step_benchmark_gpt / is_test_false_eval_results.txt
system's picture
system HF staff
Update is_test_false_eval_results.txt
b1a602b
raw
history blame contribute delete
152 Bytes
model =gpt2
total batch size=40
train num epochs=10
fp16 =True
max seq length =40
eval_acc = 0.836
eval_loss = 1.0890175614092086