chore!: moving to pydantic2 (explodinggradients#1394)

pyproject.toml

@@ -5,13 +5,14 @@ dependencies = [
     "datasets",
     "tiktoken",
     "langchain",
-    "langchain-core<0.3",
+    "langchain-core",
     "langchain-community",
     "langchain_openai",
-    "openai>1",
-    "pysbd>=0.3.4",
     "nest-asyncio",
     "appdirs",
+    "pydantic>=2",
+    "openai>1",
+    "pysbd>=0.3.4",
 ]
 dynamic = ["version", "readme"]
 

src/ragas/_analytics.py

@@ -9,7 +9,7 @@
 
 import requests
 from appdirs import user_data_dir
-from langchain_core.pydantic_v1 import BaseModel, Field
+from pydantic import BaseModel, Field
 
 from ragas.utils import get_debug_mode
 

src/ragas/cost.py

@@ -3,7 +3,7 @@
 
 from langchain_core.callbacks.base import BaseCallbackHandler
 from langchain_core.outputs import ChatGeneration, ChatResult, LLMResult
-from langchain_core.pydantic_v1 import BaseModel
+from pydantic import BaseModel
 
 from ragas.utils import get_from_dict
 
@@ -39,7 +39,9 @@ def cost(
             + self.output_tokens * cost_per_output_token
         )
 
-    def __eq__(self, other: "TokenUsage") -> bool:
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, TokenUsage):
+            return False
         return (
             self.input_tokens == other.input_tokens
             and self.output_tokens == other.output_tokens

src/ragas/llms/output_parser.py

@@ -4,7 +4,7 @@
 
 from langchain_core.exceptions import OutputParserException
 from langchain_core.output_parsers import PydanticOutputParser
-from langchain_core.pydantic_v1 import BaseModel
+from pydantic import BaseModel
 
 from ragas.llms import BaseRagasLLM
 from ragas.llms.prompt import Prompt, PromptValue

src/ragas/metrics/_answer_correctness.py

@@ -5,7 +5,7 @@
 from dataclasses import dataclass, field
 
 import numpy as np
-from langchain_core.pydantic_v1 import BaseModel
+from pydantic import BaseModel
 
 from ragas.dataset_schema import SingleTurnSample
 from ragas.llms.output_parser import RagasoutputParser, get_json_format_instructions

src/ragas/metrics/_answer_relevance.py

@@ -5,7 +5,7 @@
 from dataclasses import dataclass, field
 
 import numpy as np
-from langchain_core.pydantic_v1 import BaseModel
+from pydantic import BaseModel
 
 from ragas.dataset_schema import SingleTurnSample
 from ragas.llms.output_parser import RagasoutputParser, get_json_format_instructions

src/ragas/metrics/_context_entities_recall.py

@@ -24,7 +24,7 @@ class ContextEntitiesResponse(BaseModel):
 
 
 _output_instructions = get_json_format_instructions(
-    pydantic_object=ContextEntitiesResponse
+    pydantic_object=ContextEntitiesResponse  # type: ignore
 )
 _output_parser = RagasoutputParser(pydantic_object=ContextEntitiesResponse)
 

src/ragas/metrics/_context_precision.py

@@ -33,7 +33,7 @@ class ContextPrecisionVerifications(BaseModel):
 
 
 _verification_output_instructions = get_json_format_instructions(
-    ContextPrecisionVerification
+    ContextPrecisionVerification  # type: ignore
 )
 _output_parser = RagasoutputParser(pydantic_object=ContextPrecisionVerification)
 

src/ragas/metrics/_context_recall.py

@@ -5,7 +5,7 @@
 from dataclasses import dataclass, field
 
 import numpy as np
-from langchain_core.pydantic_v1 import BaseModel
+from pydantic import BaseModel, RootModel
 
 from ragas.dataset_schema import SingleTurnSample
 from ragas.llms.output_parser import RagasoutputParser, get_json_format_instructions
@@ -29,11 +29,11 @@ class ContextRecallClassificationAnswer(BaseModel):
     reason: str
 
 
-class ContextRecallClassificationAnswers(BaseModel):
-    __root__: t.List[ContextRecallClassificationAnswer]
+class ContextRecallClassificationAnswers(RootModel):
+    root: t.List[ContextRecallClassificationAnswer]
 
     def dicts(self) -> t.List[t.Dict]:
-        return self.dict()["__root__"]
+        return self.model_dump()
 
 
 _classification_output_instructions = get_json_format_instructions(

src/ragas/metrics/_faithfulness.py

@@ -6,7 +6,7 @@
 from dataclasses import dataclass, field
 
 import numpy as np
-from langchain_core.pydantic_v1 import BaseModel, Field
+from pydantic import BaseModel, Field, RootModel
 
 from ragas.dataset_schema import SingleTurnSample
 from ragas.llms.output_parser import RagasoutputParser, get_json_format_instructions
@@ -39,11 +39,8 @@ class Statements(BaseModel):
     simpler_statements: t.List[str] = Field(..., description="the simpler statements")
 
 
-class StatementsAnswers(BaseModel):
-    __root__: t.List[Statements]
-
-    def dicts(self) -> t.List[t.Dict]:
-        return self.dict()["__root__"]
+class StatementsAnswers(RootModel):
+    root: t.List[Statements]
 
 
 _statements_output_instructions = get_json_format_instructions(StatementsAnswers)
@@ -79,7 +76,7 @@ def dicts(self) -> t.List[t.Dict]:
                         ],
                     },
                 ]
-            ).dicts(),
+            ).model_dump(),
         }
     ],
     input_keys=["question", "answer", "sentences"],
@@ -94,11 +91,11 @@ class StatementFaithfulnessAnswer(BaseModel):
     verdict: int = Field(..., description="the verdict(0/1) of the faithfulness.")
 
 
-class StatementFaithfulnessAnswers(BaseModel):
-    __root__: t.List[StatementFaithfulnessAnswer]
+class StatementFaithfulnessAnswers(RootModel):
+    root: t.List[StatementFaithfulnessAnswer]
 
-    def dicts(self) -> t.List[t.Dict]:
-        return self.dict()["__root__"]
+    def dicts(self):
+        return self.model_dump()
 
 
 _faithfulness_output_instructions = get_json_format_instructions(
@@ -144,20 +141,20 @@ def dicts(self) -> t.List[t.Dict]:
                         "verdict": 0,
                     },
                 ]
-            ).dicts(),
+            ).model_dump(),
         },
         {
             "context": """Photosynthesis is a process used by plants, algae, and certain bacteria to convert light energy into chemical energy.""",
             "statements": ["Albert Einstein was a genius."],
-            "answer": StatementFaithfulnessAnswers.parse_obj(
+            "answer": StatementFaithfulnessAnswers.model_validate(
                 [
                     {
                         "statement": "Albert Einstein was a genius.",
                         "reason": "The context and statement are unrelated",
                         "verdict": 0,
                     }
                 ]
-            ).dicts(),
+            ).model_dump(),
         },
     ],
     input_keys=["context", "statements"],
@@ -237,9 +234,9 @@ def _create_statements_prompt(self, row: t.Dict) -> PromptValue:
     def _compute_score(self, answers: StatementFaithfulnessAnswers):
         # check the verdicts and compute the score
         faithful_statements = sum(
-            1 if answer.verdict else 0 for answer in answers.__root__
+            1 if answer.verdict else 0 for answer in answers.model_dump()
         )
-        num_statements = len(answers.__root__)
+        num_statements = len(answers.model_dump())
         if num_statements:
             score = faithful_statements / num_statements
         else:
@@ -272,7 +269,7 @@ async def _ascore(self: t.Self, row: t.Dict, callbacks: Callbacks) -> float:
         if statements is None:
             return np.nan
 
-        statements = [item["simpler_statements"] for item in statements.dicts()]
+        statements = [item["simpler_statements"] for item in statements.model_dump()]
         statements = [item for sublist in statements for item in sublist]
 
         assert isinstance(statements, t.List), "statements must be a list"
@@ -295,7 +292,7 @@ async def _ascore(self: t.Self, row: t.Dict, callbacks: Callbacks) -> float:
         ]
 
         faithfulness_list = [
-            faith.dicts() for faith in faithfulness_list if faith is not None
+            faith.model_dump() for faith in faithfulness_list if faith is not None
         ]
 
         if faithfulness_list:
@@ -385,7 +382,7 @@ async def _ascore(self: t.Self, row: t.Dict, callbacks: Callbacks) -> float:
         if statements is None:
             return np.nan
 
-        statements = [item["simpler_statements"] for item in statements.dicts()]
+        statements = [item["simpler_statements"] for item in statements.model_dump()]
         statements = [item for sublist in statements for item in sublist]
 
         assert isinstance(statements, t.List), "statements must be a list"

src/ragas/metrics/_noise_sensitivity.py

@@ -137,13 +137,12 @@ async def _evaluate_statement_faithfulness(
                 "verdict",
             )
 
-            faithfulness_list = StatementFaithfulnessAnswers.parse_obj(
+            faithfulness_list = StatementFaithfulnessAnswers.model_validate(
                 faithfulness_list
             )
 
             verdict_list = [
-                1 if statement.verdict else 0
-                for statement in faithfulness_list.__root__
+                1 if statement.verdict else 0 for statement in faithfulness_list.dicts()
             ]
             return np.array(verdict_list)
         else:
@@ -162,14 +161,17 @@ async def _decompose_answer_into_statements(
                 callbacks=callbacks,
             )
         else:
-            statements_gen = self.llm.generate(
+            statements_gen = await self.llm.generate(
                 p_value,
                 callbacks=callbacks,
             )
 
         # Await the aparse method
         statements = await _statements_output_parser.aparse(
-            statements_gen.generations[0][0].text, p_value, self.llm, self.max_retries  # type: ignore
+            statements_gen.generations[0][0].text,
+            p_value,
+            self.llm,
+            self.max_retries,  # type: ignore
         )
 
         if statements is None:

src/ragas/metrics/_summarization.py

@@ -31,13 +31,13 @@ class GenerateAnswersResponse(BaseModel):
 
 
 _output_instructions_question_generation = get_json_format_instructions(
-    pydantic_object=GenerateQuestionsResponse
+    pydantic_object=GenerateQuestionsResponse  # type: ignore
 )
 _output_instructions_answer_generation = get_json_format_instructions(
-    pydantic_object=GenerateAnswersResponse
+    pydantic_object=GenerateAnswersResponse  # type: ignore
 )
 _output_instructions_keyphrase_extraction = get_json_format_instructions(
-    pydantic_object=ExtractKeyphrasesResponse
+    pydantic_object=ExtractKeyphrasesResponse  # type: ignore
 )
 _output_parser_question_generation = RagasoutputParser(
     pydantic_object=GenerateQuestionsResponse

src/ragas/testset/prompts.py

@@ -1,3 +1,4 @@
+# type: ignore
 from langchain_core.pydantic_v1 import BaseModel
 
 from ragas.llms.output_parser import RagasoutputParser, get_json_format_instructions