Fix global_step variable definition

[arvind]

No description provided.

[googlebot]

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed (or fixed any issues), please reply here (e.g. I signed it!) and we'll verify it.

---

What to do if you already signed the CLA

Individual signers

• It's possible we don't have your GitHub username or you're using a different email address on your commit. Check your existing CLA data and verify that your email is set on your git commits.

Corporate signers

• Your company has a Point of Contact who decides which employees are authorized to participate. Ask your POC to be added to the group of authorized contributors. If you don't know who your Point of Contact is, direct the project maintainer to go/cla#troubleshoot.
• The email used to register you as an authorized contributor must be the email used for the Git commit. Check your existing CLA data and verify that your email is set on your git commits.
• The email used to register you as an authorized contributor must also be attached to your GitHub account.

[coveralls]

Pull Request Test Coverage Report for Build 84

• 1 of 1 (100.0%) changed or added relevant line in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 61.934%

---

Totals
Change from base Build 79:	0.0%
Covered Lines:	820
Relevant Lines:	1324

---

💛 - Coveralls

[googlebot]

CLAs look good, thanks!

[ludwigschubert]

This was implemented in #59.
Sorry we couldn't merge this PR for that—the global_step tensor also needed initialization.

[arvind]

Apologies for not documenting the motivation for this PR, but I don't think #59 fixes the issue that this PR was trying to tackle: global_step is not available in the T function passed to the objectives functions.

Thus, even after #59, the following custom decay objective fails:

@objectives.wrap_objective
def decay(n_steps):
  def inner(T):
    return 1 - (T('global_step') / n_steps)
  return inner

with the following error:

KeyError: "The name 'import/global_step:0' refers to a Tensor which does not exist. The operation, 'import/global_step', does not exist in the graph."

Is there an alternate way to write the decay objective?

[ludwigschubert]

As discussed in person with @arvind:

• Current workaround is to directly use tf.train.get_or_create_global_step(). Objective functions get evaluated within a scope that has the default session and graph available.
• We should rethink T in general, especially the special cases such as input. I think its plausible we don't need T at all, but should probably discuss with everyone first. :-)