fix taskmaster loader

Taskmaster.py

@@ -20,6 +20,7 @@ import json
 import os
 
 import datasets
+from phonenumbers import example_number
 
 
 # TODO: Add BibTeX citation
@@ -56,9 +57,9 @@ _LICENSE = "CC BY 4.0"
 # The HuggingFace dataset library don't host the datasets but only point to the original files
 # This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
 _URLs = {
-    "train": "https://huggingface.co/datasets/GEM/Taskmaster/train.csv",
-    "dev": "https://huggingface.co/datasets/GEM/Taskmaster/dev.csv",
-    "test": "https://huggingface.co/datasets/GEM/Taskmaster/test.csv",
+    "train": "train.csv",
+    "validation": "dev.csv",
+    "test": "test.csv",
 }
 
 # New addition
@@ -100,10 +101,10 @@ class Taskmaster(datasets.GeneratorBasedBuilder):
         features = datasets.Features(
             {
                 "gem_id": datasets.Value("string"),
-                "0": datasets.Value("string"),
-                "1": datasets.Value("string"),
-                "2": datasets.Value("string"),
-                "id": datasets.Value("string"),
+                "context": datasets.Value("string"),
+                "target": datasets.Value("string"),
+                "references": [datasets.Value("string")],
+                "conversation_id": datasets.Value("string"),
                 # "paragraphs": datasets.features.Sequence(
                 #     datasets.Value("string")),
                 # "summary": datasets.features.Sequence(
@@ -139,43 +140,31 @@ class Taskmaster(datasets.GeneratorBasedBuilder):
         # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLs
         # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
         # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
-        # _URLs = { "train": "train.csv", "test": "test.csv", "validation": "dev.csv", }
-        my_urls = {
-            "train": "train.csv",
-            "test": "test.csv",
-            "validation": "dev.csv",
-        }  # _URLs[self.config.name]
-        data_dir = dl_manager.download_and_extract(my_urls)
+        data_dir = dl_manager.download_and_extract(_URLs)
 
         return [
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
                 # These kwargs will be passed to _generate_examples
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir["train"], "train-%s.csv"
-                    ),  # % (self.config.name)),
+                    "filepath": data_dir["train"],
                     "split": "train",
                 },
             ),
             datasets.SplitGenerator(
-                name=datasets.Split.TEST,
+                name=datasets.Split.VALIDATION,
                 # These kwargs will be passed to _generate_examples
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir["test"], "test-%s.csv"
-                    ),  # % (self.config.name)),
-                    "split": "test",
+                    "filepath": data_dir["validation"],
+                    "split": "validation",
                 },
             ),
             datasets.SplitGenerator(
-                name=datasets.Split.VALIDATION,
+                name=datasets.Split.TEST,
                 # These kwargs will be passed to _generate_examples
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir["validation"], "valid-%s.csv"
-                    ),  # % (self.config.name)),
-                    "split": "dev",
+                    "filepath": data_dir["test"],
+                    "split": "test",
                 },
             ),
         ]
@@ -190,7 +179,25 @@ class Taskmaster(datasets.GeneratorBasedBuilder):
         # The `key` is here for legacy reason (tfds) and is not important in itself.
 
         with open(filepath, encoding="utf-8") as f:
-            for row in f:
-                data = csv.loads(row)
-                data["gem_id"] = "GEM-TASKMASTER-%s-%d" % (split, data["id"] + 1)
-                yield data["id"], data
+            reader = csv.DictReader(f)
+            expl_count = 0
+            other_count = 0
+            for row in reader:
+                if not row["response"].startswith("[A]"):
+                    other_count += 1
+                    continue
+                target = row["response"].replace("[A]", "")
+                gem_id = f"Taskmaster-{split}-{expl_count}"
+                expl_count += 1
+
+                ex = {
+                    "gem_id": gem_id,
+                    "context": row["context"],
+                    "target": target,
+                    "references": [target],
+                    "conversation_id": row["conversation_id"]
+                }
+
+                print(f"{other_count}/{other_count+expl_count} total now {expl_count}")
+
+                yield expl_count, ex

dataset_infos.json

@@ -0,0 +1,87 @@
+{
+    "default": {
+        "description": "The Taskmaster-3 (aka TicketTalk) dataset consists of 23,789 movie ticketing dialogs\n(located in Taskmaster/TM-3-2020/data/). By \"movie ticketing\" we mean conversations\nwhere the customer's goal is to purchase tickets after deciding on theater, time,\nmovie name, number of tickets, and date, or opt out of the transaction.\nThe columns are gem_id, 0, 1 for serial numbering, 2 for the text dialog and id\nfor the default id by the authors.\n",
+        "citation": "@article{byrne2020tickettalk,\n  title={TicketTalk: Toward human-level performance with end-to-end, transaction-based dialog systems},\n  author={Byrne, Bill and Krishnamoorthi, Karthik and Ganesh, Saravanan and Kale, Mihir Sanjay},\n  journal={arXiv preprint arXiv:2012.12458},\n  year={2020}\n}\n",
+        "homepage": "https://github.com/google-research-datasets/Taskmaster/tree/master/TM-3-2020",
+        "license": "CC BY 4.0",
+        "features": {
+            "gem_id": {
+                "dtype": "string",
+                "id": null,
+                "_type": "Value"
+            },
+            "context": {
+                "dtype": "string",
+                "id": null,
+                "_type": "Value"
+            },
+            "target": {
+                "dtype": "string",
+                "id": null,
+                "_type": "Value"
+            },
+            "references": [
+                {
+                    "dtype": "string",
+                    "id": null,
+                    "_type": "Value"
+                }
+            ],
+            "conversation_id": {
+                "dtype": "string",
+                "id": null,
+                "_type": "Value"
+            }
+        },
+        "post_processed": null,
+        "supervised_keys": null,
+        "task_templates": null,
+        "builder_name": "taskmaster",
+        "config_name": "default",
+        "version": {
+            "version_str": "3.0.0",
+            "description": null,
+            "major": 3,
+            "minor": 0,
+            "patch": 0
+        },
+        "splits": {
+            "train": {
+                "name": "train",
+                "num_bytes": 252278519,
+                "num_examples": 187182,
+                "dataset_name": "taskmaster"
+            },
+            "validation": {
+                "name": "validation",
+                "num_bytes": 31416020,
+                "num_examples": 23406,
+                "dataset_name": "taskmaster"
+            },
+            "test": {
+                "name": "test",
+                "num_bytes": 31207515,
+                "num_examples": 23316,
+                "dataset_name": "taskmaster"
+            }
+        },
+        "download_checksums": {
+            "train.csv": {
+                "num_bytes": 365307526,
+                "checksum": "c182fcfe9aa85931fd438d52e5e8dc70b67a8ec1d2bdf5404ebc3c7452528d57"
+            },
+            "dev.csv": {
+                "num_bytes": 45407763,
+                "checksum": "39958c2fe1d383ddf4d032c8d02850f973ca7f2ef05a6b7067df22758227b381"
+            },
+            "test.csv": {
+                "num_bytes": 45430002,
+                "checksum": "95089fd04b2d91c383b9c65f02dcc942ed9f2d209e6e1d8deef4677a8f8ef18d"
+            }
+        },
+        "download_size": 456145291,
+        "post_processing_size": null,
+        "dataset_size": 314902054,
+        "size_in_bytes": 771047345
+    }
+}

dev.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:38894de0ca972fba8e0872fd4d3381c46fa3a690066c2542154e496c51b90d2b
-size 45407772
+oid sha256:39958c2fe1d383ddf4d032c8d02850f973ca7f2ef05a6b7067df22758227b381
+size 45407763

test.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bf3b80f0a974a57255f87cd626ccfd8a27154f48a01c295e4d7845311b322bcb
-size 45430011
+oid sha256:95089fd04b2d91c383b9c65f02dcc942ed9f2d209e6e1d8deef4677a8f8ef18d
+size 45430002

train.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0d353eb959782ab32db4b7633209df5609179e9326c01027cd01a8f7854e200b
-size 365307535
+oid sha256:c182fcfe9aa85931fd438d52e5e8dc70b67a8ec1d2bdf5404ebc3c7452528d57
+size 365307526