I want to know what models files are

[xiaoshuguo750]

Hi!

I want to know what models files are,and how can I use them?

Thank you very much!

[tnikolla]

Hi,

these are files with the variables trained (optimized) from imagenet classification algorithm and grasp-detection algorithm. You can use them by giving them as arguments, for example:

./imagenet_classifier.py --batch_size=128 --model_path=./models/imagenet/m1/m1.ckpt --train_or_validation=train

Give your path to --model_path=your/path/to/the/ckpt_file.

For a better explanation of what they are and how to use them see tensorflow guide.

[HEUzhouhanwen]

Thank you for your answer! When I run the following command（using Cornell Grasping Dataset）, there are some problems（Step 23 | loss = nan）, what is the reason? What should I do?

./grasp_det.py --batch_size=64 --model_path=./models/imagenet/m2/m2.ckpt --train_or_validation=train some mistake happened!what's wrong?

Step 21 | loss = 177726.0 | x = [ 92.57499695 107.09999847 77.875 ] | x_hat = [ 71.45228577 70.01683807 78.42079163] | tan = [ 0.04202652 -0.11677019 -8.66753483] | tan_hat = [-15.22892761 7.36084843 -13.21743774] | h = [ 18.9166832 16.20939255 33.59136581] | h_hat = [ 8.36284542 10.77396488 17.70290947] | w = [ 15.04407215 9.35572624 12.16120529] | w_hat = [ 21.95122147 11.86730862 8.71802139] | (0.983 sec/batch

Step 22 | loss = 145800.0 | x = [ 113.22499847 101.32499695 111.29999542] | x_hat = [ 88.18299866 69.44770813 81.33192444] | tan = [-1.64852488 -inf -0.33319956] | tan_hat = [ 1.98662510e+01 -2.18759489e+00 -1.49553716e-02] | h = [ 26.99363327 19.73999977 20.29046822] | h_hat = [ 17.33011055 9.94925213 21.10087013] | w = [ 10.8564539 6.64999962 13.38611507] | w_hat = [ 19.93681526 0.93864435 17.20625496] | (0.963 sec/batch

Step 23 | loss = nan | x = [ 94.84999847 93.79999542 122.5 ] | x_hat = [ 89.2689743 98.40085602 95.08123016] | tan = [ 1.84252501 -9.06495476 -0. ] | tan_hat = [-6.25003433 3.59752798 -0.40878177] | h = [ 31.55083656 19.15188408 6.29999876] | h_hat = [ 24.60993385 24.8934269 7.01266623] | w = [ 11.33346653 9.29216862 3.75999999] | w_hat = [ 26.53094292 20.04769325 22.91687965] | (0.975 sec/batch

Step 24 | loss = nan | x = [ 107.97499847 108.5 99.57499695] | x_hat = [ nan nan nan] | tan = [-0.38987583 0.43599257 -1.42756033] | tan_hat = [ nan nan nan] | h = [ 11.26979828 29.40008545 31.7219696 ] | h_hat = [ nan nan nan] | w = [ 8.17940044 31.39706039 10.76076698] | w_hat = [ nan nan nan] | (0.964 sec/batch

Thank you very much!

[tnikolla]

Hi,

check first issue #4 .

[xiaoshuguo750]

Thank you！

At 2018-01-01 18:00:59, "Tomi Nikolla" notifications@github.com wrote:

Hi,

check first issue #4 .

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[zhang-lu153]

Hi, Can you give your ./models/imagenet/m1/m1.ckpt ?I want to know what this models file is My email is zl@hnu.edu.cn . Thank you!

[jinhuan-hit]

@lubao1995 You can see ./models/imagenet/m2/m2.ckpt in your path，not m1.