Implemented fixes in 2:4 sparsity tutorial

[lvoursl]

Description

I was running this tutorial on 2:4 sparsity and found a few issues that I'm fixing in this PR. More concrete:

1. Block with the model definition wasn't rendered on the website since there was some stop symbol in source markdown.
2. One variable wasn't defined in the code and one comma was missed in another place -- so some code wasn't working.
3. Model didn't converge to the same level of performance after pruning (gap was substantial in F1 metric). The problem seems to be related to the learning rate schedule in HuggingFace Transformers: during fine-tuning of pruned model learning rate went to 0 super quickly and the model params were not updating at all. To fix it I simply set LR schedule to constant, so we have consistent results in that case.

cc: @jcaip

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/2616

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit b804be4 with merge base b51f2b8 ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[jcaip]

lgtm, thank you for the changes!