Add missing arg in 02-transformers notebook

[pri-ax]

Add missing from_tf=True arg when creating the model AutoModel.from_pretrained() to avoid an OSError from loading a PyTorch model from a TF 2.0 checkpoint.

Also fixed two small typos in markdown text.

[codecov]

Codecov Report

Merging #5085 into master will decrease coverage by 0.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #5085      +/-   ##
==========================================
- Coverage   77.24%   77.22%   -0.02%     
==========================================
  Files         133      133              
  Lines       22146    22146              
==========================================
- Hits        17107    17103       -4     
- Misses       5039     5043       +4     

Impacted Files	Coverage Δ
src/transformers/modeling_tf_utils.py	85.86% <0.00%> (-0.78%)	⬇️
src/transformers/modeling_utils.py	90.61% <0.00%> (-0.12%)	⬇️
src/transformers/modeling_transfo_xl.py	78.81% <0.00%> (+0.19%)	⬆️
src/transformers/file_utils.py	75.40% <0.00%> (+0.40%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 7291ea0...0092c55. Read the comment docs.

[LysandreJik]

Hi, I agree with the typo change, but there's no need for the from_tf flag. bert-base-cased is available in both TensorFlow and PyTorch, so there's no need for the flag.

[pri-ax]

@LysandreJik you're right! I was getting an error unless I explicitly passed from_tf=True yesterday, but not today. Since I can't reproduce it, I'll remove that change.

[LysandreJik]

Great, thanks!