Asilarknes
/

testoneshot

Model card Files Files and versions

testoneshot / README.md

Asilarknes's picture

continue readout step 500 ppl 92.75

31f9ceb verified about 17 hours ago

|

history blame contribute delete

189 Bytes

OneShot continued full model

Latest continued checkpoint:

file: full_model_continue.pt
step: 500
ppl: 92.7529
final: False

Base config:

d=896
r=320
layers=10
vocab=8192