[doc] Summary of the models fixes

[stas00]

• improve readability - typos, punctuation, clearer sentences

2 questions:

1. https://huggingface.co/transformers/model_summary.html#t5

I think the example is incorrect in the target part of the example. I think it's missing "cute".

Before:

For instance, if we have the sentence “My dog is very cute .”, and we decide to remove the token dog, is and cute, the input becomes “My very .” and the target is “ dog is . ”

Proposed change:

For instance, if we have the sentence “My dog is very cute .”, and we decide to remove the tokens: "dog", "is" and "cute", the encoder input becomes “My very .” and the target input becomes “ dog is cute .”

At https://huggingface.co/transformers/model_summary.html#full-vs-sparse-attention

• in the "LSH attention" section:

"The attention mask is modified to mask the current token (except at the first position) because it will give a query and key equal (so very similar to each other). "

It's missing a word at the end of the sentence. Is it "equal attention"?

• Also in the "Local attention" just after the previous section, it goes:

"This is shown in Figure 2d of the paper"

but there is no link or name of the paper it's referring to.

@sgugger

[codecov]

Codecov Report

Merging #6511 into master will decrease coverage by 0.43%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #6511      +/-   ##
==========================================
- Coverage   80.37%   79.94%   -0.44%     
==========================================
  Files         156      156              
  Lines       28058    28058              
==========================================
- Hits        22552    22431     -121     
- Misses       5506     5627     +121     

Impacted Files	Coverage Δ
src/transformers/configuration_utils.py	95.91% <ø> (-0.69%)	⬇️
src/transformers/generation_tf_utils.py	86.21% <ø> (-0.26%)	⬇️
src/transformers/generation_utils.py	96.94% <ø> (ø)
src/transformers/trainer.py	37.84% <0.00%> (ø)
src/transformers/modeling_tf_openai.py	22.58% <0.00%> (-72.26%)	⬇️
src/transformers/modeling_tf_flaubert.py	24.53% <0.00%> (-63.81%)	⬇️
src/transformers/tokenization_pegasus.py	45.31% <0.00%> (-50.00%)	⬇️
src/transformers/tokenization_marian.py	66.66% <0.00%> (-32.50%)	⬇️
src/transformers/tokenization_reformer.py	51.66% <0.00%> (-30.00%)	⬇️
src/transformers/tokenization_auto.py	95.55% <0.00%> (-2.23%)	⬇️
... and 10 more

---

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 24107c2...78bd9e3. Read the comment docs.