Updated for prepackaged ct2 zip

make_triton_model_repo.sh

 #!/bin/bash
-wget -O models.zip "https://ssmt.iiit.ac.in/uploads/data_mining/models.zip" --no-check-certificate
-wget "https://cdn.iiit.ac.in/cdn/ssmt.iiit.ac.in/uploads/data_mining/HimangY-oneMT-Models-V1.zip" --no-check-certificate
-unzip models.zip
-unzip HimangY-oneMT-Models-V1.zip
-python3 -m venv ./ssmt_ct2
-source ./ssmt_ct2/bin/activate
-pip install -U pip wheel; pip install ctranslate2 "OpenNMT-py==1.2.0"
-cd HimangY-oneMT-Models-V1
-ct2-opennmt-py-converter --model_path 150001.pt --output_dir en-hi; mv 150001.src en-hi.src
-ct2-opennmt-py-converter --model_path 150002.pt --output_dir hi-en; mv 150002.src hi-en.src
-ct2-opennmt-py-converter --model_path 150003.pt --output_dir en-te; mv 150003.src en-te.src
-ct2-opennmt-py-converter --model_path 150004.pt --output_dir te-en; mv 150004.src te-en.src
-ct2-opennmt-py-converter --model_path 150005.pt --output_dir hi-te; mv 150005.src hi-te.src
-ct2-opennmt-py-converter --model_path 150006.pt --output_dir te-hi; mv 150006.src te-hi.src
-ct2-opennmt-py-converter --model_path 150007.pt --output_dir hi-ur; mv 150007.src hi-ur.src
-ct2-opennmt-py-converter --model_path 150008.pt --output_dir ur-hi; mv 150008.src ur-hi.src
-# ct2-opennmt-py-converter --model_path 150009.pt --output_dir hi-gu; mv 150009.src hi-gu.src
-ct2-opennmt-py-converter --model_path 150010.pt --output_dir gu-hi; mv 150010.src gu-hi.src
-# ct2-opennmt-py-converter --model_path 150011.pt --output_dir hi-pa; mv 150011.src hi-pa.src
-ct2-opennmt-py-converter --model_path 150013.pt --output_dir hi-or; mv 150013.src hi-or.src
-ct2-opennmt-py-converter --model_path 150014.pt --output_dir or-hi; mv 150014.src or-hi.src
-ct2-opennmt-py-converter --model_path 150015.pt --output_dir hi-ta; mv 150015.src hi-ta.src
-ct2-opennmt-py-converter --model_path 150017.pt --output_dir hi-kn; mv 150017.src hi-kn.src
-ct2-opennmt-py-converter --model_path 150018.pt --output_dir kn-hi; mv 150018.src kn-hi.src
-ct2-opennmt-py-converter --model_path 150019.pt --output_dir ta-te; mv 150019.src ta-te.src
-ct2-opennmt-py-converter --model_path 150020.pt --output_dir te-ta; mv 150020.src te-ta.src
-cd ../models
-ct2-opennmt-py-converter --model_path 1.pt --output_dir en-hi; mv 1.src en-hi.src
-ct2-opennmt-py-converter --model_path 2.pt --output_dir hi-en; mv 2.src hi-en.src
-# ct2-opennmt-py-converter --model_path 3.pt --output_dir en-te; mv 3.src en-te.src
-ct2-opennmt-py-converter --model_path 4.pt --output_dir te-en; mv 4.src te-en.src
-ct2-opennmt-py-converter --model_path 6.pt --output_dir hi-te; mv 6.src hi-te.src
-ct2-opennmt-py-converter --model_path 7.pt --output_dir te-hi; mv 7.src te-hi.src
-ct2-opennmt-py-converter --model_path 8.pt --output_dir en-gu; mv 8.src en-gu.src
-ct2-opennmt-py-converter --model_path 9.pt --output_dir gu-en; mv 9.src gu-en.src
-cd ..
+wget "https://ssmt.iiit.ac.in/uploads/data_mining/HimangY_MT_Models_v1.zip" --no-check-certificate
+unzip HimangY_MT_Models_v1.zip
 mkdir himangy_triton_repo
 cd himangy_triton_repo
 cp -r ../triton_models/nmt .
 cp -r ../triton_models/demuxer .
 cp -r ../triton_models/tokenizer .
-cp -r ../triton_models/model_ct2 himangy-en-hi;mv ../HimangY-oneMT-Models-V1/en-hi himangy-en-hi/1/translator;mv ../HimangY-oneMT-Models-V1/en-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-hi/' himangy-en-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'hi'/" himangy-en-hi/1/model.py
-cp -r ../triton_models/model_ct2 himangy-hi-en;mv ../HimangY-oneMT-Models-V1/hi-en himangy-hi-en/1/translator;mv ../HimangY-oneMT-Models-V1/hi-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-en/' himangy-hi-en/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'en'/" himangy-hi-en/1/model.py
-cp -r ../triton_models/model_ct2 himangy-en-te;mv ../HimangY-oneMT-Models-V1/en-te himangy-en-te/1/translator;mv ../HimangY-oneMT-Models-V1/en-te.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-te/' himangy-en-te/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'te'/" himangy-en-te/1/model.py
-cp -r ../triton_models/model_ct2 himangy-te-en;mv ../HimangY-oneMT-Models-V1/te-en himangy-te-en/1/translator;mv ../HimangY-oneMT-Models-V1/te-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-en/' himangy-te-en/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'en'/" himangy-te-en/1/model.py
-cp -r ../triton_models/model_ct2 himangy-hi-te;mv ../HimangY-oneMT-Models-V1/hi-te himangy-hi-te/1/translator;mv ../HimangY-oneMT-Models-V1/hi-te.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-te/' himangy-hi-te/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'te'/" himangy-hi-te/1/model.py
-cp -r ../triton_models/model_ct2 himangy-te-hi;mv ../HimangY-oneMT-Models-V1/te-hi himangy-te-hi/1/translator;mv ../HimangY-oneMT-Models-V1/te-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-hi/' himangy-te-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'hi'/" himangy-te-hi/1/model.py
-cp -r ../triton_models/model_ct2 himangy-hi-ur;mv ../HimangY-oneMT-Models-V1/hi-ur himangy-hi-ur/1/translator;mv ../HimangY-oneMT-Models-V1/hi-ur.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-ur/' himangy-hi-ur/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'ur'/" himangy-hi-ur/1/model.py
-cp -r ../triton_models/model_ct2 himangy-ur-hi;mv ../HimangY-oneMT-Models-V1/ur-hi himangy-ur-hi/1/translator;mv ../HimangY-oneMT-Models-V1/ur-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-ur-hi/' himangy-ur-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'ur', 'hi'/" himangy-ur-hi/1/model.py
-cp -r ../triton_models/model_onmt himangy-hi-gu;mv ../HimangY-oneMT-Models-V1/150009.pt himangy-hi-gu/1/translator.pt;mv ../HimangY-oneMT-Models-V1/150009.src tokenizer/1/bpe_src/hi-gu.src;sed -i 's/model_name/himangy-hi-gu/' himangy-hi-gu/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'gu'/" himangy-hi-gu/1/model.py
-cp -r ../triton_models/model_ct2 himangy-gu-hi;mv ../HimangY-oneMT-Models-V1/gu-hi himangy-gu-hi/1/translator;mv ../HimangY-oneMT-Models-V1/gu-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-gu-hi/' himangy-gu-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'gu', 'hi'/" himangy-gu-hi/1/model.py
-cp -r ../triton_models/model_onmt himangy-hi-pa;mv ../HimangY-oneMT-Models-V1/150011.pt himangy-hi-pa/1/translator.pt;mv ../HimangY-oneMT-Models-V1/150011.src tokenizer/1/bpe_src/hi-pa.src;sed -i 's/model_name/himangy-hi-pa/' himangy-hi-pa/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'pa'/" himangy-hi-pa/1/model.py
-cp -r ../triton_models/model_ct2 himangy-hi-or;mv ../HimangY-oneMT-Models-V1/hi-or himangy-hi-or/1/translator;mv ../HimangY-oneMT-Models-V1/hi-or.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-or/' himangy-hi-or/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'or'/" himangy-hi-or/1/model.py
-cp -r ../triton_models/model_ct2 himangy-or-hi;mv ../HimangY-oneMT-Models-V1/or-hi himangy-or-hi/1/translator;mv ../HimangY-oneMT-Models-V1/or-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-or-hi/' himangy-or-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'or', 'hi'/" himangy-or-hi/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-hi-ta;mv ../HimangY-oneMT-Models-V1/hi-ta himangy-hi-ta/1/translator;mv ../HimangY-oneMT-Models-V1/hi-ta.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-ta/' himangy-hi-ta/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'ta'/" himangy-hi-ta/1/model.py
-cp -r ../triton_models/model_ct2 himangy-hi-kn;mv ../HimangY-oneMT-Models-V1/hi-kn himangy-hi-kn/1/translator;mv ../HimangY-oneMT-Models-V1/hi-kn.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-kn/' himangy-hi-kn/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'kn'/" himangy-hi-kn/1/model.py
-cp -r ../triton_models/model_ct2 himangy-kn-hi;mv ../HimangY-oneMT-Models-V1/kn-hi himangy-kn-hi/1/translator;mv ../HimangY-oneMT-Models-V1/kn-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-kn-hi/' himangy-kn-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'kn', 'hi'/" himangy-kn-hi/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-ta-te;mv ../HimangY-oneMT-Models-V1/ta-te himangy-ta-te/1/translator;mv ../HimangY-oneMT-Models-V1/ta-te.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-ta-te/' himangy-ta-te/config.pbtxt;sed -i "s/input_lang, output_lang/'ta', 'te'/" himangy-ta-te/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-te-ta;mv ../HimangY-oneMT-Models-V1/te-ta himangy-te-ta/1/translator;mv ../HimangY-oneMT-Models-V1/te-ta.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-ta/' himangy-te-ta/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'ta'/" himangy-te-ta/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-en-hi;mv ../models/en-hi himangy-en-hi/1/translator;mv ../models/en-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-hi/' himangy-en-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'hi'/" himangy-en-hi/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-hi-en;mv ../models/hi-en himangy-hi-en/1/translator;mv ../models/hi-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-en/' himangy-hi-en/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'en'/" himangy-hi-en/1/model.py
-# cp -r ../triton_models/model_onnmt himangy-en-te;mv ../models/3.pt himangy-en-te/1/translator.pt;mv ../models/3.src tokenizer/1/bpe_src/en-te.src;sed -i 's/model_name/himangy-en-te/' himangy-en-te/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'te'/" himangy-en-te/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-te-en;mv ../models/te-en himangy-te-en/1/translator;mv ../models/te-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-en/' himangy-te-en/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'en'/" himangy-te-en/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-hi-te;mv ../models/hi-te himangy-hi-te/1/translator;mv ../models/hi-te.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-te/' himangy-hi-te/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'te'/" himangy-hi-te/1/model.py
-# cp -r ../triton_models/model_ct2 himangy-te-hi;mv ../models/te-hi himangy-te-hi/1/translator;mv ../models/te-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-hi/' himangy-te-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'hi'/" himangy-te-hi/1/model.py
-cp -r ../triton_models/model_ct2 himangy-en-gu;mv ../models/en-gu himangy-en-gu/1/translator;mv ../models/en-gu.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-gu/' himangy-en-gu/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'gu'/" himangy-en-gu/1/model.py
-cp -r ../triton_models/model_ct2 himangy-gu-en;mv ../models/gu-en himangy-gu-en/1/translator;mv ../models/gu-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-gu-en/' himangy-gu-en/config.pbtxt;sed -i "s/input_lang, output_lang/'gu', 'en'/" himangy-gu-en/1/model.py
+cp -r ../triton_models/model_ct2 himangy-en-hi;mv ../HimangY_MT_Models_v1/en-hi himangy-en-hi/1/translator;mv ../HimangY_MT_Models_v1/en-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-hi/' himangy-en-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'hi'/" himangy-en-hi/1/model.py
+cp -r ../triton_models/model_ct2 himangy-hi-en;mv ../HimangY_MT_Models_v1/hi-en himangy-hi-en/1/translator;mv ../HimangY_MT_Models_v1/hi-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-en/' himangy-hi-en/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'en'/" himangy-hi-en/1/model.py
+cp -r ../triton_models/model_ct2 himangy-en-te;mv ../HimangY_MT_Models_v1/en-te himangy-en-te/1/translator;mv ../HimangY_MT_Models_v1/en-te.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-te/' himangy-en-te/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'te'/" himangy-en-te/1/model.py
+cp -r ../triton_models/model_ct2 himangy-te-en;mv ../HimangY_MT_Models_v1/te-en himangy-te-en/1/translator;mv ../HimangY_MT_Models_v1/te-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-en/' himangy-te-en/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'en'/" himangy-te-en/1/model.py
+cp -r ../triton_models/model_ct2 himangy-hi-te;mv ../HimangY_MT_Models_v1/hi-te himangy-hi-te/1/translator;mv ../HimangY_MT_Models_v1/hi-te.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-te/' himangy-hi-te/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'te'/" himangy-hi-te/1/model.py
+cp -r ../triton_models/model_ct2 himangy-te-hi;mv ../HimangY_MT_Models_v1/te-hi himangy-te-hi/1/translator;mv ../HimangY_MT_Models_v1/te-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-te-hi/' himangy-te-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'te', 'hi'/" himangy-te-hi/1/model.py
+cp -r ../triton_models/model_ct2 himangy-hi-ur;mv ../HimangY_MT_Models_v1/hi-ur himangy-hi-ur/1/translator;mv ../HimangY_MT_Models_v1/hi-ur.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-ur/' himangy-hi-ur/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'ur'/" himangy-hi-ur/1/model.py
+cp -r ../triton_models/model_ct2 himangy-ur-hi;mv ../HimangY_MT_Models_v1/ur-hi himangy-ur-hi/1/translator;mv ../HimangY_MT_Models_v1/ur-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-ur-hi/' himangy-ur-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'ur', 'hi'/" himangy-ur-hi/1/model.py
+cp -r ../triton_models/model_onmt himangy-hi-gu;mv ../HimangY_MT_Models_v1/hi-gu.pt himangy-hi-gu/1/translator.pt;mv ../HimangY_MT_Models_v1/hi-gu.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-gu/' himangy-hi-gu/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'gu'/" himangy-hi-gu/1/model.py
+cp -r ../triton_models/model_ct2 himangy-gu-hi;mv ../HimangY_MT_Models_v1/gu-hi himangy-gu-hi/1/translator;mv ../HimangY_MT_Models_v1/gu-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-gu-hi/' himangy-gu-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'gu', 'hi'/" himangy-gu-hi/1/model.py
+cp -r ../triton_models/model_onmt himangy-hi-pa;mv ../HimangY_MT_Models_v1/hi-pa.pt himangy-hi-pa/1/translator.pt;mv ../HimangY_MT_Models_v1/hi-pa.src tokenizer/1/bpe_src/hi-pa.src;sed -i 's/model_name/himangy-hi-pa/' himangy-hi-pa/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'pa'/" himangy-hi-pa/1/model.py
+cp -r ../triton_models/model_ct2 himangy-hi-or;mv ../HimangY_MT_Models_v1/hi-or himangy-hi-or/1/translator;mv ../HimangY_MT_Models_v1/hi-or.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-hi-or/' himangy-hi-or/config.pbtxt;sed -i "s/input_lang, output_lang/'hi', 'or'/" himangy-hi-or/1/model.py
+cp -r ../triton_models/model_ct2 himangy-or-hi;mv ../HimangY_MT_Models_v1/or-hi himangy-or-hi/1/translator;mv ../HimangY_MT_Models_v1/or-hi.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-or-hi/' himangy-or-hi/config.pbtxt;sed -i "s/input_lang, output_lang/'or', 'hi'/" himangy-or-hi/1/model.py
+cp -r ../triton_models/model_ct2 himangy-en-gu;mv ../HimangY_MT_Models_v1/en-gu himangy-en-gu/1/translator;mv ../HimangY_MT_Models_v1/en-gu.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-en-gu/' himangy-en-gu/config.pbtxt;sed -i "s/input_lang, output_lang/'en', 'gu'/" himangy-en-gu/1/model.py
+cp -r ../triton_models/model_ct2 himangy-gu-en;mv ../HimangY_MT_Models_v1/gu-en himangy-gu-en/1/translator;mv ../HimangY_MT_Models_v1/gu-en.src tokenizer/1/bpe_src;sed -i 's/model_name/himangy-gu-en/' himangy-gu-en/config.pbtxt;sed -i "s/input_lang, output_lang/'gu', 'en'/" himangy-gu-en/1/model.py
 cd ..
-deactivate
-rm -rf ssmt_ct2;rm -rf HimangY-oneMT-Models-V1;rm -rf models;rm -f models.zip;rm -f HimangY-oneMT-Models-V1.zip
\ No newline at end of file
+rm -rf HimangY_MT_Models_v1
+rm -f HimangY_MT_Models_v1.zip

triton_models/model_ct2/1/model.py

@@ -58,6 +58,8 @@ class TritonPythonModel:
                 max_batch_size=128,
                 max_input_length=100,
                 max_decoding_length=100,
+                beam_size=15,
+                replace_unknowns=True,
             )
         ]
         responses = [

triton_models/model_onmt/1/model.py

@@ -23,7 +23,7 @@ class TritonPythonModel:
                     tgt_prefix=False,
                     alpha=0.0,
                     batch_type="sents",
-                    beam_size=5,
+                    beam_size=15,
                     beta=-0.0,
                     block_ngram_repeat=0,
                     coverage_penalty="none",
@@ -43,7 +43,7 @@ class TritonPythonModel:
                     random_sampling_temp=1.0,
                     random_sampling_topk=1,
                     ratio=-0.0,
-                    replace_unk=False,
+                    replace_unk=True,
                     report_align=False,
                     report_time=False,
                     seed=829,
@@ -59,7 +59,7 @@ class TritonPythonModel:
                     tgt_prefix=False,
                     alpha=0.0,
                     batch_type="sents",
-                    beam_size=5,
+                    beam_size=15,
                     beta=-0.0,
                     block_ngram_repeat=0,
                     coverage_penalty="none",
@@ -79,7 +79,7 @@ class TritonPythonModel:
                     random_sampling_temp=1.0,
                     random_sampling_topk=1,
                     ratio=-0.0,
-                    replace_unk=False,
+                    replace_unk=True,
                     report_align=False,
                     report_time=False,
                     seed=829,

ulca_model.json

@@ -88,18 +88,6 @@
             "targetLanguage": "hi",
             "targetScriptCode": "Deva"
         },
-        {
-            "sourceLanguage": "hi",
-            "sourceScriptCode": "Deva",
-            "targetLanguage": "kn",
-            "targetScriptCode": "Knda"
-        },
-        {
-            "sourceLanguage": "kn",
-            "sourceScriptCode": "Knda",
-            "targetLanguage": "hi",
-            "targetScriptCode": "Deva"
-        },
         {
             "sourceLanguage": "en",
             "sourceScriptCode": "Latn",
@@ -156,4 +144,4 @@
             }
         ]
     }
-}
\ No newline at end of file
+}