA
Viewer • Updated • 10.6k • 99 • 8 -
qiaojin/PubMedQA
Viewer • Updated • 274k • 3.05k • 157 -
knkarthick/dialogsum
Viewer • Updated • 14.5k • 11.7k • 180