metadata

language:
  - en
  - hi
  - multilingual
license: cc-by-4.0
tags:
  - translation
  - opus-mt-tc
model-index:
  - name: opus-mt-tc-base-en-hi
    results:
      - task:
          type: translation
          name: Translation eng-hin
        dataset:
          name: tatoeba-test-v2021-08-07
          type: tatoeba_mt
          args: eng-hin
        metrics:
          - type: bleu
            value: 22.2
            name: BLEU

Opus Tatoeba English-Hindi

This model was obtained by running the script convert_marian_to_pytorch.py. The original models were trained by J�rg Tiedemann using the MarianNMT library. See all available MarianMTModel models on the profile of the Helsinki NLP group.

dataset: opus+bt
model: transformer-align
source language(s): eng
target language(s): hin
model: transformer-align
pre-processing: normalization + SentencePiece (spm32k,spm32k)
download: opus+bt-2021-04-10.zip
test set translations: opus+bt-2021-04-10.test.txt
test set scores: opus+bt-2021-04-10.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newsdev2014.eng-hin	13.9	0.421	520	9538	1.000
newstest2014-hien.eng-hin	17.4	0.442	2507	60878	0.989
Tatoeba-test.eng-hin	22.2	0.485	5000	32904	1.000
tico19-test.eng-hin	30.6	0.539	2100	62738	0.988