working on generation

dev_packages/chatsky_llm_autoconfig/chatsky_llm_autoconfig/algorithms/base.py

@@ -40,24 +40,6 @@ def invoke(self, graph: BaseGraph, start_node: int = 1, end_node: int = 0, topic
         raise NotImplementedError
 
 
-class DialogAugmentation(BaseAlgorithm):
-    """
-    Base class for augmenting Dialogues.
-
-    This class takes a Dialogue as input and returns an augmented Dialogue as output.
-    It is designed for data augmentation or other manipulations of Dialogues.
-
-    :param dialogue: The Dialogue object to be augmented.
-    :param topic: The topic to guide the augmentation process (optional).
-    """
-
-    def __init__(self):
-        super().__init__()
-
-    def invoke(self, dialogue: Dialogue, topic: str = "") -> Dialogue:
-        raise NotImplementedError
-
-
 class GraphAugmentation(BaseAlgorithm):
     """Graph generator that works only with topics."""
 

dev_packages/chatsky_llm_autoconfig/chatsky_llm_autoconfig/algorithms/dialogue_augmentation.py

@@ -2,14 +2,29 @@
 from chatsky_llm_autoconfig.schemas import DialogueMessage
 from langchain.prompts import PromptTemplate
 from langchain_core.output_parsers import JsonOutputParser
-from typing import List
-
+from typing import List, Any
 from langchain_openai import ChatOpenAI
-from pydantic import BaseModel
+from pydantic import BaseModel, SecretStr
 import os
+from chatsky_llm_autoconfig.algorithms.base import BaseAlgorithm
+from chatsky_llm_autoconfig.autometrics.registry import AlgorithmRegistry
+from langchain_core.runnables import RunnableSerializable
+
+
+class DialogueSequence(BaseModel):
+    result: List[DialogueMessage]
+
+
+@AlgorithmRegistry.register(input_type=Dialogue, output_type=Dialogue)
+class DialogAugmentation(BaseAlgorithm):
+    """Base class for augmenting Dialogues."""
+
+    augmentation_prompt: PromptTemplate = ""
 
-augmentation_prompt = PromptTemplate.from_template(
-    """
+    def __init__(self):
+        super().__init__()
+        self.augmentation_prompt = PromptTemplate.from_template(
+            """
 You are tasked with augmenting a dialogue by adding variations to existing utterances while maintaining the original dialogue flow and intent.
 
 THEME: {topic}
@@ -46,21 +61,7 @@
     {{"text": "What kind of package is it?", "participant": "assistant"}}
 ]
 """
-)
-
-
-class DialogueSequence(BaseModel):
-    result: List[DialogueMessage]
-
-
-class DialogAugmentation(BaseModel):
-    """Base class for augmenting Dialogues."""
-
-    def __init__(self, **data):
-        super().__init__(**data)
-        self.parser = JsonOutputParser(pydantic_object=DialogueSequence)
-        self.model = ChatOpenAI(model="gpt-4o-mini", api_key=os.getenv("OPENAI_API_KEY"), base_url=os.getenv("OPENAI_BASE_URL"), temperature=0.7)
-        self.chain = augmentation_prompt | self.model | self.parser
+        )
 
     def invoke(self, *, dialogue: Dialogue, topic: str = "") -> Dialogue:
         """
@@ -73,14 +74,15 @@ def invoke(self, *, dialogue: Dialogue, topic: str = "") -> Dialogue:
         Returns:
             Dialogue: Augmented dialogue object
         """
-        # Convert dialogue to string format for prompt
-        # Предполагая, что у Dialogue есть str представление
-        dialogue_str = str(dialogue)
+        parser: JsonOutputParser = JsonOutputParser(pydantic_object=DialogueSequence)
+        model: ChatOpenAI = ChatOpenAI(
+            model="gpt-4o-mini", api_key=SecretStr(os.getenv("OPENAI_API_KEY") or ""), base_url=os.getenv("OPENAI_BASE_URL"), temperature=0.7
+        )
+        chain: RunnableSerializable[Any, Any] = self.augmentation_prompt | model | parser
 
-        # Get augmented messages
-        result = self.chain.invoke({"topic": topic, "dialogue": dialogue_str})
+        dialogue_str: str = str(dialogue)
+        result: List[DialogueMessage] = chain.invoke({"topic": topic, "dialogue": dialogue_str})
 
-        # Create new Dialogue object with augmented messages
         return Dialogue(messages=result, topic=topic)
 
     async def ainvoke(self, *, dialogue: Dialogue, topic: str = "") -> Dialogue:
@@ -94,8 +96,13 @@ async def ainvoke(self, *, dialogue: Dialogue, topic: str = "") -> Dialogue:
         Returns:
             Dialogue: Augmented dialogue object
         """
-        dialogue_str = str(dialogue)
-
-        result = await self.chain.ainvoke({"topic": topic, "dialogue": dialogue_str})
+        parser: JsonOutputParser = JsonOutputParser(pydantic_object=DialogueSequence)
+        model: ChatOpenAI = ChatOpenAI(
+            model="gpt-4o-mini", api_key=SecretStr(os.getenv("OPENAI_API_KEY") or ""), base_url=os.getenv("OPENAI_BASE_URL"), temperature=0.7
+        )
+        chain: RunnableSerializable[Any, Any] = self.augmentation_prompt | model | parser
+
+        dialogue_str: str = str(dialogue)
+        result: List[DialogueMessage] = await chain.ainvoke({"topic": topic, "dialogue": dialogue_str})
 
         return Dialogue(messages=result, topic=topic)

dev_packages/chatsky_llm_autoconfig/chatsky_llm_autoconfig/autometrics/run_autometrics.py

@@ -66,7 +66,7 @@ def run_all_algorithms():
         elif algorithms[class_]["input_type"] is str and algorithms[class_]["output_type"] is BaseGraph:
             metrics = {"is_theme_valid": [], "are_triplets_valid": []}
             for case in topic_to_graph:
-                test_topic = case['topic']
+                test_topic = case["topic"]
                 result = class_instance.invoke(test_topic)
 
                 metrics["are_triplets_valid"].append(are_triplets_valid(result, model, topic=test_topic)["value"])

dev_packages/chatsky_llm_autoconfig/chatsky_llm_autoconfig/metrics/llm_metrics.py

@@ -5,8 +5,11 @@
 This module contains functions that checks Graphs and Dialogues for various metrics using LLM calls.
 """
 
+from langchain.chat_models.base import BaseChatModel
+from langchain.output_parsers import PydanticOutputParser
+from typing import TypedDict, List
 from chatsky_llm_autoconfig.graph import BaseGraph
-from typing import List, Tuple
+from typing import List
 from langchain_core.language_models.chat_models import BaseChatModel
 from langchain.prompts import PromptTemplate
 from pydantic import BaseModel, Field
@@ -18,7 +21,12 @@
 logging.basicConfig(level=logging.INFO)
 
 
-def are_triplets_valid(G: BaseGraph, model: BaseChatModel, topic: str) -> dict[str]:
+class ValidationResult(TypedDict):
+    value: bool
+    description: str
+
+
+def are_triplets_valid(G: BaseGraph, model: BaseChatModel, topic: str) -> ValidationResult:
     """
     Validates the dialog graph structure and logical transitions between nodes.
 
@@ -28,7 +36,7 @@ def are_triplets_valid(G: BaseGraph, model: BaseChatModel, topic: str) -> dict[s
         topic (str): The topic of the dialog
 
     Returns:
-        dict: {'value': bool, 'description': str}
+        ValidationResult: Dictionary with structure {'value': bool, 'description': str}
     """
     # Define prompt template and parser inside the function since they're only used here
     triplet_validate_prompt_template = """
@@ -74,7 +82,7 @@ class TransitionValidationResult(BaseModel):
     # Create a mapping from node IDs to node data for quick access
     node_map = {node["id"]: node for node in graph["nodes"]}
     overall_valid = True
-    descriptions = []
+    descriptions: List[str] = []
 
     for edge in graph["edges"]:
         source_id = edge["source"]
@@ -83,13 +91,15 @@ class TransitionValidationResult(BaseModel):
 
         # Check if source and target nodes exist
         if source_id not in node_map:
-            description = f"Invalid edge: source node {source_id} does not exist."
+            description = f"Invalid edge: source node {
+                source_id} does not exist."
             logging.info(description)
             overall_valid = False
             descriptions.append(description)
             continue
         if target_id not in node_map:
-            description = f"Invalid edge: target node {target_id} does not exist."
+            description = f"Invalid edge: target node {
+                target_id} does not exist."
             logging.info(description)
             overall_valid = False
             descriptions.append(description)
@@ -115,11 +125,12 @@ class TransitionValidationResult(BaseModel):
 
         if not response.isValid:
             overall_valid = False
-            description = f"Invalid transition from {source_utterances} to {target_utterances} via edge '{edge_utterances}': {response.description}"
+            description = f"Invalid transition from {source_utterances} to {
+                target_utterances} via edge '{edge_utterances}': {response.description}"
             logging.info(description)
             descriptions.append(description)
 
-    result = {"value": overall_valid, "description": " ".join(descriptions) if descriptions else "All transitions are valid."}
+    result: ValidationResult = {"value": overall_valid, "description": " ".join(descriptions) if descriptions else "All transitions are valid."}
     return result
 
 

experiments/2024.10.06_synthetic_data_based_on_existing_graphs/gen_data.ipynb

@@ -367,7 +367,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "dff-llm-integration-VcuUrJCU-py3.12",
+   "display_name": ".venv",
    "language": "python",
    "name": "python3"
   },
@@ -381,7 +381,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.12.3"
+   "version": "3.12.7"
   }
  },
  "nbformat": 4,