s2s distillation uses AutoModelForSeqToSeqLM

[sshleifer]

hard coding bart caused subsequent AutoTokenizer to fail for marian/mbart distillation.

[codecov]

Codecov Report

Merging #6761 into master will decrease coverage by 0.24%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #6761      +/-   ##
==========================================
- Coverage   79.70%   79.45%   -0.25%     
==========================================
  Files         157      157              
  Lines       28479    28479              
==========================================
- Hits        22698    22627      -71     
- Misses       5781     5852      +71     

Impacted Files	Coverage Δ
src/transformers/modeling_tf_electra.py	25.13% <0.00%> (-73.83%)	⬇️
src/transformers/modeling_tf_openai.py	22.58% <0.00%> (-72.26%)	⬇️
src/transformers/tokenization_xlnet.py	66.66% <0.00%> (-23.43%)	⬇️
src/transformers/tokenization_ctrl.py	78.64% <0.00%> (-17.48%)	⬇️
src/transformers/tokenization_reformer.py	81.66% <0.00%> (-13.34%)	⬇️
src/transformers/tokenization_openai.py	71.21% <0.00%> (-11.37%)	⬇️
src/transformers/modeling_t5.py	77.63% <0.00%> (-6.21%)	⬇️
src/transformers/modeling_openai.py	80.96% <0.00%> (-1.30%)	⬇️
src/transformers/modeling_utils.py	87.50% <0.00%> (-0.56%)	⬇️
src/transformers/modeling_tf_utils.py	86.97% <0.00%> (-0.33%)	⬇️
... and 9 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 05e7150...88974cc. Read the comment docs.