Datasets:

Modalities:
Text
Formats:
parquet
Languages:
English
ArXiv:
Tags:
License:
prompteus commited on
Commit
8078537
1 Parent(s): e57df94

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +23 -2
README.md CHANGED
@@ -92,10 +92,13 @@ The answers in the original dataset were in a structured but non-standard format
92
  were evaluated using a sympy-based calculator, the outputs were checked to be consistent with the intermediate results and exported
93
  into a simple html-like language that BeautifulSoup can parse.
94
 
 
 
 
95
 
96
  ## Content and Data splits
97
 
98
- For convenience, we created a validation set by sampling 200 random examples from the original train split. You can load this using:
99
 
100
  ```python3
101
  datasets.load_dataset("MU-NLPC/Calc-gsm8k")
@@ -110,6 +113,24 @@ datasets.load_dataset("MU-NLPC/Calc-gsm8k", "original-splits")
110
  For more info about the content of the dataset, see [gsm8k HF dataset](https://huggingface.co/datasets/gsm8k) and the [official repository](https://github.com/openai/grade-school-math).
111
 
112
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
113
  ## Licence
114
 
115
  MIT, consistently with the original dataset.
@@ -124,7 +145,7 @@ If you use this version of the dataset in research, please cite the [original GS
124
  title = "Calc-X and Calcformers: Empowering Arithmetical Chain-of-Thought through Interaction with Symbolic Systems",
125
  author = "Marek Kadlčík and Michal Štefánik and Ondřej Sotolář and Vlastimil Martinek",
126
  booktitle = "Proceedings of the The 2023 Conference on Empirical Methods in Natural Language Processing: Main track",
127
- month = december,
128
  year = "2023",
129
  address = "Singapore, Singapore",
130
  publisher = "Association for Computational Linguistics",
 
92
  were evaluated using a sympy-based calculator, the outputs were checked to be consistent with the intermediate results and exported
93
  into a simple html-like language that BeautifulSoup can parse.
94
 
95
+ We also perform in-dataset and cross-dataset data-leak detection within the [Calc-X collection](https://huggingface.co/collections/MU-NLPC/calc-x-652fee9a6b838fd820055483)
96
+ However, in case of gsm8k, we found no data leaks and removed no examples from the data.
97
+
98
 
99
  ## Content and Data splits
100
 
101
+ For convenience, we created a validation set by sampling 200 random examples from the original train split. This is the default variant:
102
 
103
  ```python3
104
  datasets.load_dataset("MU-NLPC/Calc-gsm8k")
 
113
  For more info about the content of the dataset, see [gsm8k HF dataset](https://huggingface.co/datasets/gsm8k) and the [official repository](https://github.com/openai/grade-school-math).
114
 
115
 
116
+
117
+ ## Related work
118
+
119
+ This dataset was created as a part of a larger effort in training models capable of using a calculator during inference, which we call Calcformers.
120
+
121
+ - [**Calc-X collection**](https://huggingface.co/collections/MU-NLPC/calc-x-652fee9a6b838fd820055483) - datasets for training Calcformers
122
+ - [**Calcformers collection**](https://huggingface.co/collections/MU-NLPC/calcformers-65367392badc497807b3caf5) - calculator-using models we trained and published on HF
123
+ - [**Calc-X and Calcformers paper**](https://arxiv.org/abs/2305.15017)
124
+ - [**Calc-X and Calcformers repo**](https://github.com/prompteus/calc-x)
125
+
126
+ Here are links to the original dataset:
127
+
128
+ - [**original gsm8k dataset**](https://huggingface.co/datasets/gsm8k)
129
+ - [**original gsm8k paper**](https://arxiv.org/abs/2110.14168)
130
+ - [**original gsm8k repo**](https://github.com/openai/grade-school-math)
131
+
132
+
133
+
134
  ## Licence
135
 
136
  MIT, consistently with the original dataset.
 
145
  title = "Calc-X and Calcformers: Empowering Arithmetical Chain-of-Thought through Interaction with Symbolic Systems",
146
  author = "Marek Kadlčík and Michal Štefánik and Ondřej Sotolář and Vlastimil Martinek",
147
  booktitle = "Proceedings of the The 2023 Conference on Empirical Methods in Natural Language Processing: Main track",
148
+ month = dec,
149
  year = "2023",
150
  address = "Singapore, Singapore",
151
  publisher = "Association for Computational Linguistics",