judgeval 0.0.29__py3-none-any.whl → 0.0.31__py3-none-any.whl

judgeval/scorers/judgeval_scorer.py

@@ -34,6 +34,7 @@ class JudgevalScorer:
     async_mode: bool = True  # Whether to run the scorer in async mode
     verbose_mode: bool = True  # Whether to run the scorer in verbose mode
     include_reason: bool = False  # Whether to include the reason in the output
+    custom_example: bool = False  # Whether the scorer corresponds to CustomExamples
     error: Optional[str] = None  # The error message if the scorer failed
     evaluation_cost: Optional[float] = None  # The cost of running the scorer
     verbose_logs: Optional[str] = None  # The verbose logs of the scorer
@@ -52,6 +53,7 @@ class JudgevalScorer:
         async_mode: bool = True, 
         verbose_mode: bool = True, 
         include_reason: bool = False, 
+        custom_example: bool = False,
         error: Optional[str] = None, 
         evaluation_cost: Optional[float] = None, 
         verbose_logs: Optional[str] = None, 
@@ -78,6 +80,7 @@ class JudgevalScorer:
             self.async_mode = async_mode
             self.verbose_mode = verbose_mode
             self.include_reason = include_reason
+            self.custom_example = custom_example
             self.error = error
             self.evaluation_cost = evaluation_cost
             self.verbose_logs = verbose_logs

judgeval/scorers/judgeval_scorers/__init__.py

@@ -15,6 +15,7 @@ from judgeval.scorers.judgeval_scorers.api_scorers import (
     ComparisonScorer as APIComparisonScorer,
     InstructionAdherenceScorer as APIInstructionAdherenceScorer,
     GroundednessScorer as APIGroundednessScorer,
+    DerailmentScorer as APIDerailmentScorer,
 )
 
 from judgeval.scorers.judgeval_scorers.local_implementations import (
@@ -153,6 +154,11 @@ GroundednessScorer = ScorerWrapper(
     api_implementation=APIGroundednessScorer,
 )
 
+DerailmentScorer = ScorerWrapper(
+    api_implementation=APIDerailmentScorer,
+    local_implementation=LocalInstructionAdherenceScorer # TODO: add local implementation
+)
+
 __all__ = [
     "ExecutionOrderScorer",
     "JSONCorrectnessScorer",
@@ -166,4 +172,5 @@ __all__ = [
     "Text2SQLScorer",
     "ComparisonScorer",
     "GroundednessScorer",
+    "DerailmentScorer",
 ]

judgeval/scorers/judgeval_scorers/api_scorers/__init__.py

@@ -11,7 +11,7 @@ from judgeval.scorers.judgeval_scorers.api_scorers.answer_correctness import Ans
 from judgeval.scorers.judgeval_scorers.api_scorers.comparison import ComparisonScorer
 from judgeval.scorers.judgeval_scorers.api_scorers.instruction_adherence import InstructionAdherenceScorer
 from judgeval.scorers.judgeval_scorers.api_scorers.groundedness import GroundednessScorer
-
+from judgeval.scorers.judgeval_scorers.api_scorers.derailment_scorer import DerailmentScorer
 __all__ = [
     "ExecutionOrderScorer",
     "JSONCorrectnessScorer",
@@ -26,4 +26,5 @@ __all__ = [
     "ComparisonScorer",
     "InstructionAdherenceScorer",
     "GroundednessScorer",
+    "DerailmentScorer",
 ]

judgeval/scorers/judgeval_scorers/api_scorers/derailment_scorer.py

@@ -0,0 +1,21 @@
+"""
+`judgeval` answer relevancy scorer
+
+TODO add link to docs page for this scorer
+
+"""
+
+# Internal imports
+from judgeval.scorers.api_scorer import APIJudgmentScorer
+from judgeval.constants import APIScorer
+
+class DerailmentScorer(APIJudgmentScorer):
+    def __init__(self, threshold: float):
+        super().__init__(
+            threshold=threshold, 
+            score_type=APIScorer.DERAILMENT,
+        )
+
+    @property
+    def __name__(self):
+        return "Derailment"

judgeval/scorers/score.py

@@ -11,6 +11,7 @@ from rich.progress import Progress, SpinnerColumn, TextColumn
 
 from judgeval.data import (
     Example, 
+    CustomExample,
     ScoringResult,
     generate_scoring_result,
     create_scorer_data,
@@ -240,7 +241,7 @@ async def score_with_indicator(
 
 
 async def a_execute_scoring(
-    examples: List[Example],
+    examples: Union[List[Example], List[CustomExample]],
     scorers: List[JudgevalScorer],
     model: Optional[Union[str, List[str], JudgevalJudge]] = None,
     ignore_errors: bool = True,
@@ -256,7 +257,7 @@ async def a_execute_scoring(
     Each `Example` will be evaluated by all of the `JudgevalScorer`s in the `scorers` list.
 
     Args:
-        examples (List[Example]): A list of `Example` objects to be evaluated.
+        examples (Union[List[Example], List[CustomExample]]): A list of `Example` objects to be evaluated.
         scorers (List[JudgevalScorer]): A list of `JudgevalScorer` objects to evaluate the examples.
         model (Union[str, List[str], JudgevalJudge]): The model to use for evaluation.
         ignore_errors (bool): Whether to ignore errors during evaluation.
@@ -313,7 +314,7 @@ async def a_execute_scoring(
                             debug(f"Scorer threshold: {scorer.threshold}")
                         if hasattr(scorer, 'model'):
                             debug(f"Scorer model: {type(scorer.model).__name__}")
-                if isinstance(ex, Example):
+                if isinstance(ex, Example) or isinstance(ex, CustomExample):
                     if len(scorers) == 0:
                         pbar.update(1)
                         continue
@@ -339,7 +340,7 @@ async def a_execute_scoring(
             await asyncio.gather(*tasks)
     else:
         for i, ex in enumerate(examples):
-            if isinstance(ex, Example):
+            if isinstance(ex, Example) or isinstance(ex, CustomExample):
                 if len(scorers) == 0:
                     continue
 
@@ -366,7 +367,7 @@ async def a_execute_scoring(
 
 async def a_eval_examples_helper(
     scorers: List[JudgevalScorer],
-    example: Example,
+    example: Union[Example, CustomExample],
     scoring_results: List[ScoringResult],
     score_index: int,
     ignore_errors: bool,

{judgeval-0.0.29.dist-info → judgeval-0.0.31.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: judgeval
-Version: 0.0.29
+Version: 0.0.31
 Summary: Judgeval Package
 Project-URL: Homepage, https://github.com/JudgmentLabs/judgeval
 Project-URL: Issues, https://github.com/JudgmentLabs/judgeval/issues

{judgeval-0.0.29.dist-info → judgeval-0.0.31.dist-info}/RECORD

@@ -1,46 +1,49 @@
 judgeval/__init__.py,sha256=dtXxsCmI4eEsZdGSUMy8P_pA0bc2-OSGAgb2C__yJoA,252
 judgeval/clients.py,sha256=6VQmEqmfCngUdS2MuPBIpHvtDFqOENm8-_BmMvjLyRQ,944
-judgeval/constants.py,sha256=ksAXhAXovzJKH0uHOdQtREs168uCJRG79PooHNmEbYQ,5313
-judgeval/evaluation_run.py,sha256=6Kft3wZDWkdBDZoMwOhWf7zSAOF4naI7Pcg_YlZaZY4,6394
-judgeval/judgment_client.py,sha256=uf0V1-eu3qnFTwrQ_Ckcv8IiWRVv7dbvou4P4KjU6hM,26794
+judgeval/constants.py,sha256=XTqijsuuLEhUBXTjzNJVsee5U_Gl14ULLO5uQVW_nEE,5398
+judgeval/evaluation_run.py,sha256=WGzx-Ug2qhSmunFo8NrmSstBRsOUc5KpKq0Lc51rqsM,6739
+judgeval/judgment_client.py,sha256=FncHkjyFx2vfXv4cu4DzbOO0ideHNOWtHVbc8pSXNxk,29754
 judgeval/rules.py,sha256=B0ZL0pn72D4Jnlr0zMQ6CPHi7D8AQQRariXCVsiCMiI,20542
-judgeval/run_evaluation.py,sha256=N2ppmEE5WoSReChKjr_n0NcdAUlUR6Nua7M1C_3zHQ8,24949
+judgeval/run_evaluation.py,sha256=2Mv1iLthJeFQZSVhjLOcJKRZ52Sy6OxLb2KyQ_yVwnA,28484
 judgeval/common/__init__.py,sha256=7d24BRxtncpMj3AAJCj8RS7TqgjXmW777HVZH6-3sBs,289
 judgeval/common/exceptions.py,sha256=U-TxHLn7oVMezsMuoYouNDb2XuS8RCggfntYf5_6u4E,565
 judgeval/common/logger.py,sha256=KO75wWXCxhUHUMvLaTU31ZzOk6tkZBa7heQ7y0f-zFE,6062
-judgeval/common/tracer.py,sha256=b-eQyC_MPwMAVQJS6wtVW0_7hzk8tC9EV6NZZoNjWos,58188
+judgeval/common/tracer.py,sha256=9Qga-7rLFlQK-oM5eK1O_8Mn1SewIrPtFwWbSZFtSII,59651
 judgeval/common/utils.py,sha256=LUQV5JfDr6wj7xHAJoNq-gofNZ6mjXbeKrGKzBME1KM,33533
-judgeval/data/__init__.py,sha256=dG5ytBOeOWCTd5o0KP7IblqtW4G1EBaGreLWepM3jas,345
-judgeval/data/custom_api_example.py,sha256=uW_ZBzkDLWumtudmfRHAJQkVYpm2qWgcDf7vBNLpS-o,3444
-judgeval/data/example.py,sha256=BhGBhamFWgH6wtvrRYM8dGtDfXh-cDxDhtNL5Gbdz_M,5892
-judgeval/data/result.py,sha256=BT4f2FF5EFuiRjOmS4vuIXsrEwSlG16Vw3QaWi6PZzc,3122
+judgeval/data/__init__.py,sha256=xuKx_KCVHGp6CXvQuVmKl3v7pJp-qDaz0NccKxwjtO0,481
+judgeval/data/custom_example.py,sha256=QRBqiRiZS8UgVeTRHY0r1Jzm6yAYsyg6zmHxQGxdiQs,739
+judgeval/data/example.py,sha256=cJrmPGLel_P2sy1UaRvuVSAi35EnA9XMR11Lhp4aDLo,5930
+judgeval/data/result.py,sha256=Gb9tiSDsk1amXgh0cFG6JmlW_BMKxS2kuTwNA0rrHjA,3184
 judgeval/data/scorer_data.py,sha256=JVlaTx1EP2jw2gh3Vgx1CSEsvIFABAN26IquKyxwiJQ,3273
+judgeval/data/sequence.py,sha256=DlQUjyWQJB6iNmiftDZ9N6C-nPtrOC1e0JZ57U00zZk,2387
+judgeval/data/sequence_run.py,sha256=GrnYSZBcZmt4tKQYA_1v09MFB8n3ccrkOJd4qyweHMg,1987
 judgeval/data/datasets/__init__.py,sha256=IdNKhQv9yYZ_op0rdBacrFaFVmiiYQ3JTzXzxOTsEVQ,176
 judgeval/data/datasets/dataset.py,sha256=AFYjksV_wXx5CqFYJsl3aN8yZ6hC50O1myRuOJ8s8_E,12867
-judgeval/data/datasets/eval_dataset_client.py,sha256=P9fEmcNrjPPaiYbbLiEiBziZrIexA39HN9qzClt6uPE,12691
-judgeval/integrations/langgraph.py,sha256=fGDZOTlVbxTO4ErC-m9OSg3h-RkOIIWXCfhjgkKRh4E,11187
+judgeval/data/datasets/eval_dataset_client.py,sha256=xzXlBJRBEEmwsB79_eepm0Da-Bz8yRodX7ttk-u-BxU,14986
+judgeval/integrations/langgraph.py,sha256=J-cQfFP52TjJewdSTe-fcsUC4HDvjNbXoxmbmF0SgiE,11743
 judgeval/judges/__init__.py,sha256=6X7VSwrwsdxGBNxCyapVRWGghhKOy3MVxFNMQ62kCXM,308
 judgeval/judges/base_judge.py,sha256=ch_S7uBB7lyv44Lf1d7mIGFpveOO58zOkkpImKgd9_4,994
 judgeval/judges/litellm_judge.py,sha256=EIL58Teptv8DzZUO3yP2RDQCDq-aoBB6HPZzPdK6KTg,2424
 judgeval/judges/mixture_of_judges.py,sha256=IJoi4Twk8ze1CJWVEp69k6TSqTCTGrmVYQ0qdffer60,15549
 judgeval/judges/together_judge.py,sha256=l00hhPerAZXg3oYBd8cyMtWsOTNt_0FIqoxhKJKQe3k,2302
 judgeval/judges/utils.py,sha256=9lvUxziGV86ISvVFxYBWc09TWFyAQgUTyPf_a9mD5Rs,2686
-judgeval/scorers/__init__.py,sha256=gkeKJvjXhswCnkEyjijrVvGVM3Om86egrZ-PUOGvNvI,1158
+judgeval/scorers/__init__.py,sha256=Z_88Sr45gLFAIbMHzG1BF24TUQGCDiuP9QpmVFvSYJM,1204
 judgeval/scorers/api_scorer.py,sha256=NQ_CrrUPhSUk1k2Q8rKpCG_TU2FT32sFEqvb-Yi54B0,2688
 judgeval/scorers/base_scorer.py,sha256=xdUlY3CnLdCQ1Z5iUeY22Bim5v-OQruZmaVF_4Y1mC0,2183
 judgeval/scorers/exceptions.py,sha256=eGW5CuJgZ5YJBFrE4FHDSF651PO1dKAZ379mJ8gOsfo,178
-judgeval/scorers/judgeval_scorer.py,sha256=jq_rzfTG0XBTuLCaa6TlaK4YcT-LlgsO1LEm6hpOYdg,6601
+judgeval/scorers/judgeval_scorer.py,sha256=79-JJurqHP-qTaWNWInx4SjvQYwXc9lvfPPNgwsh2yA,6773
 judgeval/scorers/prompt_scorer.py,sha256=PaAs2qRolw1P3_I061Xvk9qzvF4O-JR8g_39RqXnHcM,17728
-judgeval/scorers/score.py,sha256=ObFAlMbNRcGrfBpH4WW_6OA3CjrneC539xSWhGH60GQ,18578
+judgeval/scorers/score.py,sha256=r9QiT4-LIvivcJ6XxByrbswKSO8eQTtAD1UlXT_lcmo,18741
 judgeval/scorers/utils.py,sha256=iHQVTlIANbmCTXz9kTeSdOytgUZ_T74Re61ajqsk_WQ,6827
-judgeval/scorers/judgeval_scorers/__init__.py,sha256=xFRb62sp4JmBUSeuAB_pC_7kEGp-lGdqCRIu9--Bbdg,5992
-judgeval/scorers/judgeval_scorers/api_scorers/__init__.py,sha256=mZ6b_5Dl04k3PaG24ICBajB_j43ody1II1OJhO1DkXo,1648
+judgeval/scorers/judgeval_scorers/__init__.py,sha256=kSmQWKeBvLeZMfLYNQSc2qbJYo1MFIQnf3P-D4ltuSM,6232
+judgeval/scorers/judgeval_scorers/api_scorers/__init__.py,sha256=_sDUBxSG536KGqXNi6dFpaYKghjEAadxBxaaxV9HuuE,1764
 judgeval/scorers/judgeval_scorers/api_scorers/answer_correctness.py,sha256=Fnd9CVIOZ73sWEWymsU5eBrrZqPFjMZ0BKpeW-PDyTg,711
 judgeval/scorers/judgeval_scorers/api_scorers/answer_relevancy.py,sha256=oETeN9K0HSIRdL2SDqn82Vskpwh5SlKnZvs5VDm2OBU,658
 judgeval/scorers/judgeval_scorers/api_scorers/comparison.py,sha256=kuzf9OWvpY38yYSwlBgneLkUZwJNM4FQqvbS66keA90,1249
 judgeval/scorers/judgeval_scorers/api_scorers/contextual_precision.py,sha256=tpSuzFAaW8X9xqA0aLLKwh7qmBK0Pc_bJZMIe_q412U,770
 judgeval/scorers/judgeval_scorers/api_scorers/contextual_recall.py,sha256=pFVhk4pLtQ-FnNlbI-dFF-SIh69Jza7erHqiPkFWoBo,758
 judgeval/scorers/judgeval_scorers/api_scorers/contextual_relevancy.py,sha256=RQ6DZwEhChfecd89Ey-T7ke--7qTaXZlRsNxwH8gaME,823
+judgeval/scorers/judgeval_scorers/api_scorers/derailment_scorer.py,sha256=V9WPuwNMm097V7IknKs8UkmAk0yjnBXTcJha_BHXxTA,475
 judgeval/scorers/judgeval_scorers/api_scorers/execution_order.py,sha256=Pb3CiNF2Ca826B92wJCVAi_68lJjLhqqCKwQKaflSUg,1294
 judgeval/scorers/judgeval_scorers/api_scorers/faithfulness.py,sha256=-BwOapqjryYNKNydtdkUiKIij76dY0O1jBmdc6dKazQ,692
 judgeval/scorers/judgeval_scorers/api_scorers/groundedness.py,sha256=ntEEeTANEOsGlcbiTAF_3r6BeSJEaVDns8po8T0L6Vg,692
@@ -87,7 +90,7 @@ judgeval/scorers/judgeval_scorers/local_implementations/summarization/prompts.py
 judgeval/scorers/judgeval_scorers/local_implementations/summarization/summarization_scorer.py,sha256=Qk7lwHgRPYeGoxTOyclAh1VfGItfvHJ6l1t7Nk3SWFM,20927
 judgeval/tracer/__init__.py,sha256=wy3DYpH8U_z0GO_K_gOSkK0tTTD-u5eLDo0T5xIBoAc,147
 judgeval/utils/alerts.py,sha256=O19Xj7DA0YVjl8PWiuH4zfdZeu3yiLVvHfY8ah2wG0g,2759
-judgeval-0.0.29.dist-info/METADATA,sha256=UAJZhCDCKCMOSnvYVZ9M2v3O0QdMErKdYWeN2NBp2_M,5418
-judgeval-0.0.29.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-judgeval-0.0.29.dist-info/licenses/LICENSE.md,sha256=tKmCg7k5QOmxPK19XMfzim04QiQJPmgIm0pAn55IJwk,11352
-judgeval-0.0.29.dist-info/RECORD,,
+judgeval-0.0.31.dist-info/METADATA,sha256=g9288fIE7NDwXuqUylqCV0mby5hAY7yEztR8TOn5sNk,5418
+judgeval-0.0.31.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+judgeval-0.0.31.dist-info/licenses/LICENSE.md,sha256=tKmCg7k5QOmxPK19XMfzim04QiQJPmgIm0pAn55IJwk,11352
+judgeval-0.0.31.dist-info/RECORD,,

judgeval/data/custom_api_example.py

@@ -1,91 +0,0 @@
-from typing import List, Optional, Dict, Any, Union
-from pydantic import BaseModel, ConfigDict, model_validator
-
-from judgeval.data.example import Example
-from judgeval.data.custom_example import CustomExample
-from judgeval.data.scorer_data import ScorerData
-from judgeval.common.logger import debug, error
-
-class ProcessExample(BaseModel):
-    """
-    ProcessExample is an `Example` object that contains intermediate information 
-    about an undergoing evaluation on the original `Example`. It is used purely for
-    internal operations and keeping track of the evaluation process.
-    """
-    name: str
-    # input: Optional[str] = None
-    # actual_output: Optional[Union[str, List[str]]] = None
-    # expected_output: Optional[Union[str, List[str]]] = None
-    # context: Optional[list] = None
-    # retrieval_context: Optional[list] = None
-    # tools_called: Optional[list] = None
-    # expected_tools: Optional[list] = None
-
-    # make these optional, not all test cases in a conversation will be evaluated
-    success: Optional[bool] = None
-    scorers_data: Optional[List[ScorerData]] = None
-    run_duration: Optional[float] = None 
-    evaluation_cost: Optional[float] = None
-
-    order: Optional[int] =  None
-    # These should map 1 to 1 from golden
-    additional_metadata: Optional[Dict] = None
-    comments: Optional[str] = None
-    trace_id: Optional[str] = None
-    model_config = ConfigDict(arbitrary_types_allowed=True)
-
-    def update_scorer_data(self, scorer_data: ScorerData):
-        """
-        Updates scorer data field of test case after the scorers have been
-        evaluated on this test case.
-        """
-        debug(f"Updating scorer data for example '{self.name}' with scorer: {scorer_data}")
-        # self.scorers_data is a list of ScorerData objects that contain the 
-        # evaluation results of each scorer on this test case
-        if self.scorers_data is None:
-            self.scorers_data = [scorer_data]
-        else:
-            self.scorers_data.append(scorer_data)
-
-        if self.success is None:
-            # self.success will be None when it is a message
-            # in that case we will be setting success for the first time
-            self.success = scorer_data.success
-        else:
-            if scorer_data.success is False:
-                debug(f"Example '{self.name}' marked as failed due to scorer: {scorer_data}")
-                self.success = False
-
-    def update_run_duration(self, run_duration: float):
-        self.run_duration = run_duration
-
-
-def create_process_custom_example(
-    example: CustomExample,
-) -> ProcessExample:
-    """
-    When an LLM Test Case is executed, we track its progress using an ProcessExample.
-
-    This will track things like the success of the test case, as well as the metadata (such as verdicts and claims in Faithfulness).
-    """
-    success = True
-    if example.name is not None:
-        name = example.name
-    else:
-        name = "Test Case Placeholder"
-        debug(f"No name provided for example, using default name: {name}")
-    order = None
-    scorers_data = []
-
-    debug(f"Creating ProcessExample for: {name}")
-    process_ex = ProcessExample(
-        name=name,
-        success=success,
-        scorers_data=scorers_data,
-        run_duration=None,
-        evaluation_cost=None,
-        order=order,
-        additional_metadata=example.additional_metadata,
-        trace_id=example.trace_id
-    )
-    return process_ex