booper
· runs
live training runs — each point is a step the trainer reported. refreshes every 15s.
run
refresh
loss
train
train (avg 5)
val
throughput (tokens / s)
learning rate
latest samples
eval table