sillsdev / sillsdev/silnlp

1 Cor. 1:14 contains unexpected output that does not match source verse

Open
#666 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug needs to be revisited pipeline 6: infer
Dominant language
Python
Stars
39
Forks
7
Avg merge
1d 9h
Merged PRs (30d)
5

Description

1 Cor 1:14 includes very unexpected output in the bep-to-Indonesian back translation, despite the team being generally happy with the output (GEN, RUT, 1-2CO) otherwise. Model scores look good: BLEU 52.83, chrf3++ 68.55.

Of note is that the training set in almost entirely OT, with the exception of Eph 1-4.

bep (source) extract:

Rasiꞌnanto i oloꞌmi tiꞌarakou ara to kuriu, bateꞌraꞌ peá Kirisipu hai Gaiu,

In English (NIV),

I thank God that I did not baptize any of you except Crispus and Gaius,

Indonesian (target) inference:

Kepada pemimpin penyanyi. Dinyanyikan menurut lagu "Bunga Bakung." Lagu yang dikarang Daudiꞌ.
Translated Ind-Eng by Google Translate:

To the leader of the singers. Sung to the tune of "Daffodils." Song composed by Daudiꞌ.

The config for this model and the inference are in S:\MT\experiments\Indonesia\Behoa\NLLB.1.3B.bep-ABe.id-TBBe.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the model configuration and inference referenced at S:\MT\experiments\Indonesia\Behoa\NLLB.1.3B.bep-ABe.id-TBBe, then compare the source verse, target inference, and training-set coverage described in the issue. Reproduce the unexpected 1 Cor. 1:14 output and document the cause or a confirmed reproduction path.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.