versionhq 1.2.3.0__py3-none-any.whl → 1.2.3.4__py3-none-any.whl

versionhq/__init__.py

@@ -32,7 +32,7 @@ from versionhq.agent_network.formation import form_agent_network
 from versionhq.task_graph.draft import workflow
 
 
-__version__ = "1.2.3.0"
+__version__ = "1.2.3.4"
 __all__ = [
     "Agent",
 

versionhq/clients/workflow/model.py

@@ -2,92 +2,25 @@ import uuid
 from abc import ABC
 from datetime import datetime
 from typing import Any, Dict, List, Optional
-from typing_extensions import Self
+
 from pydantic import UUID4, InstanceOf, BaseModel, ConfigDict, Field, field_validator, model_validator
 from pydantic_core import PydanticCustomError
 
-from versionhq.clients.product.model import Product
-from versionhq.clients.customer.model import Customer
 from versionhq.agent.model import Agent
 from versionhq.agent_network.model import AgentNetwork
+from versionhq.clients.product.model import Product
+from versionhq.clients.customer.model import Customer
 from versionhq.tool.composio_tool_vars import ComposioAppName
 
 
-class ScoreFormat:
-    def __init__(self, rate: float | int = 0, weight: int = 1):
-        self.rate = rate
-        self.weight = weight
-        self.aggregate = rate * weight
-
-
-class Score:
-    """
-    Evaluate the score on 0 (no performance) to 1 scale.
-    `rate`: Any float from 0.0 to 1.0 given by an agent.
-    `weight`: Importance of each factor to the aggregated score.
-    """
-
-    def __init__(
-        self,
-        brand_tone: ScoreFormat = ScoreFormat(0, 0),
-        audience: ScoreFormat = ScoreFormat(0, 0),
-        track_record: ScoreFormat = ScoreFormat(0, 0),
-        **kwargs: Optional[Dict[str, ScoreFormat]],
-    ):
-        self.brand_tone = brand_tone
-        self.audience = audience
-        self.track_record = track_record
-        self.kwargs = kwargs
-
-
-    def result(self) -> int:
-        aggregate_score = int(self.brand_tone.aggregate) + int(self.audience.aggregate) + int(self.track_record.aggregate)
-        denominator = self.brand_tone.weight + self.audience.weight + self.track_record.weight
-
-        for k, v in self.kwargs.items():
-            aggregate_score += v.aggregate
-            denominator += v.weight
-
-        if denominator == 0:
-            return 0
-
-        return round(aggregate_score / denominator, 2)
-
-
-
 class MessagingComponent(ABC, BaseModel):
     layer_id: int = Field(default=0, description="add id of the layer: 0, 1, 2")
     message: str = Field(default=None, max_length=1024, description="text message content to be sent")
-    score: InstanceOf[Score] = Field(default=None)
+    score: Optional[float | int] = Field(default=None)
     condition: str = Field(default=None, description="condition to execute the next component")
     interval: Optional[str] = Field(default=None, description="ideal interval to set to assess the condition")
 
 
-    def store_scoring_result(self, subject: str, score_raw: int | Score | ScoreFormat = None) -> Self:
-        """
-        Set up the `score` field
-        """
-
-        if isinstance(score_raw, Score):
-            setattr(self, "score", score_raw)
-
-        elif isinstance(score_raw, ScoreFormat):
-            score_instance = Score()
-            setattr(score_instance, subject, score_raw)
-            setattr(self, "score", score_instance)
-
-        elif isinstance(score_raw, int) or isinstance(score_raw, float):
-            score_instance, score_format_instance = Score(), ScoreFormat(rate=score_raw, weight=1)
-            setattr(score_instance, "kwargs", { subject: score_format_instance })
-            setattr(self, "score", score_instance)
-
-        else:
-            pass
-
-        return self
-
-
-
 class MessagingWorkflow(ABC, BaseModel):
     """
     Store 3 layers of messaging workflow sent to `customer` on the `product`

versionhq/storage/ltm_sqlite_storage.py

@@ -97,7 +97,7 @@ class LTMSQLiteStorage:
                     ]
 
         except sqlite3.Error as e:
-            self._logger.log(level="error", message=f"MEMORY ERROR: An error occurred while querying LTM: {str(e)}",color="red")
+            self._logger.log(level="error", message=f"MEMORY ERROR: An error occurred while querying LTM: {str(e)}", color="red")
             return None
 
 

versionhq/task/TEMPLATES/Description.py

@@ -1,4 +1,4 @@
-EVALUATE="""Evaluate the provided task output against the given task description, assigning a score between 0 (worst) and 1 (best) based on the specified criteria. Scores should be numerical (integers or decimals). Provide specific suggestions for improvement. Do not assign identical scores to different criteria unless otherwise you have clear reasons to do so:
+EVALUATE="""Evaluate the provided task output against the given task description, assigning a score between 0 (worst) and 1 (best) based on the specified criteria. Scores should be numerical (integers or decimals). Weight should be numerical (integers or decimals) and represents importance of the criteria to the final result. Provide specific suggestions for improvement. Do not assign identical scores to different criteria unless otherwise you have clear reasons to do so:
 Task output: {task_output}
 Task description: {task_description}
 Evaluation criteria: {eval_criteria}

versionhq/task/evaluation.py

@@ -1,64 +1,22 @@
-from typing import List, Optional, Dict, Any
+from typing import List, Any
 from typing_extensions import Self
 
 from pydantic import BaseModel, model_validator
+import pandas as pd
+from sklearn.preprocessing import MinMaxScaler
 
 from versionhq.memory.model import MemoryMetadata
 
-"""
-Evaluate task output from accuracy, token consumption, and latency perspectives, and mark the score from 0 to 1.
-"""
-
-
-class ScoreFormat:
-    def __init__(self, rate: float | int = 0, weight: int = 1):
-        self.rate = rate
-        self.weight = weight
-        self.aggregate = rate * weight
-
-
-class Score:
-    """
-    Evaluate the score on 0 (no performance) to 1 scale.
-    `rate`: Any float from 0.0 to 1.0 given by an agent.
-    `weight`: Importance of each factor to the aggregated score.
-    """
-
-    def __init__(self, config: Optional[Dict[str, ScoreFormat]] = None):
-        self.config = config
-
-        if self.config:
-            for k, v in self.config.items():
-                if isinstance(v, ScoreFormat):
-                    setattr(self, k, v)
-
-
-    def result(self) -> float:
-        aggregate_score, denominator = 0, 0
-
-        for k, v in self.__dict__.items():
-            aggregate_score += v.aggregate
-            denominator += v.weight
-
-        if denominator == 0:
-            return 0
-
-        return round(aggregate_score / denominator, 3)
-
 
 class EvaluationItem(BaseModel):
     """
     A Pydantic class to store the evaluation result with scoring and suggestion based on the given criteria.
+    This class will be used as a response format for the eval task.
     """
     criteria: str
     suggestion: str
     score: float
-
-    def _format_score(self, weight: int = 1) -> ScoreFormat | None:
-        if self.score and isinstance(self.score, float):
-            return ScoreFormat(rate=self.score, weight=weight)
-
-        else: return None
+    weight: int = 1
 
 
 class Evaluation(BaseModel):
@@ -111,33 +69,43 @@ class Evaluation(BaseModel):
         return shot_prompt
 
 
-    @property
-    def aggregate_score(self) -> float:
+    def _normalize_df(self) -> pd.DataFrame:
         """
-        Calcurate aggregate score from evaluation items.
+        Creates a pandas DataFrame from a list of EvaluationItem objects containing 'weight' and 'score' columns, and normalizes them using MinMaxScaler.
+
+        Args:
+            items: A list of EvaluationItem objects.
+
+        Returns:
+            A pandas DataFrame with normalized 'weight' and 'score' columns, or an empty DataFrame if the input is empty.
         """
         if not self.items:
-            return 0
+            return pd.DataFrame()
 
-        aggregate_score = 0
-        denominator = 0
+        data = { 'weight': [item.weight for item in self.items], 'score': [item.score for item in self.items] }
+        df = pd.DataFrame(data)
 
-        for item in self.items:
-            score_format = item._format_score()
-            aggregate_score += score_format.aggregate if score_format else 0
-            denominator += score_format.weight if score_format else 0
+        scaler = MinMaxScaler(feature_range=(0, 1))
+        df[['weight', 'score']] = scaler.fit_transform(df[['weight', 'score']])
 
-        if denominator == 0:
+        return df
+
+
+    @property
+    def aggregate_score(self) -> int | float:
+        if not self.items:
             return 0
 
-        return round(aggregate_score / denominator, 2)
+        df = self._normalize_df()
+        df['weighted_score'] = df['weight'] * df['score']
+        aggregate_score = round(df['weighted_score'].sum(), 3)
+        return aggregate_score
 
 
     @property
     def suggestion_summary(self) -> str | None:
-        """
-        Returns a summary of the suggestions
-        """
+        """Returns a summary of the suggestions"""
+
         if not self.items:
             return None
 

versionhq/task/model.py

@@ -228,13 +228,21 @@ class TaskOutput(BaseModel):
             self._tokens += task_eval._tokens
 
             if res.pydantic:
-                item = EvaluationItem(score=res.pydantic.score, suggestion=res.pydantic.suggestion, criteria=res.pydantic.criteria)
+                item = EvaluationItem(
+                    score=res.pydantic.score,
+                    weight=res.pydantic.weight,
+                    suggestion=res.pydantic.suggestion,
+                    criteria=res.pydantic.criteria
+                )
                 self.evaluation.items.append(item)
 
             else:
                 try:
                     item = EvaluationItem(
-                        score=float(res.json_dict["score"]), suggestion=res.json_dict["suggestion"], criteria=res.json_dict["criteria"]
+                        score=float(res.json_dict["score"]),
+                        weight=float(res.json_dict["weight"]),
+                        suggestion=res.json_dict["suggestion"],
+                        criteria=res.json_dict["criteria"]
                     )
                     self.evaluation.items.append(item)
                 except Exception as e:
@@ -246,10 +254,7 @@ class TaskOutput(BaseModel):
 
     @property
     def aggregate_score(self) -> float | int:
-        if self.evaluation is None:
-            return 0
-        else:
-            self.evaluation.aggregate_score
+        return self.evaluation.aggregate_score if self.evaluation is not None else 0
 
 
     @property

{versionhq-1.2.3.0.dist-info → versionhq-1.2.3.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: versionhq
-Version: 1.2.3.0
+Version: 1.2.3.4
 Summary: An agentic orchestration framework for building agent networks that handle task automation.
 Author-email: Kuriko Iwai <kuriko@versi0n.io>
 License: MIT License
@@ -48,8 +48,8 @@ Requires-Dist: regex==2024.11.6
 Requires-Dist: requests>=2.32.3
 Requires-Dist: pydantic>=2.10.6
 Requires-Dist: werkzeug>=3.1.3
-Requires-Dist: typing
-Requires-Dist: json-repair
+Requires-Dist: typing>=0.0.0
+Requires-Dist: json-repair>=0.0.0
 Requires-Dist: litellm>=1.55.8
 Requires-Dist: openai>=1.64.0
 Requires-Dist: composio-openai>=0.6.9
@@ -68,27 +68,24 @@ Requires-Dist: composio-core==0.7.0
 Requires-Dist: networkx>=3.4.2
 Requires-Dist: matplotlib>=3.10.0
 Requires-Dist: boto3>=1.37.1
-Provides-Extra: torch
-Requires-Dist: torch>=2.6.0; extra == "torch"
-Requires-Dist: torchvision>=0.21.0; extra == "torch"
-Requires-Dist: pytorch-triton-xpu>=3.2.0; sys_platform == "linux" and extra == "torch"
 Provides-Extra: docling
 Requires-Dist: docling>=2.25.2; extra == "docling"
 Provides-Extra: mem0ai
 Requires-Dist: mem0ai>=0.1.55; extra == "mem0ai"
 Provides-Extra: pdfplumber
 Requires-Dist: pdfplumber>=0.11.5; extra == "pdfplumber"
-Provides-Extra: pandas
-Requires-Dist: pandas>=2.2.3; extra == "pandas"
-Provides-Extra: numpy
-Requires-Dist: numpy>=1.26.4; extra == "numpy"
 Provides-Extra: pygraphviz
 Requires-Dist: pygraphviz>=1.14; extra == "pygraphviz"
 Provides-Extra: tools
 Requires-Dist: html2text>=2024.2.26; extra == "tools"
 Requires-Dist: sec-api>=1.0.28; extra == "tools"
-Provides-Extra: eval
-Requires-Dist: scikit-learn>=1.6.1; extra == "eval"
+Provides-Extra: torch
+Requires-Dist: torch>=2.6.0; extra == "torch"
+Requires-Dist: torchvision>=0.21.0; extra == "torch"
+Provides-Extra: evals
+Requires-Dist: scikit-learn>=1.6.1; extra == "evals"
+Requires-Dist: numpy>=1.26.4; extra == "evals"
+Requires-Dist: pandas>=2.2.3; extra == "evals"
 
 # Overview
 

{versionhq-1.2.3.0.dist-info → versionhq-1.2.3.4.dist-info}/RECORD

@@ -1,4 +1,4 @@
-versionhq/__init__.py,sha256=Co8oha5gA6GLy73HAkQ-HmC8IaXLGOK5YeIOT6KMnDc,2980
+versionhq/__init__.py,sha256=OB2B9WOc-P9RyhFLEhs2zS0E34d7GwED1POMAfSp4Cg,2980
 versionhq/_utils/__init__.py,sha256=d-vYVcORZKG-kkLe_fzE8VbViDpAk9DDOKe2fVK25ew,178
 versionhq/_utils/i18n.py,sha256=TwA_PnYfDLA6VqlUDPuybdV9lgi3Frh_ASsb_X8jJo8,1483
 versionhq/_utils/llm_as_a_judge.py,sha256=RM0oYfoeanuUyUL3Ewl6_8Xn1F5Axd285UMH46kxG1I,2378
@@ -23,7 +23,7 @@ versionhq/clients/customer/model.py,sha256=_AtaVVMm9MgCwrQ-HTRQ2oXUMKrSCEfZwE2Jd
 versionhq/clients/product/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/clients/product/model.py,sha256=3w__pug9XRe4LIm9wX8C8WKqi40r081Eb1q2vWk9UaU,3694
 versionhq/clients/workflow/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-versionhq/clients/workflow/model.py,sha256=fqmTx8Y6P6i-sK045ENO88GDbf4WYY4hUA26RfEH8Ek,6011
+versionhq/clients/workflow/model.py,sha256=_yCbmwzexjQqEGli7XX0vFG6yFnAOe9QDoMWQaU_qZE,3920
 versionhq/knowledge/__init__.py,sha256=qW7IgssTA4_bFFV9ziOcYRfGjlq1c8bkb-HnfWknpuQ,567
 versionhq/knowledge/_utils.py,sha256=YWRF8U533cfZes_gZqUvdj-K24MD2ri1R0gjc_aPYyc,402
 versionhq/knowledge/embedding.py,sha256=KfHc__1THxb5jrg1EMrF-v944RDuIr2hE0l-MtM3Bp0,6826
@@ -39,17 +39,17 @@ versionhq/memory/contextual_memory.py,sha256=QEMVvHuEXxY7M6-12S8HhyFKf108KfX8Zzt
 versionhq/memory/model.py,sha256=VQR1229t7GQPMItlGAHLtJrb6LrZfSoRA1DRW4z0SOU,8234
 versionhq/storage/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/storage/base.py,sha256=p-Jas0fXQan_qotnRD6seQxrT2lj-uw9-SmHQhdppcs,355
-versionhq/storage/ltm_sqlite_storage.py,sha256=K0Assani1QfFuq5FMxkcMmRlmLUOvw-RNkKyMV70U3g,3956
+versionhq/storage/ltm_sqlite_storage.py,sha256=LeJE4ZPUWjyY1E5nNCHoKujTHFDR2BO_LAMvAOX-WHg,3957
 versionhq/storage/mem0_storage.py,sha256=ZY8MELBWaINRv9YuRW5MxH7dj2cII-L0i3xSD6o1-2M,3781
 versionhq/storage/rag_storage.py,sha256=bS2eE874obarYl-4hT6ZWYWTRsqtfuGpKgKzERmM6Uo,7433
 versionhq/storage/task_output_storage.py,sha256=M8vInLJ5idGAq17w1juHKXtyPyF-B-rK_P8UcqD-Px8,5357
 versionhq/storage/utils.py,sha256=r5ghA_ktdR2IuzlzKqZYCjsNxztEMzyhWLneA4cFuWY,748
 versionhq/task/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-versionhq/task/evaluation.py,sha256=iRLzppqwKaiGpbsr9gMbf6T7NQe6rxTA6OBcWhmiCKs,4473
+versionhq/task/evaluation.py,sha256=yKnqj3UwdNAMaFEBZmWWf0Hc09-M9hM8VETnaovlFco,3817
 versionhq/task/formatter.py,sha256=N8Kmk9vtrMtBdgJ8J7RmlKNMdZWSmV8O1bDexmCWgU0,643
-versionhq/task/model.py,sha256=rJmyEUM1DnX1mkN_0etYXqBJP_YADXVXZhFR5R-h8ZA,28915
+versionhq/task/model.py,sha256=8HQLzzAfM03gHL5M_7oL7UW-mZvI7rPCJ2Jxsv49VtI,29122
 versionhq/task/structured_response.py,sha256=4q-hQPu7oMMHHXEzh9YW4SJ7N5eCZ7OfZ65juyl_jCI,5000
-versionhq/task/TEMPLATES/Description.py,sha256=EkwJHc65G32MjWyn3rcp0ATmMaVPHuYKaykyByU5r4g,751
+versionhq/task/TEMPLATES/Description.py,sha256=hKhpbz0ztbkUMXz9KiL-P40fis9OB5ICOdL9jCtgAhU,864
 versionhq/task_graph/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/task_graph/colors.py,sha256=naJCx4Vho4iuJtbW8USUXb-M5uYvd5ds2p8qbjUfRus,669
 versionhq/task_graph/draft.py,sha256=l18XacRsbDhAv6CvKMnUMI26IDuizA1UNWHbL1q5gn4,5099
@@ -62,8 +62,8 @@ versionhq/tool/decorator.py,sha256=C4ZM7Xi2gwtEMaSeRo-geo_g_MAkY77WkSLkAuY0AyI,1
 versionhq/tool/model.py,sha256=Nc2f9frTK5tH4kh6EeEAk1Fi1w19kEXLOcsBwHCS1a4,12189
 versionhq/tool/rag_tool.py,sha256=dW5o-83V4bMFFJEj3PUm7XjblwrYJGmZVBlCpPj6CeM,3852
 versionhq/tool/tool_handler.py,sha256=2m41K8qo5bGCCbwMFferEjT-XZ-mE9F0mDUOBkgivOI,1416
-versionhq-1.2.3.0.dist-info/LICENSE,sha256=cRoGGdM73IiDs6nDWKqPlgSv7aR4n-qBXYnJlCMHCeE,1082
-versionhq-1.2.3.0.dist-info/METADATA,sha256=ecBan7bE1D3h9QCoGrmez3swRMqLhJsxilByj0pJCyg,21535
-versionhq-1.2.3.0.dist-info/WHEEL,sha256=jB7zZ3N9hIM9adW7qlTAyycLYW9npaWKLRzaoVcLKcM,91
-versionhq-1.2.3.0.dist-info/top_level.txt,sha256=DClQwxDWqIUGeRJkA8vBlgeNsYZs4_nJWMonzFt5Wj0,10
-versionhq-1.2.3.0.dist-info/RECORD,,
+versionhq-1.2.3.4.dist-info/LICENSE,sha256=cRoGGdM73IiDs6nDWKqPlgSv7aR4n-qBXYnJlCMHCeE,1082
+versionhq-1.2.3.4.dist-info/METADATA,sha256=6ctQ8pQeU6rOuUOO0AxAmVy4v0MYDk2QJbZyCWHFBR0,21418
+versionhq-1.2.3.4.dist-info/WHEEL,sha256=jB7zZ3N9hIM9adW7qlTAyycLYW9npaWKLRzaoVcLKcM,91
+versionhq-1.2.3.4.dist-info/top_level.txt,sha256=DClQwxDWqIUGeRJkA8vBlgeNsYZs4_nJWMonzFt5Wj0,10
+versionhq-1.2.3.4.dist-info/RECORD,,