An inclusive natural language to code library built over allennlp 2.
This repository is using code from the following resources:
In this repo, we adapt the code to the latest allennlp version.
PYTHONPATH=src ANNOTATION_DIR=$PWD/data/annotations/atis/ \ allennlp train configs/atis/seq2seq/defaults.jsonnet \ -s data/output/atis/seq2seq --include-package andushuPYTHONPATH=src ANNOTATION_DIR=$PWD/data/annotations/atis/ \ allennlp predict \ --output-file=data/output/atis/seq2seq/seq2seq.jsonl \ --predictor seq2seq \ --include-package andushu \ data/output/atis/seq2seq/model.tar.gz \ data/annotations/atis/atis_test.jsonl PYTHONPATH=src ANNOTATION_DIR=$PWD/data/annotations/geoquery/ \ allennlp train configs/geoquery/seq2seq/defaults.jsonnet \ -s data/output/geoquery/seq2seq --include-package andushuPYTHONPATH=src ANNOTATION_DIR=$PWD/data/annotations/geoquery/ \ allennlp predict \ --output-file=data/output/geoquery/seq2seq/seq2seq.jsonl \ --predictor seq2seq \ --include-package andushu data/output/geoquery/seq2seq/model.tar.gz \ data/annotations/geoquery/geo_test.jsonl This repo includes the code for
@article{Tan2021InvestigatingMW, title={Investigating Math Word Problems using Pretrained Multilingual Language Models}, author={Minghuan Tan and Lei Wang and Lingxiao Jiang and Jing Jiang}, journal={ArXiv}, year={2021}, volume={abs/2105.08928} }Annotations used for this paper can be found at
Training Math23K using bert-base-multilingual-cased.
CUDA_VISIBLE_DEVICES=2 PROJECT=math2tree SUB_PROJECT=math23k CONFIG=transformer_vocab SPACY_LANGUAGE=zh MODEL_NAME=bert-base-multilingual-cased OP_TYPE=disallow_pow bash docker_train.shTraining over MathQA-Adapted without Pow.
CUDA_VISIBLE_DEVICES=2 PROJECT=math2tree SUB_PROJECT=mathqa CONFIG=transformer_vocab SPACY_LANGUAGE=zh MODEL_NAME=bert-base-multilingual-cased OP_TYPE=disallow_pow bash docker_train.shTraining over MathQA-Adapted with Pow.
CUDA_VISIBLE_DEVICES=2 PROJECT=math2tree SUB_PROJECT=mathqa CONFIG=transformer_vocab SPACY_LANGUAGE=zh MODEL_NAME=bert-base-multilingual-cased OP_TYPE=allow_pow bash docker_train.shTraining over MathXLing without Pow.
CUDA_VISIBLE_DEVICES=2 PROJECT=math2tree SUB_PROJECT=mathxling CONFIG=transformer_vocab SPACY_LANGUAGE=zh MODEL_NAME=bert-base-multilingual-cased OP_TYPE=disallow_pow bash docker_train.shTraining over MathXLing with Pow.
CUDA_VISIBLE_DEVICES=2 PROJECT=math2tree SUB_PROJECT=mathxling CONFIG=transformer_vocab SPACY_LANGUAGE=zh MODEL_NAME=bert-base-multilingual-cased OP_TYPE=allow_pow bash docker_train.shCUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/mathqa/transformer_vocab_en_disallow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/mathqa/transformer_vocab_en_allow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/mathqa/transformer_vocab_en_disallow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/math23k/transformer_vocab_zh_disallow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/math23k/transformer_vocab_zh_allow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/math23k/transformer_vocab_en_allow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/math23k/transformer_vocab_en_disallow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/mathxling/transformer_vocab_zh_sallow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/mathxling/transformer_vocab_zh_allow_pow_bert-base-multilingual-cased evaluate CUDA_VISIBLE_DEVICES=7 bash docker_eval.sh math2tree data/output/mathxling/transformer_vocab_zh_allow_pow_xlm-roberta-base evaluate