pietrolesci commited on
Commit
94c64c8
1 Parent(s): 2f914a5

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -6
README.md CHANGED
@@ -6,9 +6,10 @@ Original dataset is available on the HuggingFace Hub [here](https://huggingface.
6
  This is the same as the `snli_format` split of the SciTail dataset available on the HuggingFace Hub (i.e., same data, same splits, etc).
7
  The only differences are the following:
8
 
9
- - selecting only the columns `["sentence1", "sentence2", "gold_label"]`
10
- - renaming columns with the following mapping `{"sentence1": "premise", "sentence2": "hypothesis", "gold_label": "label"}`
11
- - encoding labels with the following mapping `{"entailment": 0, "neutral": 1, "contradiction": 2}`
 
12
 
13
  Note that there are 10 overlapping instances (as found by merging on columns "label", "premise", and "hypothesis") between
14
  `train` and `test` splits.
@@ -29,16 +30,17 @@ for name, df_ in dd.items():
29
  df = df[["sentence1", "sentence2", "gold_label"]]
30
 
31
  # rename columns
32
- df = df.rename(columns={"sentence1": "premise", "sentence2": "hypothesis", "gold_label": "label"})
33
 
34
  # encode labels
35
- df["label"] = df["label"].map({"entailment": 0, "neutral": 1, "contradiction": 2})
 
36
 
37
  # cast to dataset
38
  features = Features({
39
  "premise": Value(dtype="string", id=None),
40
  "hypothesis": Value(dtype="string", id=None),
41
- "label": ClassLabel(num_classes=3, names=["entailment", "neutral", "contradiction"]),
42
  })
43
  ds[name] = Dataset.from_pandas(df, features=features)
44
 
6
  This is the same as the `snli_format` split of the SciTail dataset available on the HuggingFace Hub (i.e., same data, same splits, etc).
7
  The only differences are the following:
8
 
9
+ - selecting only the columns `["sentence1", "sentence2", "gold_label", "label"]`
10
+ - renaming columns with the following mapping `{"sentence1": "premise", "sentence2": "hypothesis"}`
11
+ - creating a new column "label" from "gold_label" with the following mapping `{"entailment": "entailment", "neutral": "not_entailment"}`
12
+ - encoding labels with the following mapping `{"not_entailment": 0, "entailment": 1}`
13
 
14
  Note that there are 10 overlapping instances (as found by merging on columns "label", "premise", and "hypothesis") between
15
  `train` and `test` splits.
30
  df = df[["sentence1", "sentence2", "gold_label"]]
31
 
32
  # rename columns
33
+ df = df.rename(columns={"sentence1": "premise", "sentence2": "hypothesis"})
34
 
35
  # encode labels
36
+ df["label"] = df["gold_label"].map({"entailment": "entailment", "neutral": "not_entailment"})
37
+ df["label"] = df["label"].map({"not_entailment": 0, "entailment": 1})
38
 
39
  # cast to dataset
40
  features = Features({
41
  "premise": Value(dtype="string", id=None),
42
  "hypothesis": Value(dtype="string", id=None),
43
+ "label": ClassLabel(num_classes=2, names=["not_entailment", "entailment"]),
44
  })
45
  ds[name] = Dataset.from_pandas(df, features=features)
46