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 • 3.81M • • 2.66k -
meta-llama/Llama-2-7b-chat-hf
Text Generation • Updated • 1.22M • • 4.26k -
EleutherAI/pythia-160m-deduped
Text Generation • Updated • 46k • 3 -
EleutherAI/pythia-410m-deduped
Text Generation • Updated • 22k • 20 -
EleutherAI/pythia-1b-deduped
Text Generation • Updated • 22.3k • 19 -
bigscience/bloomz-560m
Text Generation • Updated • 293k • 117 -
bigscience/mt0-base
Text2Text Generation • Updated • 3.76k • • 30 -
facebook/opt-350m
Text Generation • Updated • 970k • • 137 -
rajpurkar/squad
Viewer • Updated • 98.2k • 64k • 292 -
google-research-datasets/qed
Updated • 141 • 3 -
GEM/FairytaleQA
Viewer • Updated • 10.6k • 364 • 8 -
qiaojin/PubMedQA
Viewer • Updated • 274k • 6.53k • 176 -
knkarthick/dialogsum
Viewer • Updated • 14.5k • 17.3k • 194