judgeval 0.4.0__py3-none-any.whl → 0.6.0__py3-none-any.whl

{judgeval-0.4.0.dist-info → judgeval-0.6.0.dist-info}/RECORD

@@ -1,53 +1,57 @@
-judgeval/__init__.py,sha256=HM1M8hmqRum6G554QKkXhB4DF4f5eh_xtYo0Kf-t3kw,332
-judgeval/clients.py,sha256=JnB8n90GyXiYaGmSEYaA67mdJSnr3SIrzArao7NGebw,980
-judgeval/constants.py,sha256=hWed25HwGUJy-tePbtoUZ0_Zg0X_MkAH84KiH-OHHFI,4150
-judgeval/dataset.py,sha256=rjV54XNTslNNtf-Uu2ndDIh602ZwSCFhPg2NuckDJ-w,6081
-judgeval/evaluation_run.py,sha256=edNpO444Fwt2ykWsflIzlYdDJUlUfbpXHHQSKfFS4y0,2876
-judgeval/judgment_client.py,sha256=vPoxbmxAlhbG5rXXqxWjMbyEqOI044BaQanr1fev2CE,11723
+judgeval/__init__.py,sha256=5Lm1JMYFREJGN_8X-Wpruu_ovwGLJ08gCzNAt-u-pQE,419
+judgeval/cli.py,sha256=IcL4_bGr9CtEeea1-AFqM_TEV_VomDlArlxh4IomiSQ,1754
+judgeval/clients.py,sha256=HHul68PV1om0dxsVZZu90TtCiy5zaqAwph16jXTQzQo,989
+judgeval/constants.py,sha256=UNoTLHgbpZHRInPM2ZaI3m0XokPkee5ILlg20reqhzo,4180
+judgeval/dataset.py,sha256=vOrDKam2I-K1WcVF5IBkQruCDvXTc8PRaFm4-dV0lXs,6220
+judgeval/judgment_client.py,sha256=-7xcBFowzXKedMINwfZCOL4FKucECWPNEY9QVMo_cys,13644
+judgeval/local_eval_queue.py,sha256=GmlXeZt7bfAJe1hPUjDg_irth4RkNqL2Zdi7VzboBzI,6984
 judgeval/rules.py,sha256=CoQjqmP8daEXewMkplmA-7urubDtweOr5O6z8klVwLI,20031
-judgeval/run_evaluation.py,sha256=7J6FHhWhB-IDPMSOcWkrjTpSNm2v3s_KBq8Np3y2pys,27652
+judgeval/run_evaluation.py,sha256=gs-_v_i95LKlJj95G2RmQXvIyBfoldnd1pWCNO4UqsM,21985
 judgeval/version_check.py,sha256=FoLEtpCjDw2HuDQdpw5yT29UtwumSc6ZZN6AV_c9Mnw,1057
 judgeval/common/__init__.py,sha256=KH-QJyWtQ60R6yFIBDYS3WGRiNpEu1guynpxivZvpBQ,309
 judgeval/common/exceptions.py,sha256=OkgDznu2wpBQZMXiZarLJYNk1HIcC8qYW7VypDC3Ook,556
 judgeval/common/logger.py,sha256=514eFLYWS_UL8VY-zAR2ePUlpQe4rbYlleLASFllLE4,1511
 judgeval/common/utils.py,sha256=oxGDRVWOICKWeyGgsoc36_yAyHSYF4XtH842Mkznwis,34739
 judgeval/common/api/__init__.py,sha256=-E7lpZz1fG8puR_aYUMfPmQ-Vyhd0bgzoaU5EhIuFjQ,114
-judgeval/common/api/api.py,sha256=0Vd81CXoUYF1a7hOlI2PqXJcQO7eqWEbS8sX2W0CJMY,14184
-judgeval/common/api/constants.py,sha256=DXej0m8HEhb871SdiR8t_o4fzeMoQjHYqb_X0Plj8wY,4577
+judgeval/common/api/api.py,sha256=fWtMNln0o1wOhJ9wangWpyY_j3WF7P3at_LYPJEicP0,13670
+judgeval/common/api/constants.py,sha256=y0BDcQqHBZ7MwLd4gT5hLUF8UMs_GVwsJGC-ibfxCAw,4698
+judgeval/common/api/json_encoder.py,sha256=QQgCe2FBmW1uWKx8yvuhr4U7_b4D0sG97GZtXHKnBdk,5881
 judgeval/common/storage/__init__.py,sha256=a-PI7OL-ydyzugGUKmJKRBASnK-Q-gs82L9K9rSyJP8,90
 judgeval/common/storage/s3_storage.py,sha256=0-bNKheqJJyBZ92KGrzQtd1zocIRWBlfn_58L4a-Ay0,3719
 judgeval/common/tracer/__init__.py,sha256=tJCJsmVmrL89Phv88gNCJ-j0ITPez6lh8vhMAAlLNSc,795
 judgeval/common/tracer/constants.py,sha256=yu5y8gMe5yb1AaBkPtAH-BNwIaAR3NwYCRoSf45wp5U,621
-judgeval/common/tracer/core.py,sha256=blHEh61CE5kZLYCgyRF4kU6dVzi_Ko6DrnBpw2-jByI,73973
+judgeval/common/tracer/core.py,sha256=TQ80NODaJx7gzmntevDLA3evVJ3m2Zy2s0Pwd7APG9Y,84867
 judgeval/common/tracer/otel_exporter.py,sha256=kZLlOQ6afQE4dmb9H1wgU4P3H5PG1D_zKyvnpWcT5Ak,3899
-judgeval/common/tracer/otel_span_processor.py,sha256=W7SM62KnxJ48vC9WllIHRKaLlvxkCwqYoT4KqZLfGNs,6497
-judgeval/common/tracer/span_processor.py,sha256=eFjTgSWSkM6BWE94CrvgafDg_WkxLsFL_MafwBG-p9M,1145
-judgeval/common/tracer/span_transformer.py,sha256=nCnwRC52OKfYRFnsOwGdPaqb_U17yn5S_9jfhv1GaLM,7803
+judgeval/common/tracer/otel_span_processor.py,sha256=BD-FKXaZft5_3zqy1Qe_tpkudVOLop9AGhBjZUgp-Z8,6502
+judgeval/common/tracer/providers.py,sha256=3c3YOtKuoBjlTL0rc2HAGnUpppqvsyzrN5H6EKCqEi0,2733
+judgeval/common/tracer/span_processor.py,sha256=1NQxNSVWcb8qCFLmslSVMnaWdkOZmiFJnxeeN0i6vnU,1150
+judgeval/common/tracer/span_transformer.py,sha256=cfzz6RpTCOG9Io9knNlwtAW34p3wyK-u8jSNMu24p1w,7382
 judgeval/common/tracer/trace_manager.py,sha256=ltiXcWC-68DRc8uSa28qHiWRSIBf6NpYOPkZYooR8tg,3086
 judgeval/data/__init__.py,sha256=1QagDcSQtfnJ632t9Dnq8d7XjAqhmY4mInOWt8qH9tM,455
+judgeval/data/evaluation_run.py,sha256=IirmYZ1_9N99eep7DDuoyshwjmpNK9bQCxCWXnnhhuI,4053
 judgeval/data/example.py,sha256=kRskIgsjwcvv2Y8jaPwV-PND7zlmMbFsvRVQ_b7SZY0,914
-judgeval/data/judgment_types.py,sha256=KE1HrFLfSxiu1zutaiZ7B7La9PGXIAsoWpo_5iy645c,8336
+judgeval/data/judgment_types.py,sha256=3nGCUZ1YJhXajhFlAQvax0SOJ8eLuORtquwwjMreJFw,9826
 judgeval/data/result.py,sha256=OtSnBUrdQpjyAqxXRLTW3wC9v9lOm_GqzL14ccRQxrg,2124
 judgeval/data/scorer_data.py,sha256=5QBHtvOIWOq0Rn9_uPJzAMRYMlWxMB-rXnG_6kV4Z4Y,2955
 judgeval/data/tool.py,sha256=iWQSdy5uNbIeACu3gQy1DC2oGYxRVYNfkkczWdQMAiA,99
-judgeval/data/trace.py,sha256=tDOuYFPUssQInjsmwyxcXq-W3IB29Vq340VzqafuKJc,6942
-judgeval/data/trace_run.py,sha256=c6pRSv09Vj016hxM49I3kMftCwWg8hhkfT_1kBXluSI,1600
+judgeval/data/trace.py,sha256=S781vVU1BvQ_kTS3s7UGYdmYVVxVGjDzWJHZpHedyf0,2834
+judgeval/data/trace_run.py,sha256=Oo1vDrJYX_itt4tt7PJf7fNKd0HE3fnBJxuIkRY8Wrg,1585
 judgeval/data/scripts/fix_default_factory.py,sha256=lvp2JwYZqz-XpD9LZNa3mANZVP-jJSZoNzolI6JWERM,591
 judgeval/data/scripts/openapi_transform.py,sha256=Sm04JClzyP1ga8KA3gkIdsae8Hlx-XU7-x0gHCQYOhg,3877
-judgeval/integrations/langgraph.py,sha256=kJXLsgBY7DgsUTZyVQ47deDgHm887brFHfyIbuyerGw,29986
+judgeval/integrations/langgraph.py,sha256=XsTNpKvXZmSf4TJBtRKSd5AB7S-Td9GTG5wZW9Npj6k,30062
 judgeval/judges/__init__.py,sha256=6X7VSwrwsdxGBNxCyapVRWGghhKOy3MVxFNMQ62kCXM,308
 judgeval/judges/base_judge.py,sha256=_dz0qWsKRxzXxpRY9l6mrxTRYPSF2FE4ZXkrzhZ4gbY,986
-judgeval/judges/litellm_judge.py,sha256=yt6QvwKMmxZcrUtjbn3EiO5aVg7CHM2YZkBCSQLS8jk,2509
-judgeval/judges/mixture_of_judges.py,sha256=cecQ8mRmz2-dDoZl2MGsrhZICkpIvRovGPK3su0kc8s,14889
+judgeval/judges/litellm_judge.py,sha256=K9yCGOmozt7sYO0u8CHWyZNi8mXnSR3pPkP8yVsvuRc,2561
+judgeval/judges/mixture_of_judges.py,sha256=iTNjTX4Le1nCwGRm9qfMCv1lQjgqoIw3OE0teiLubwo,14946
 judgeval/judges/together_judge.py,sha256=5FADUhs6-FN1ZVV_1D3-8_gu9mPbZiG0PYTpme41SfM,2336
-judgeval/judges/utils.py,sha256=0CF9qtIUQUL3-W-qTGpmTjZbkUUBAM6TslDsrCHnTBU,2725
+judgeval/judges/utils.py,sha256=_t6oYN9q63wyP7D4jI8X0bNmvVw7OfaE7uMTYDVS14E,2782
 judgeval/scorers/__init__.py,sha256=4H_cinTQ4EogZv59YEV-3U9EOTLppNwgAPTi1-jI9Fw,746
 judgeval/scorers/agent_scorer.py,sha256=TjwD_YglSywr3EowEojiCyg5qDgCRa5LRGc5nFdmIBc,703
 judgeval/scorers/api_scorer.py,sha256=xlhqkeMUBFxl8daSXOTWOYwZjBAz7o6b4sVD5f8cIHw,2523
-judgeval/scorers/base_scorer.py,sha256=eDfQk8N8TQfM1ayJDWr0NTdSQxcbk9-VZHd0Igb9EbI,2878
+judgeval/scorers/base_scorer.py,sha256=hKrLLh2DaxTgAfze8p_IapvsrogRCevYgfaNCDeOJzc,2869
 judgeval/scorers/example_scorer.py,sha256=2n45y3LMV1Q-ARyXLHqvVWETlnY1DqS7OLzPu9IBGz8,716
 judgeval/scorers/exceptions.py,sha256=ACDHK5-TWiF3NTk-wycaedpbrdobm-CvvC1JA_iP-Mk,179
-judgeval/scorers/score.py,sha256=SPoN5GG5sGKAt2aw_NgDj4-w5Mv1CGgaD9q-Sj1jRII,6626
+judgeval/scorers/score.py,sha256=SWyoqOOvyLpLy39tLyb_Q94sdh9r_IuDv6YNREw52lg,7546
 judgeval/scorers/utils.py,sha256=HQOYTJtNnsi_aPfMssePAaBbXpAv7LXgwUlWlDFuN2g,3965
 judgeval/scorers/judgeval_scorers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 judgeval/scorers/judgeval_scorers/api_scorers/__init__.py,sha256=GX4KkwPR2p-c0Y5mZingJa8EUfjAbMGhrmRBDBunOGw,1484
@@ -58,14 +62,16 @@ judgeval/scorers/judgeval_scorers/api_scorers/execution_order.py,sha256=NABO_iBd
 judgeval/scorers/judgeval_scorers/api_scorers/faithfulness.py,sha256=ps51bTgQsD9xGYsk1v9bx0WxQMqywSllCE9_xlJkLd8,531
 judgeval/scorers/judgeval_scorers/api_scorers/hallucination.py,sha256=SnFLvU4FGsMeUVUp0SGHSy_6wgfwr_vHPGnZx5YJl_Q,691
 judgeval/scorers/judgeval_scorers/api_scorers/instruction_adherence.py,sha256=aQzu-TiGqG74JDQ927evv5yGmnZw2AOolyHvlIhiUbI,683
-judgeval/scorers/judgeval_scorers/api_scorers/prompt_scorer.py,sha256=nx73DeoVkSqJTP1hYxMsJobG9HVWgMDN5-xFOXt_8Ts,7348
+judgeval/scorers/judgeval_scorers/api_scorers/prompt_scorer.py,sha256=1FsUGjQu3oa2rF-oqt32j-yA2YM33_trGTJ0HgagFJ0,7793
 judgeval/scorers/judgeval_scorers/api_scorers/tool_dependency.py,sha256=Mcp1CjMNyOax9UkvoRdSyUYdO2Os1-Nko43y89m2Luo,594
 judgeval/scorers/judgeval_scorers/api_scorers/tool_order.py,sha256=Z2FLGBC7m_CLx-CMgXVuTvYvN0vY5yOcWA0ImBkeBfY,787
 judgeval/tracer/__init__.py,sha256=wkuXtOGDCrwgPPXlh_sSJmvGuWaAMHyNzk1TzB5f9aI,148
 judgeval/utils/alerts.py,sha256=3w_AjQrgfmOZvfqCridW8WAnHVxHHXokX9jNzVFyGjA,3297
+judgeval/utils/async_utils.py,sha256=uNx1SopEc0quSjc8GBQqyba0SmCMAzv2NKIq6xYwttc,989
 judgeval/utils/file_utils.py,sha256=PWHRs8dUr8iDwpglSSk4Yjd7C6ZhDzUaO-jV3m7riHM,1987
 judgeval/utils/requests.py,sha256=K3gUKrwL6TvwYKVYO5OeLWdUHn9NiUPmnIXhZEiEaHU,1534
-judgeval-0.4.0.dist-info/METADATA,sha256=6ZJMXn13RNp-LZfoBUNtmjt7nCvthWU9qT9YNBNe7Wo,10348
-judgeval-0.4.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-judgeval-0.4.0.dist-info/licenses/LICENSE.md,sha256=tKmCg7k5QOmxPK19XMfzim04QiQJPmgIm0pAn55IJwk,11352
-judgeval-0.4.0.dist-info/RECORD,,
+judgeval-0.6.0.dist-info/METADATA,sha256=CulXMs0v5YrHjR3ntVX8xWKcZyxwEpo_nOYs_hkaeN8,10403
+judgeval-0.6.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+judgeval-0.6.0.dist-info/entry_points.txt,sha256=-eoeD-oDLn4A7MSgeBS9Akwanf3_0r0cgEleBcIOjg0,46
+judgeval-0.6.0.dist-info/licenses/LICENSE.md,sha256=tKmCg7k5QOmxPK19XMfzim04QiQJPmgIm0pAn55IJwk,11352
+judgeval-0.6.0.dist-info/RECORD,,

judgeval-0.6.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+judgeval = judgeval.cli:app

judgeval/evaluation_run.py

@@ -1,76 +0,0 @@
-from typing import List, Optional, Union
-from pydantic import BaseModel, field_validator, Field
-
-from judgeval.data import Example
-from judgeval.scorers import BaseScorer, APIScorerConfig
-from judgeval.constants import ACCEPTABLE_MODELS
-
-
-class EvaluationRun(BaseModel):
-    """
-    Stores example and evaluation scorers together for running an eval task
-
-    Args:
-        project_name (str): The name of the project the evaluation results belong to
-        eval_name (str): A name for this evaluation run
-        examples (List[Example]): The examples to evaluate
-        scorers (List[Union[JudgmentScorer, BaseScorer]]): A list of scorers to use for evaluation
-        model (str): The model used as a judge when using LLM as a Judge
-        metadata (Optional[Dict[str, Any]]): Additional metadata to include for this evaluation run, e.g. comments, dataset name, purpose, etc.
-    """
-
-    organization_id: Optional[str] = None
-    project_name: Optional[str] = Field(default=None, validate_default=True)
-    eval_name: Optional[str] = Field(default=None, validate_default=True)
-    examples: List[Example]
-    scorers: List[Union[APIScorerConfig, BaseScorer]]
-    model: Optional[str] = "gpt-4.1"
-    trace_span_id: Optional[str] = None
-    # API Key will be "" until user calls client.run_eval(), then API Key will be set
-    override: Optional[bool] = False
-    append: Optional[bool] = False
-
-    def model_dump(self, **kwargs):
-        data = super().model_dump(**kwargs)
-
-        data["scorers"] = [
-            scorer.model_dump() for scorer in self.scorers
-        ]  # Pydantic has problems with properly calling model_dump() on the scorers, so we need to do it manually
-        data["examples"] = [example.model_dump() for example in self.examples]
-
-        return data
-
-    @field_validator("examples")
-    def validate_examples(cls, v):
-        if not v:
-            raise ValueError("Examples cannot be empty.")
-        return v
-
-    @field_validator("scorers", mode="before")
-    def validate_scorers(cls, v):
-        if not v:
-            raise ValueError("Scorers cannot be empty.")
-        if not all(
-            isinstance(scorer, BaseScorer) or isinstance(scorer, APIScorerConfig)
-            for scorer in v
-        ):
-            raise ValueError(
-                "All scorers must be of type BaseScorer or APIScorerConfig."
-            )
-        return v
-
-    @field_validator("model")
-    def validate_model(cls, v, values):
-        if not v:
-            raise ValueError("Model cannot be empty.")
-
-        # Check if model is string or list of strings
-        if isinstance(v, str):
-            if v not in ACCEPTABLE_MODELS:
-                raise ValueError(
-                    f"Model name {v} not recognized. Please select a valid model name.)"
-                )
-            return v
-
-    class Config:
-        arbitrary_types_allowed = True