Change hints to instructions

Author: YoanSallami

README.md

@@ -42,7 +42,7 @@ Synalinks is designed for a diverse range of users, from professionals and AI re
 
 Developping a successful LM application in a profesional context, beyond stateless chatbots, is difficult and typically include:
 
-- **Building optimized prompts with examples/hints at each step**: Synalinks uses advanced In-Context Reinforcement Learning techniques to optimize each prompt.
+- **Building optimized prompts with examples/instructions at each step**: Synalinks uses advanced In-Context Reinforcement Learning techniques to optimize each prompt.
 - **Pipelines that change over time**: Easily edit your pipelines, re-run your training, and you're good to go.
 - **Ensuring the correctness of the LMs output**: Synalinks combines constrained structured output with In-Context RL to ensure both format and content correctness.
 - **Async Optimization**: Synalinks automatically optimizes your pipelines by detecting parallel processes.

coverage-badge.svg

@@ -41,7 +41,7 @@ Training a whole Language Model (LM) from scratch is out of the reach of most co
 
 LMs have the capability to leverage their prompt to mimick the examples given, but it means that one have to update the examples each time you change the pipelines as you experiment. Making it cumberstone, but even with that, their is no guaranty that the examples you gave yield to the best results.
 
-To select the best examples and hints to give to the LMs, it needs a complex system like Synalinks that automate the generation and selection.
+To select the best examples and instructions to give to the LMs, it needs a complex system like Synalinks that automate the generation and selection.
 
 ---
 

docs/FAQ.md

@@ -20,7 +20,7 @@ Synalinks is designed for a diverse range of users, from professionals and AI re
 
 Developping a successful LM application in a profesional context, beyond stateless chatbots, is difficult and typically include:
 
-- **Building optimized prompts with examples/hints at each step**: Synalinks uses advanced In-Context Reinforcement Learning techniques to optimize each prompt.
+- **Building optimized prompts with examples/instructions at each step**: Synalinks uses advanced In-Context Reinforcement Learning techniques to optimize each prompt.
 - **Pipelines that change over time**: Easily edit your pipelines, re-run your training, and you're good to go.
 - **Ensuring the correctness of the LMs output**: Synalinks combines constrained structured output with In-Context RL to ensure both format and content correctness.
 - **Async Optimization**: Synalinks automatically optimizes your pipelines by detecting parallel processes.

docs/Introduction.md

@@ -54,7 +54,7 @@
     from synalinks.src.backend.pydantic.base import Weight
     from synalinks.src.backend.pydantic.base import Document
     from synalinks.src.backend.pydantic.base import Prediction
-    from synalinks.src.backend.pydantic.base import Hints
+    from synalinks.src.backend.pydantic.base import Instructions
     from synalinks.src.backend.pydantic.base import is_chat_message
     from synalinks.src.backend.pydantic.base import is_chat_messages
     from synalinks.src.backend.pydantic.base import is_edge
@@ -63,7 +63,7 @@
     from synalinks.src.backend.pydantic.base import is_entities
     from synalinks.src.backend.pydantic.base import is_entity
     from synalinks.src.backend.pydantic.base import is_document
-    from synalinks.src.backend.pydantic.base import is_hints
+    from synalinks.src.backend.pydantic.base import is_instructions
     from synalinks.src.backend.pydantic.base import is_prediction
     from synalinks.src.backend.pydantic.base import is_knowledge_graph
     from synalinks.src.backend.pydantic.base import is_knowledge_graphs

synalinks/src/backend/__init__.py

@@ -33,18 +33,18 @@ class Variable:
     from typing import List
     import synalinks
 
-    class Hints(synalinks.DataModel):
-        hints: List[str] = []
+    class Instructions(synalinks.DataModel):
+        instructions: List[str] = []
 
     initial_data = {
-        "hints": [
+        "instructions": [
             "For any problem involving division, always round the quotient to "
             "the nearest even number, regardless of the remainder."
         ],
     }
     variable_from_dict = synalinks.Variable(
         initializer=initial_data,
-        data_model=Hints,
+        data_model=Instructions,
     )
     ```
 
@@ -54,21 +54,21 @@ class Hints(synalinks.DataModel):
     from typing import List
     import synalinks
 
-    class Hints(synalinks.DataModel):
-        hints: List[str] = []
+    class Instructions(synalinks.DataModel):
+        instructions: List[str] = []
 
     from synalinks.initializers import Empty
 
     variable_from_initializer = synalinks.Variable(
-        initializer=Empty(data_model=Hints)
+        initializer=Empty(data_model=Instructions)
     )
     ```
 
     **Updating the value of a `Variable`:**
 
     ```python
     new_json = {
-        "hints": [
+        "instructions": [
             "When performing division, always check if the division results "
             "in a whole number. If not, express the result as a fraction or "
             "a decimal, depending on the context of the problem."
@@ -83,13 +83,13 @@ class Hints(synalinks.DataModel):
     from typing import List
     import synalinks
 
-    class Hints(synalinks.DataModel):
-        hints: List[str] = []
+    class Instructions(synalinks.DataModel):
+        instructions: List[str] = []
 
     from synalinks.initializers import Empty
 
     non_trainable_variable = synalinks.Variable(
-        initializer=Empty(data_model=Hints), trainable=False
+        initializer=Empty(data_model=Instructions), trainable=False
     )
     ```
 

synalinks/src/backend/common/variables.py

@@ -12,51 +12,51 @@
 
 class VariablesTest(testing.TestCase):
     def test_initialize_variable_with_dict(self):
-        class Hints(DataModel):
-            hints: List[str] = []
+        class Instructions(DataModel):
+            instructions: List[str] = []
 
         initial_data = {
-            "hints": [
+            "instructions": [
                 "For any problem involving division, always round the quotient to "
                 "the nearest even number, regardless of the remainder."
             ],
         }
         variable_from_dict = Variable(
             initializer=initial_data,
-            data_model=Hints,
+            data_model=Instructions,
         )
         self.assertEqual(variable_from_dict.get_json(), initial_data)
         self.assertEqual(
             variable_from_dict.get_schema(),
-            standardize_schema(Hints.get_schema()),
+            standardize_schema(Instructions.get_schema()),
         )
 
     def test_initialize_variable_with_callable_initializer(self):
-        class Hints(DataModel):
-            hints: List[str] = []
+        class Instructions(DataModel):
+            instructions: List[str] = []
 
         from synalinks.src.initializers import Empty
 
-        variable_from_initializer = Variable(initializer=Empty(data_model=Hints))
-        self.assertEqual(variable_from_initializer.get_json(), Hints().get_json())
+        variable_from_initializer = Variable(initializer=Empty(data_model=Instructions))
+        self.assertEqual(variable_from_initializer.get_json(), Instructions().get_json())
         self.assertEqual(
             variable_from_initializer.get_schema(),
-            standardize_schema(Hints.get_schema()),
+            standardize_schema(Instructions.get_schema()),
         )
 
     def test_assign_variable_from_dict(self):
-        class Hints(DataModel):
-            hints: List[str] = []
+        class Instructions(DataModel):
+            instructions: List[str] = []
 
         initial_data = {
-            "hints": [
+            "instructions": [
                 "For any problem involving division, always round the quotient to "
                 "the nearest even number, regardless of the remainder."
             ],
         }
-        variable_from_dict = Variable(initializer=initial_data, data_model=Hints)
+        variable_from_dict = Variable(initializer=initial_data, data_model=Instructions)
         new_value = {
-            "hints": [
+            "instructions": [
                 "When performing division, always check if the division results in "
                 "a whole number. If not, express the result as a fraction or a "
                 "decimal, depending on the context of the problem."
@@ -66,22 +66,22 @@ class Hints(DataModel):
         self.assertEqual(variable_from_dict.get_json(), new_value)
         self.assertEqual(
             variable_from_dict.get_schema(),
-            standardize_schema(Hints.get_schema()),
+            standardize_schema(Instructions.get_schema()),
         )
 
     def test_assign_variable_from_dataype(self):
-        class Hints(DataModel):
-            hints: List[str] = []
+        class Instructions(DataModel):
+            instructions: List[str] = []
 
         initial_data = {
-            "hints": [
+            "instructions": [
                 "For any problem involving division, always round the quotient to "
                 "the nearest even number, regardless of the remainder."
             ],
         }
-        variable_from_dict = Variable(initializer=initial_data, data_model=Hints)
+        variable_from_dict = Variable(initializer=initial_data, data_model=Instructions)
         new_value = {
-            "hints": [
+            "instructions": [
                 "When performing division, always check if the division results in "
                 "a whole number. If not, express the result as a fraction or a "
                 "decimal, depending on the context of the problem."
@@ -91,5 +91,5 @@ class Hints(DataModel):
         self.assertEqual(variable_from_dict.get_json(), new_value)
         self.assertEqual(
             variable_from_dict.get_schema(),
-            standardize_schema(Hints.get_schema()),
+            standardize_schema(Instructions.get_schema()),
         )

synalinks/src/backend/common/variables_test.py

@@ -8,7 +8,7 @@
 from synalinks.src.backend.pydantic.base import Entity
 from synalinks.src.backend.pydantic.base import Document
 from synalinks.src.backend.pydantic.base import Prediction
-from synalinks.src.backend.pydantic.base import Hints
+from synalinks.src.backend.pydantic.base import Instructions
 from synalinks.src.backend.pydantic.base import KnowledgeGraph
 from synalinks.src.backend.pydantic.base import KnowledgeGraphs
 from synalinks.src.backend.pydantic.base import Label
@@ -24,7 +24,7 @@
 from synalinks.src.backend.pydantic.base import is_entities
 from synalinks.src.backend.pydantic.base import is_entity
 from synalinks.src.backend.pydantic.base import is_document
-from synalinks.src.backend.pydantic.base import is_hints
+from synalinks.src.backend.pydantic.base import is_instructions
 from synalinks.src.backend.pydantic.base import is_prediction
 from synalinks.src.backend.pydantic.base import is_knowledge_graph
 from synalinks.src.backend.pydantic.base import is_knowledge_graphs

synalinks/src/backend/pydantic/__init__.py

@@ -380,26 +380,26 @@ def is_prediction(x):
 
 @synalinks_export(
     [
-        "synalinks.backend.Hints",
-        "synalinks.Hints",
+        "synalinks.backend.Instructions",
+        "synalinks.Instructions",
     ]
 )
-class Hints(Entity):
-    """The generator's hints"""
+class Instructions(Entity):
+    """The generator's instructions"""
 
-    label: str = "Hints"
-    hints: List[str]
+    label: str = "Instructions"
+    instructions: List[str]
     reward: Optional[float] = None # None if not yet backpropagated
 
 
 @synalinks_export(
     [
-        "synalinks.backend.is_hints",
-        "synalinks.is_hints",
+        "synalinks.backend.is_instructions",
+        "synalinks.is_instructions",
     ]
 )
-def is_hints(x):
-    """Checks if the given data model is a hints
+def is_instructions(x):
+    """Checks if the given data model is a instructions
 
     Args:
         x (DataModel | JsonDataModel | SymbolicDataModel | Variable):
@@ -408,7 +408,7 @@ def is_hints(x):
     Returns:
         (bool): True if the condition is met
     """
-    if contains_schema(x.get_schema(), Hints.get_schema()):
+    if contains_schema(x.get_schema(), Instructions.get_schema()):
         return True
     return False
 

synalinks/src/backend/pydantic/base.py

@@ -9,21 +9,21 @@
 
 class EmptyInitializerTest(testing.TestCase):
     def test_empty_initializer(self):
-        class Hints(DataModel):
-            hints: List[str] = []
+        class Instructions(DataModel):
+            instructions: List[str] = []
 
-        initializer = Empty(data_model=Hints)
+        initializer = Empty(data_model=Instructions)
         empty_data_model = initializer()
-        self.assertEqual(initializer.get_schema(), Hints.get_schema())
-        self.assertEqual(empty_data_model, Hints().get_json())
+        self.assertEqual(initializer.get_schema(), Instructions.get_schema())
+        self.assertEqual(empty_data_model, Instructions().get_json())
 
     def test_empty_initializer_from_config(self):
-        class Hints(DataModel):
-            hints: List[str] = []
+        class Instructions(DataModel):
+            instructions: List[str] = []
 
-        initializer = Empty(data_model=Hints)
+        initializer = Empty(data_model=Instructions)
         config = initializer.get_config()
         initializer = Empty.from_config(config)
         empty_data_model = initializer()
-        self.assertEqual(initializer.get_schema(), Hints.get_schema())
-        self.assertEqual(empty_data_model, Hints().get_json())
+        self.assertEqual(initializer.get_schema(), Instructions.get_schema())
+        self.assertEqual(empty_data_model, Instructions().get_json())