darkforest go training

th train.lua –nGPU 1 –datasource kgs –num_forward_models 512 –nthread 4 –alpha 0.05 –epoch_size 128000 –data_augmentation –batchsize 64

it finished after several days, with score ~53% for top 1 and ~85% for top5.

stared ~10pm 12/11 and finished ~11pm 12/13. 173 epochs. 17mins / epoch on average.

in epoch 60, it can achieve ~50% for top1 and ~82 for top5.

in epoch 47, ~47% for top1 and ~80% for top5.

from https://github.com/facebookresearch/darkforestGo/tree/master/train

With 4 GPUs, the training procedure gives 56.1% top-1 accuracy in KGS dataset in 3.5 days, and 57.1% top-1 in 6.5 days (see the simple log below). The parameters used are the following: –epoch_size 256000 –GPU 4 –data_augmentation –alpha 0.1 –nthread 4

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s