Migrate Wit workflows to new data design, delete deprecated Word Tagg…

…ing tasks, models and data handlers (#910) Summary: Pull Request resolved: #910 as title Differential Revision: D16827032 fbshipit-source-id: 6e0a90f7bdad2033c8e8fce35201e41fe65d3fb3
facebookresearch · Sep 5, 2019 · 134bebb · 134bebb
1 parent 8aded25
commit 134bebb
Show file tree

Hide file tree

Showing 18 changed files with 141 additions and 892 deletions.
diff --git a/pytext/builtin_task.py b/pytext/builtin_task.py
@@ -27,10 +27,8 @@
     SemanticParsingTask,
     SemanticParsingTask_Deprecated,
     SeqNNTask,
-    SeqNNTask_Deprecated,
     SquadQATask,
     WordTaggingTask,
-    WordTaggingTask_Deprecated,
 )
 
 
@@ -84,9 +82,7 @@ def register_builtin_tasks():
             SemanticParsingTask,
             SemanticParsingTask_Deprecated,
             SeqNNTask,
-            SeqNNTask_Deprecated,
             SquadQATask,
             WordTaggingTask,
-            WordTaggingTask_Deprecated,
         )
     )
diff --git a/pytext/config/config_adapter.py b/pytext/config/config_adapter.py
@@ -315,7 +315,6 @@ def old_tasks_deprecated(json_config):
     deprecate(json_config, "ShallowTaggingTask")
     deprecate(json_config, "SpanClassificationTask")
     deprecate(json_config, "TreeParserTask")
-    deprecate(json_config, "WordTaggingTask")
 
     return json_config
 
@@ -385,7 +384,6 @@ def new_tasks_rename(json_config):
         "QueryDocumentPairwiseRankingModel",
         "QueryDocumentPairwiseRankingModel_Deprecated",
     )
-    rename(json_config, "WordTaggingModel", "WordTaggingModel_Deprecated")
     # New
     rename(json_config, "NewDocModel", "DocModel")
     rename(json_config, "NewDocRegressionModel", "DocRegressionModel")
@@ -461,19 +459,15 @@ def v12_to_v13(json_config):
 
     [(task_name, task)] = json_config["task"].items()
     if task_name not in (
-        "EnsembleTask_Deprecated",
         "EnsembleTask",
         "DocClassificationTask_Deprecated",
         "DocumentClassificationTask",
         "PairwiseClassificationTask",
-        "SeqNNTask_Deprecated",
         "SeqNNTask",
         "ShallowClassificationTask_Deprecated",
         "KDDocClassificationTask_Deprecated",
-        "ElmoDocClassificationTask_Deprecated",
         "PairwiseAttentionClassificationTask_Deprecated",
         "ElmoFineTunePairwiseClassificationTask_Deprecated",
-        "ElmoKDClassificationTask_Deprecated",
         "XLMDocumentClassification",
         "XLMPairClassification",
         "NewBertClassificationTask",
@@ -542,13 +536,7 @@ def update_model_config(model_config):
 
 @register_adapter(from_version=14)
 def flatten_deprecated_ensemble_config(json_config):
-    [(task_name, task)] = json_config["task"].items()
-    if task_name != "EnsembleTask_Deprecated":
-        return json_config
-
-    # Joint ensemble model is deleted, so we no longer have a union type, hence
-    # we flatten the config.
-    task["model"] = task["model"]["BaggingDocEnsemble_Deprecated"]
+    # Deprecated ensemble is removed from codebase, so this is now just a no-op
     return json_config
 
 

diff --git a/pytext/config/test/json_config/v6.json b/pytext/config/test/json_config/v6.json
@@ -24,109 +24,5 @@
       },
       "version": 6
     }
-  },
-  {
-    "original": {
-      "task": {
-        "WordTaggingTask": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 5
-    },
-    "adapted": {
-      "task": {
-        "WordTaggingTask_Deprecated": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 6
-    }
-  },
-  {
-    "original": {
-      "task": {
-        "WordTaggingTask": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 5
-    },
-    "adapted": {
-      "task": {
-        "WordTaggingTask_Deprecated": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 6
-    }
-  },
-  {
-    "original": {
-      "task": {
-        "WordTaggingTask": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 5
-    },
-    "adapted": {
-      "task": {
-        "WordTaggingTask_Deprecated": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 6
-    }
-  },
-  {
-    "original": {
-      "task": {
-        "SeqNNTask": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 5
-    },
-    "adapted": {
-      "task": {
-        "SeqNNTask_Deprecated": {
-          "data_handler": {
-            "train_path": "tests/data/train_data_tiny.tsv",
-            "eval_path": "tests/data/test_data_tiny.tsv",
-            "test_path": "tests/data/test_data_tiny.tsv"
-          }
-        }
-      },
-      "version": 6
-    }
   }
 ]
diff --git a/pytext/data/__init__.py b/pytext/data/__init__.py
@@ -14,8 +14,6 @@
 from .disjoint_multitask_data import DisjointMultitaskData
 from .disjoint_multitask_data_handler import DisjointMultitaskDataHandler
 from .doc_classification_data_handler import DocClassificationDataHandler, RawData
-from .joint_data_handler import JointModelDataHandler
-from .seq_data_handler import SeqModelDataHandler
 from .tensorizers import Tensorizer
 
 
@@ -33,11 +31,9 @@
     "DocClassificationDataHandler",
     "EvalBatchSampler",
     "generator_iterator",
-    "JointModelDataHandler",
     "PoolingBatcher",
     "RandomizedBatchSampler",
     "RawData",
     "RoundRobinBatchSampler",
-    "SeqModelDataHandler",
     "Tensorizer",
 ]
diff --git a/pytext/data/joint_data_handler.py b/pytext/data/joint_data_handler.py