[RavSinghChandan]

decode_sequence rejected an index at or above the mapping size, but a negative one was read from the end of the mapping. The default EOS of encode_sequences is -1, so decode_sequence(encode_sequences(["ab"], "abc")[0], "abc") returned "abc": the EOS came back as the last char of the vocab, with no error. Lists were not checked at all, so [2, 10] raised a bare IndexError.

Negative indices now raise the same AssertionError as the indices that are too large, for arrays and lists. The recognition post-processors pass non-negative argmax indices, so they are not affected (the post-processor tests pass)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.75%. Comparing base (9ae237d) to head (21ff071).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2189   +/-   ##
=======================================
  Coverage   97.75%   97.75%           
=======================================
  Files         170      170           
  Lines       10959    10961    +2     
=======================================
+ Hits        10713    10715    +2     
  Misses        246      246           

Flag	Coverage Δ
unittests	97.75% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[felixdittrich92]

Thanks 👍