issues
search
mozilla
/
translations
The code, training pipeline, and models that power Firefox Translations
https://mozilla.github.io/translations/
Mozilla Public License 2.0
155
stars
34
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
fix: get rid of `hunspell` python package
#938
bhearsum
opened
2 hours ago
1
Cjk corpora fixes
#937
ZJaume
opened
1 day ago
0
Check for float16 precision support when running translate-* tasks
#936
gregtatum
opened
1 day ago
0
[skip ci] Fix typo in the rebuild docker-images/toolchains docs
#935
gabrielBusta
closed
1 day ago
0
Adjust default values for batching
#934
gregtatum
closed
1 day ago
2
decoding-teacher config property is not being used in translate.sh or translate-nbest.sh
#933
gregtatum
opened
4 days ago
0
Linters needs to ignore node_modules
#932
gregtatum
opened
1 week ago
0
Experiment with distillation data inference
#931
gregtatum
opened
1 week ago
2
Add a tsconfig.json file for JS code within this repository
#930
nordzilla
opened
1 week ago
0
Use PyMarian for COMET evaluations
#929
marco-c
opened
1 week ago
1
Single-side deduplication
#928
ZJaume
opened
1 week ago
1
Test WASM Translations in CI
#927
nordzilla
closed
1 week ago
2
Ctranslate2 ci 2
#926
gregtatum
opened
1 week ago
0
CI Run check
#925
gregtatum
closed
2 weeks ago
0
Create an `analyze-datasets` step in the pipeline
#924
gregtatum
opened
2 weeks ago
3
Investigate merging document sentences in HPLT
#923
eu9ene
opened
2 weeks ago
3
Ctranslate2 draft
#922
gregtatum
opened
2 weeks ago
0
Make `npm` available to `local` and `inference` docker images
#921
nordzilla
closed
2 weeks ago
0
Setup WASM test infrastructure for CI
#920
nordzilla
closed
1 week ago
1
Add --run-as-user flag to docker-run.py
#919
nordzilla
closed
2 weeks ago
0
Add emsdk as a git submodule
#918
nordzilla
closed
2 weeks ago
0
Add better support for reporting training continuation values
#917
gregtatum
opened
2 weeks ago
0
Rename docker tags following repository rename
#916
nordzilla
closed
3 weeks ago
1
Reduce monolingual data for en-lt to investigate distillation performance
#915
gregtatum
opened
3 weeks ago
1
Rename repo
#914
gregtatum
closed
3 weeks ago
2
Allow for split vocabs
#913
gregtatum
opened
3 weeks ago
1
[meta] Kick off a 2024-H2 training run
#912
gregtatum
opened
3 weeks ago
0
Do not use WMTNews as training!
#911
ZJaume
closed
2 weeks ago
3
More corpora specific fixes
#910
ZJaume
opened
3 weeks ago
0
Fix shortlist pruning for CJK
#909
eu9ene
closed
2 weeks ago
0
Switch bestbleu to chrF
#908
eu9ene
closed
2 weeks ago
0
Use GCP standard instances for alignment tasks
#907
eu9ene
closed
3 weeks ago
1
Configure vocab for CJK
#906
eu9ene
closed
2 weeks ago
2
Limit the amount of data used for distillation
#905
gregtatum
opened
3 weeks ago
3
Update training to support CJK
#904
eu9ene
closed
2 weeks ago
3
Check if issues with short sentences were caused by bicleaner hard rules
#903
eu9ene
opened
4 weeks ago
0
Rework wasm build scripts for gecko
#902
nordzilla
closed
3 weeks ago
2
Remove max_words filtering from data importers
#901
eu9ene
closed
2 weeks ago
4
Adjust data cleaning for CJK
#900
eu9ene
closed
2 weeks ago
1
Investigate word-based filtering for CJK
#899
eu9ene
opened
4 weeks ago
1
Run dhat or similar memory tools on a native built version of the the browsermt marian-dev fork
#898
gregtatum
opened
4 weeks ago
0
Update data importer to support CJK
#897
eu9ene
closed
2 weeks ago
1
Add support for Chinese Traditional
#896
eu9ene
opened
1 month ago
0
Fix taskcluster train scripts
#895
eu9ene
closed
1 month ago
0
Experiment with student model parameters
#894
gregtatum
opened
1 month ago
7
Student training continuation is regressed
#893
gregtatum
closed
1 month ago
3
Disable bilceaner hard rules completely
#892
eu9ene
closed
4 weeks ago
0
[meta] Retrain older models
#891
eu9ene
opened
1 month ago
1
Fine-tune students with 8-bit
#890
ZJaume
closed
3 weeks ago
3
Consider adding NTREX-128 for evaluation
#889
ZJaume
opened
1 month ago
0
Next