Attend the Women in ML Symposium on December 7 Register now

americas_nli

References:

aym

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/aym')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

bzd

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/bzd')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

cni

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/cni')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 658
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

gn

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/gn')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

hch

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/hch')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

nah

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/nah')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 738
'validation' 376
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

oto

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/oto')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 748
'validation' 222
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

quy

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/quy')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

shp

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/shp')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

tar

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/tar')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 750
'validation' 743
  • Features:
{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}

all_languages

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:americas_nli/all_languages')
  • Description:
AmericasNLI is an extension of XNLI (Conneau et al., 2018) – a natural language inference (NLI) dataset covering 15 high-resource languages – to 10 low-resource indigenous languages spoken in the Americas: Ashaninka, Aymara, Bribri, Guarani, Nahuatl, Otomi, Quechua, Raramuri, Shipibo-Konibo, and Wixarika. As with MNLI, the goal is to predict textual entailment (does sentence A imply/contradict/neither sentence B) and is a classification task (given two sentences, predict one of three labels).
  • License: No known license
  • Version: 1.0.0
  • Splits:
Split Examples
'test' 7486
'validation' 6457
  • Features:
{
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}