modified templates and IOFields for structured data task cards

IBM · Feb 22, 2024 · 87f1d9a · 87f1d9a
1 parent 4b8fd85
commit 87f1d9a
Show file tree

Hide file tree

Showing 13 changed files with 19 additions and 15 deletions.
diff --git a/prepare/cards/dart.py b/prepare/cards/dart.py
@@ -19,7 +19,7 @@
             field_to_field={"annotations/text/0": "output"},
             use_query=True,
         ),
-        AddFields(fields={"type_of_input": "Triples"}),
+        AddFields(fields={"type_of_input": "Triples", "type_of_output": "Text"}),
     ],
     task="tasks.generation",
     templates="templates.generation.all",

diff --git a/prepare/cards/tablerow_classify.py b/prepare/cards/tablerow_classify.py
@@ -23,7 +23,7 @@
         MapInstanceValues(mappers={"label": {"0": "Normal", "1": "Heart Disease"}}),
         AddFields(
             fields={
-                "text_type": "Person",
+                "text_type": "Person medical record",
                 "type_of_class": "Heart Disease Possibility",
             }
         ),

diff --git a/prepare/cards/wiki_bio.py b/prepare/cards/wiki_bio.py
@@ -21,7 +21,9 @@
         ),
         SerializeKeyValPairs(field_to_field=[["kvpairs", "input"]]),
         RenameFields(field_to_field={"target_text": "output"}),
-        AddFields(fields={"type_of_input": "Key-Value pairs"}),
+        AddFields(
+            fields={"type_of_input": "Key-Value pairs", "type_of_output": "Text"}
+        ),
     ],
     task="tasks.generation",
     templates="templates.generation.all",

diff --git a/prepare/tasks/generation.py b/prepare/tasks/generation.py
@@ -3,7 +3,7 @@
 
 add_to_catalog(
     FormTask(
-        inputs=["input", "type_of_input"],
+        inputs=["input", "type_of_input", "type_of_output"],
         outputs=["output"],
         metrics=["metrics.normalized_sacrebleu"],
         augmentable_inputs=["input"],

diff --git a/prepare/templates/classification/classification.py b/prepare/templates/classification/classification.py
@@ -5,7 +5,7 @@
 
 add_to_catalog(
     InputOutputTemplate(
-        input_format="Classify the {type_of_class} of following {text_type} to one of these options: {classes}. Text: {text}",
+        input_format="Classify the {type_of_class} of following {text_type} to one of these options: {classes}. {text_type}: {text}",
         output_format="{label}",
         postprocessors=[
             "processors.take_first_non_empty_line",

diff --git a/prepare/templates/generation/generation.py b/prepare/templates/generation/generation.py
@@ -5,7 +5,7 @@
 
 add_to_catalog(
     InputOutputTemplate(
-        input_format="Given the following {type_of_input}, generate the corresponding text. {type_of_input}: {input}",
+        input_format="Given the following {type_of_input}, generate the corresponding {type_of_output}. {type_of_input}: {input}",
         output_format="{output}",
         postprocessors=[
             "processors.take_first_non_empty_line",

diff --git a/src/unitxt/catalog/cards/dart.json b/src/unitxt/catalog/cards/dart.json
@@ -31,7 +31,8 @@
         {
             "type": "add_fields",
             "fields": {
-                "type_of_input": "Triples"
+                "type_of_input": "Triples",
+                "type_of_output": "Text"
             }
         }
     ],

diff --git a/src/unitxt/catalog/cards/tablerow_classify.json b/src/unitxt/catalog/cards/tablerow_classify.json
@@ -31,7 +31,7 @@
         {
             "type": "add_fields",
             "fields": {
-                "text_type": "Person",
+                "text_type": "Person medical record",
                 "type_of_class": "Heart Disease Possibility"
             }
         },

diff --git a/src/unitxt/catalog/cards/wiki_bio.json b/src/unitxt/catalog/cards/wiki_bio.json
@@ -40,7 +40,8 @@
         {
             "type": "add_fields",
             "fields": {
-                "type_of_input": "Key-Value pairs"
+                "type_of_input": "Key-Value pairs",
+                "type_of_output": "Text"
             }
         }
     ],

diff --git a/src/unitxt/catalog/tasks/generation.json b/src/unitxt/catalog/tasks/generation.json
@@ -2,7 +2,8 @@
     "type": "form_task",
     "inputs": [
         "input",
-        "type_of_input"
+        "type_of_input",
+        "type_of_output"
     ],
     "outputs": [
         "output"

diff --git a/src/unitxt/catalog/templates/classification/multi_class/default.json b/src/unitxt/catalog/templates/classification/multi_class/default.json
@@ -1,6 +1,6 @@
 {
     "type": "input_output_template",
-    "input_format": "Classify the {type_of_class} of following {text_type} to one of these options: {classes}. Text: {text}",
+    "input_format": "Classify the {type_of_class} of following {text_type} to one of these options: {classes}. {text_type}: {text}",
     "output_format": "{label}",
     "postprocessors": [
         "processors.take_first_non_empty_line",

diff --git a/src/unitxt/catalog/templates/generation/default.json b/src/unitxt/catalog/templates/generation/default.json
@@ -1,6 +1,6 @@
 {
     "type": "input_output_template",
-    "input_format": "Given the following {type_of_input}, generate the corresponding text. {type_of_input}: {input}",
+    "input_format": "Given the following {type_of_input}, generate the corresponding {type_of_output}. {type_of_input}: {input}",
     "output_format": "{output}",
     "postprocessors": [
         "processors.take_first_non_empty_line",

diff --git a/tests/library/test_struct_data_operators.py b/tests/library/test_struct_data_operators.py
@@ -1,6 +1,6 @@
 import unittest
 
-from unitxt.struct_data_operators import (
+from src.unitxt.struct_data_operators import (
     ListToKeyValPairs,
     SerializeKeyValPairs,
     SerializeTableAsIndexedRowMajor,
@@ -11,13 +11,12 @@
     TruncateTableCells,
     TruncateTableRows,
 )
-
 from src.unitxt.test_utils.operators import (
     check_operator,
 )
 
 
-class TestTableOperators(unittest.TestCase):
+class TestStructDataOperators(unittest.TestCase):
     """Tests for tabular data processing operators."""
 
     def test_serializetable_markdown(self):