opik 1.9.54__tar.gz → 1.9.74__tar.gz

{opik-1.9.54/src/opik.egg-info → opik-1.9.74}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: opik
-Version: 1.9.54
+Version: 1.9.74
 Summary: Comet tool for logging and evaluating LLM traces
 Home-page: https://www.comet.com
 Author: Comet ML Inc.

{opik-1.9.54 → opik-1.9.74}/src/opik/api_objects/helpers.py

@@ -13,7 +13,7 @@ from ..rest_api.types import (
     trace_filter_public,
     trace_thread_filter,
 )
-from ..types import FeedbackScoreDict
+from ..types import BatchFeedbackScoreDict
 
 LOGGER = logging.getLogger(__name__)
 
@@ -139,7 +139,7 @@ def parse_search_expressions(
 
 
 def parse_feedback_score_messages(
-    scores: List[FeedbackScoreDict],
+    scores: List[BatchFeedbackScoreDict],
     project_name: str,
     parsed_item_class: Type[ScoreMessageT],
     logger: logging.Logger,
@@ -155,9 +155,13 @@ def parse_feedback_score_messages(
 
     score_messages = [
         parsed_item_class(
+            id=score_dict["id"],
+            name=score_dict["name"],
+            value=score_dict["value"],
             source=constants.FEEDBACK_SCORE_SOURCE_SDK,
-            project_name=project_name,
-            **score_dict,
+            project_name=score_dict.get("project_name") or project_name,
+            reason=score_dict.get("reason"),
+            category_name=score_dict.get("category_name"),
         )
         for score_dict in valid_scores
     ]

{opik-1.9.54 → opik-1.9.74}/src/opik/api_objects/opik_client.py

@@ -2,7 +2,7 @@ import atexit
 import datetime
 import functools
 import logging
-from typing import Any, Dict, List, Optional, TypeVar, Union, Literal
+from typing import Any, Dict, List, Optional, TypeVar, Union, Literal, cast
 
 import httpx
 
@@ -55,7 +55,13 @@ from ..rest_api.types import (
     span_filter_public,
     trace_filter_public,
 )
-from ..types import ErrorInfoDict, FeedbackScoreDict, LLMProvider, SpanType
+from ..types import (
+    BatchFeedbackScoreDict,
+    ErrorInfoDict,
+    FeedbackScoreDict,
+    LLMProvider,
+    SpanType,
+)
 
 LOGGER = logging.getLogger(__name__)
 
@@ -285,7 +291,9 @@ class Opik:
             for feedback_score in feedback_scores:
                 feedback_score["id"] = id
 
-            self.log_traces_feedback_scores(feedback_scores, project_name)
+            self.log_traces_feedback_scores(
+                cast(List[BatchFeedbackScoreDict], feedback_scores), project_name
+            )
 
         if attachments is not None:
             for attachment_data in attachments:
@@ -460,7 +468,9 @@ class Opik:
             for feedback_score in feedback_scores:
                 feedback_score["id"] = id
 
-            self.log_spans_feedback_scores(feedback_scores, project_name)
+            self.log_spans_feedback_scores(
+                cast(List[BatchFeedbackScoreDict], feedback_scores), project_name
+            )
 
         return span.span_client.create_span(
             trace_id=trace_id,
@@ -629,23 +639,34 @@ class Opik:
         )
 
     def log_spans_feedback_scores(
-        self, scores: List[FeedbackScoreDict], project_name: Optional[str] = None
+        self, scores: List[BatchFeedbackScoreDict], project_name: Optional[str] = None
     ) -> None:
         """
         Log feedback scores for spans.
 
         Args:
-            scores (List[FeedbackScoreDict]): A list of feedback score dictionaries.
+            scores (List[BatchFeedbackScoreDict]): A list of feedback score dictionaries.
                 Specifying a span id via `id` key for each score is mandatory.
             project_name: The name of the project in which the spans are logged. If not set, the project name
                 which was configured when the Opik instance was created will be used.
+                Deprecated: use `project_name` in the feedback score dictionary that's listed in the `scores` parameter.
 
         Returns:
             None
+
+        Example:
+            >>> from opik import Opik
+            >>> client = Opik()
+            >>> # Batch logging across multiple projects
+            >>> scores = [
+            >>>     {"id": span1_id, "name": "accuracy", "value": 0.95, "project_name": "project-A"},
+            >>>     {"id": span2_id, "name": "accuracy", "value": 0.88, "project_name": "project-B"},
+            >>> ]
+            >>> client.log_spans_feedback_scores(scores=scores)
         """
         score_messages = helpers.parse_feedback_score_messages(
             scores=scores,
-            project_name=project_name or self._project_name,
+            project_name=project_name or self.project_name,
             parsed_item_class=messages.FeedbackScoreMessage,
             logger=LOGGER,
         )
@@ -667,23 +688,34 @@ class Opik:
             self._streamer.put(add_span_feedback_scores_batch_message)
 
     def log_traces_feedback_scores(
-        self, scores: List[FeedbackScoreDict], project_name: Optional[str] = None
+        self, scores: List[BatchFeedbackScoreDict], project_name: Optional[str] = None
     ) -> None:
         """
         Log feedback scores for traces.
 
         Args:
-            scores (List[FeedbackScoreDict]): A list of feedback score dictionaries.
+            scores (List[BatchFeedbackScoreDict]): A list of feedback score dictionaries.
                 Specifying a trace id via `id` key for each score is mandatory.
             project_name: The name of the project in which the traces are logged. If not set, the project name
                 which was configured when the Opik instance was created will be used.
+                Deprecated: use `project_name` in the feedback score dictionary that's listed in the `scores` parameter.
 
         Returns:
             None
+
+        Example:
+            >>> from opik import Opik
+            >>> client = Opik()
+            >>> # Batch logging across multiple projects
+            >>> scores = [
+            >>>     {"id": trace1_id, "name": "accuracy", "value": 0.95, "project_name": "project-A"},
+            >>>     {"id": trace2_id, "name": "accuracy", "value": 0.88, "project_name": "project-B"},
+            >>> ]
+            >>> client.log_traces_feedback_scores(scores=scores)
         """
         score_messages = helpers.parse_feedback_score_messages(
             scores=scores,
-            project_name=project_name or self._project_name,
+            project_name=project_name or self.project_name,
             parsed_item_class=messages.FeedbackScoreMessage,
             logger=LOGGER,
         )
@@ -706,16 +738,17 @@ class Opik:
             self._streamer.put(add_trace_feedback_scores_batch_message)
 
     def log_threads_feedback_scores(
-        self, scores: List[FeedbackScoreDict], project_name: Optional[str] = None
+        self, scores: List[BatchFeedbackScoreDict], project_name: Optional[str] = None
     ) -> None:
         """
         Log feedback scores for threads.
 
         Args:
-            scores (List[FeedbackScoreDict]): A list of feedback score dictionaries.
+            scores (List[BatchFeedbackScoreDict]): A list of feedback score dictionaries.
                 Specifying a thread id via `id` key for each score is mandatory.
             project_name: The name of the project in which the threads are logged. If not set, the project name
                 which was configured when the Opik instance was created will be used.
+                Deprecated: use `project_name` in the feedback score dictionary that's listed in the `scores` parameter.
 
         Returns:
             None
@@ -723,13 +756,10 @@ class Opik:
         Example:
             >>> from opik import Opik
             >>> client = Opik()
+            >>> # Batch logging across multiple projects
             >>> scores = [
-            >>>     {
-            >>>         "id": "thread_123",
-            >>>         "name": "user_satisfaction",
-            >>>         "value": 0.85,
-            >>>         "reason": "User seemed satisfied with the conversation"
-            >>>     }
+            >>>     {"id": "thread_123", "name": "user_satisfaction", "value": 0.85, "project_name": "project-A"},
+            >>>     {"id": "thread_456", "name": "user_satisfaction", "value": 0.92, "project_name": "project-B"},
             >>> ]
             >>> client.log_threads_feedback_scores(scores=scores)
         """
@@ -1145,7 +1175,7 @@ class Opik:
                 - `start_time`, `end_time`: =, >, <, >=, <=
                 - `input`, `output`: =, contains, not_contains
                 - `metadata`: =, contains, >, <
-                - `feedback_scores`: =, >, <, >=, <=
+                - `feedback_scores`: =, >, <, >=, <=, is_empty, is_not_empty
                 - `tags`: contains (only)
                 - `usage.total_tokens`, `usage.prompt_tokens`, `usage.completion_tokens`, `duration`, `number_of_messages`, `total_estimated_cost`: =, !=, >, <, >=, <=
 
@@ -1155,6 +1185,8 @@ class Opik:
                 - `input contains "question"` - Filter by input content
                 - `usage.total_tokens > 1000` - Filter by token usage
                 - `feedback_scores.accuracy > 0.8` - Filter by feedback score
+                - `feedback_scores.my_metric is_empty` - Filter traces with empty feedback score
+                - `feedback_scores.my_metric is_not_empty` - Filter traces with non-empty feedback score
                 - `tags contains "production"` - Filter by tag
                 - `metadata.model = "gpt-4"` - Filter by metadata field
                 - `thread_id = "thread_123"` - Filter by thread ID
@@ -1237,7 +1269,7 @@ class Opik:
                 - `start_time`, `end_time`: =, >, <, >=, <=
                 - `input`, `output`: =, contains, not_contains
                 - `metadata`: =, contains, >, <
-                - `feedback_scores`: =, >, <, >=, <=
+                - `feedback_scores`: =, >, <, >=, <=, is_empty, is_not_empty
                 - `tags`: contains (only)
                 - `usage.total_tokens`, `usage.prompt_tokens`, `usage.completion_tokens`, `duration`, `number_of_messages`, `total_estimated_cost`: =, !=, >, <, >=, <=
 
@@ -1247,6 +1279,8 @@ class Opik:
                 - `input contains "question"` - Filter by input content
                 - `usage.total_tokens > 1000` - Filter by token usage
                 - `feedback_scores.accuracy > 0.8` - Filter by feedback score
+                - `feedback_scores.my_metric is_empty` - Filter spans with empty feedback score
+                - `feedback_scores.my_metric is_not_empty` - Filter spans with non-empty feedback score
                 - `tags contains "production"` - Filter by tag
                 - `metadata.model = "gpt-4"` - Filter by metadata field
                 - `thread_id = "thread_123"` - Filter by thread ID

{opik-1.9.54 → opik-1.9.74}/src/opik/api_objects/opik_query_language.py

@@ -54,7 +54,7 @@ SUPPORTED_OPERATORS = {
     ],
     "output": ["=", "contains", "not_contains"],
     "metadata": ["=", "contains", ">", "<"],
-    "feedback_scores": ["=", ">", "<", ">=", "<="],
+    "feedback_scores": ["=", ">", "<", ">=", "<=", "is_empty", "is_not_empty"],
     "tags": ["contains"],
     "usage.total_tokens": ["=", "!=", ">", "<", ">=", "<="],
     "usage.prompt_tokens": ["=", "!=", ">", "<", ">=", "<="],
@@ -132,6 +132,8 @@ SUPPORTED_OPERATORS = {
     ],
 }
 
+OPERATORS_WITHOUT_VALUES = {"is_empty", "is_not_empty"}
+
 
 class OpikQueryLanguage:
     """
@@ -384,8 +386,12 @@ class OpikQueryLanguage:
             # Parse operators
             parsed_operator = self._parse_operator(parsed_field["field"])
 
-            # Parse values
-            parsed_value = self._parse_value()
+            operator_name = parsed_operator.get("operator", "")
+            if operator_name in OPERATORS_WITHOUT_VALUES:
+                # For operators without values, use empty string as value
+                parsed_value = {"value": ""}
+            else:
+                parsed_value = self._parse_value()
 
             expressions.append({**parsed_field, **parsed_operator, **parsed_value})
 

{opik-1.9.54 → opik-1.9.74}/src/opik/api_objects/prompt/chat/chat_prompt.py

@@ -1,9 +1,11 @@
 import copy
 import json
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Tuple, Type
+
 from typing_extensions import override
 
 from opik.rest_api import types as rest_api_types
+from opik.validation import chat_prompt_messages, validator
 from . import chat_prompt_template
 from .. import client as prompt_client
 from .. import types as prompt_types
@@ -16,6 +18,10 @@ class ChatPrompt(base_prompt.BasePrompt):
     Similar to Prompt but uses a list of chat messages instead of a string template.
     """
 
+    _parameter_validators: List[Tuple[str, Type[validator.RaisableValidator]]] = [
+        ("messages", chat_prompt_messages.ChatPromptMessagesValidator),
+    ]
+
     def __init__(
         self,
         name: str,
@@ -37,8 +43,12 @@ class ChatPrompt(base_prompt.BasePrompt):
 
         Raises:
             PromptTemplateStructureMismatch: If a text prompt with the same name already exists (template structure is immutable).
+            ValidationError: If messages structure is invalid.
         """
 
+        # Validate messages structure
+        self._validate_inputs(messages=messages)
+
         self._chat_template = chat_prompt_template.ChatPromptTemplate(
             messages=messages,
             template_type=type,
@@ -54,6 +64,13 @@ class ChatPrompt(base_prompt.BasePrompt):
 
         self._sync_with_backend()
 
+    def _validate_inputs(self, **kwargs: Any) -> None:
+        for parameter, validator_class in self._parameter_validators:
+            if parameter in kwargs:
+                validator_instance = validator_class(kwargs[parameter])
+                validator_instance.validate()
+                validator_instance.raise_if_validation_failed()
+
     def _sync_with_backend(self) -> None:
         from opik.api_objects import opik_client
 

{opik-1.9.54 → opik-1.9.74}/src/opik/api_objects/threads/threads_client.py

@@ -3,7 +3,7 @@ from typing import List, Optional
 
 import opik
 from opik.rest_api import TraceThread
-from opik.types import FeedbackScoreDict
+from opik.types import BatchFeedbackScoreDict
 
 from .. import helpers, rest_stream_parser, constants
 from ... import config
@@ -74,7 +74,7 @@ class ThreadsClient:
                 - `start_time`, `end_time`: =, >, <, >=, <=
                 - `input`, `output`: =, contains, not_contains
                 - `metadata`: =, contains, >, <
-                - `feedback_scores`: =, >, <, >=, <=
+                - `feedback_scores`: =, >, <, >=, <=, is_empty, is_not_empty
                 - `tags`: contains (only)
                 - `usage.total_tokens`, `usage.prompt_tokens`, `usage.completion_tokens`, `duration`, `number_of_messages`, `total_estimated_cost`: =, !=, >, <, >=, <=
 
@@ -84,6 +84,8 @@ class ThreadsClient:
                 - `duration > 300` - Filter by thread duration (seconds)
                 - `number_of_messages >= 5` - Filter by message count
                 - `feedback_scores.user_frustration > 0.5` - Filter by feedback score
+                - `feedback_scores.my_metric is_empty` - Filter threads with empty feedback score
+                - `feedback_scores.my_metric is_not_empty` - Filter threads with non-empty feedback score
                 - `tags contains "important"` - Filter by tag
 
                 If not provided, all threads in the project will be returned up to the limit.
@@ -127,7 +129,7 @@ class ThreadsClient:
         return threads
 
     def log_threads_feedback_scores(
-        self, scores: List[FeedbackScoreDict], project_name: Optional[str] = None
+        self, scores: List[BatchFeedbackScoreDict], project_name: Optional[str] = None
     ) -> None:
         """
         Logs feedback scores for threads in a specific project. This method processes the given
@@ -138,7 +140,8 @@ class ThreadsClient:
             scores: A list of dictionaries containing feedback scores
                 for threads to be logged. Specifying a thread id via `id` key for each score is mandatory.
             project_name: The name of the project to associate with the logged
-                scores. If not provided, the scores won't be associated with any specific project.
+                scores. If not provided, the project name configured in the Opik client will be used.
+                This parameter is used as a fallback if `project_name` is not specified in the score dictionary.
         """
         project_name = project_name or self._opik_client.project_name
 

{opik-1.9.54 → opik-1.9.74}/src/opik/api_objects/validation_helpers.py

@@ -1,7 +1,7 @@
 import logging
 from typing import Any, Optional, cast, Union, Dict
 
-from ..types import FeedbackScoreDict
+from ..types import BatchFeedbackScoreDict
 from ..validation import feedback_score as feedback_score_validator
 from .. import logging_messages, llm_usage
 from opik.types import LLMProvider
@@ -38,7 +38,7 @@ def validate_and_parse_usage(
 
 def validate_feedback_score(
     feedback_score: Any, logger: logging.Logger
-) -> Optional[FeedbackScoreDict]:
+) -> Optional[BatchFeedbackScoreDict]:
     feedback_score_validator_ = feedback_score_validator.FeedbackScoreValidator(
         feedback_score
     )
@@ -51,4 +51,4 @@ def validate_feedback_score(
         )
         return None
 
-    return cast(FeedbackScoreDict, feedback_score)
+    return cast(BatchFeedbackScoreDict, feedback_score)

{opik-1.9.54 → opik-1.9.74}/src/opik/decorator/arguments_helpers.py

@@ -52,6 +52,7 @@ class StartSpanParameters(BaseArguments):
     project_name: Optional[str] = None
     model: Optional[str] = None
     provider: Optional[str] = None
+    thread_id: Optional[str] = None  # used for traces only
 
 
 @dataclasses.dataclass

{opik-1.9.54 → opik-1.9.74}/src/opik/decorator/span_creation_handler.py

@@ -113,6 +113,7 @@ def create_span_respecting_context(
             metadata=start_span_arguments.metadata,
             tags=start_span_arguments.tags,
             project_name=start_span_arguments.project_name,
+            thread_id=start_span_arguments.thread_id,
         )
 
         current_span_data = arguments_helpers.create_span_data(

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/conversation/llm_judges/conversational_coherence/metric.py

@@ -93,7 +93,9 @@ class ConversationalCoherenceMetric(ConversationThreadMetric):
         if isinstance(model, base_model.OpikBaseModel):
             self._model = model
         else:
-            self._model = models_factory.get(model_name=model, temperature=temperature)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, temperature=temperature
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/conversation/llm_judges/session_completeness/metric.py

@@ -80,7 +80,9 @@ class SessionCompletenessQuality(ConversationThreadMetric):
         if isinstance(model, base_model.OpikBaseModel):
             self._model = model
         else:
-            self._model = models_factory.get(model_name=model, temperature=temperature)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, temperature=temperature
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/conversation/llm_judges/user_frustration/metric.py

@@ -92,7 +92,9 @@ class UserFrustrationMetric(ConversationThreadMetric):
         if isinstance(model, base_model.OpikBaseModel):
             self._model = model
         else:
-            self._model = models_factory.get(model_name=model, temperature=temperature)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, temperature=temperature
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/heuristics/equals.py

@@ -42,22 +42,26 @@ class Equals(base_metric.BaseMetric):
         self._case_sensitive = case_sensitive
 
     def score(
-        self, output: str, reference: str, **ignored_kwargs: Any
+        self, output: Any, reference: Any, **ignored_kwargs: Any
     ) -> score_result.ScoreResult:
         """
-        Calculate the score based on whether the output string exactly matches the expected output.
+        Calculate the score based on whether the output exactly matches the expected output.
 
         Args:
-            output: The output string to check.
-            reference: The expected output string to compare against.
+            output: The output to check. Will be converted to string for comparison.
+            reference: The expected output to compare against. Will be converted to string for comparison.
             **ignored_kwargs: Additional keyword arguments that are ignored.
 
         Returns:
-            score_result.ScoreResult: A ScoreResult object with a value of 1.0 if the strings match,
+            score_result.ScoreResult: A ScoreResult object with a value of 1.0 if the values match,
                 0.0 otherwise.
         """
-        value_left = output if self._case_sensitive else output.lower()
-        value_right = reference if self._case_sensitive else reference.lower()
+        # Convert to string to handle numeric and other types
+        output_str = str(output)
+        reference_str = str(reference)
+
+        value_left = output_str if self._case_sensitive else output_str.lower()
+        value_right = reference_str if self._case_sensitive else reference_str.lower()
 
         if value_left == value_right:
             return score_result.ScoreResult(value=1.0, name=self.name)

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/answer_relevance/metric.py

@@ -88,7 +88,9 @@ class AnswerRelevance(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def _init_few_shot_examples(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/context_precision/metric.py

@@ -76,7 +76,9 @@ class ContextPrecision(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/context_recall/metric.py

@@ -74,7 +74,9 @@ class ContextRecall(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/factuality/metric.py

@@ -63,7 +63,7 @@ class Factuality(base_metric.BaseMetric):
         if isinstance(model, base_model.OpikBaseModel):
             self._model = model
         else:
-            self._model = models_factory.get(model_name=model)
+            self._model = models_factory.get(model_name=model, track=self.track)
 
     def score(
         self, input: str, output: str, context: List[str], **ignored_kwargs: Any

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/g_eval/metric.py

@@ -127,7 +127,9 @@ class GEval(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
         if (
             hasattr(self._model, "supported_params")

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/hallucination/metric.py

@@ -73,7 +73,9 @@ class Hallucination(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/moderation/metric.py

@@ -70,7 +70,9 @@ class Moderation(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(self, output: str, **ignored_kwargs: Any) -> score_result.ScoreResult:
         """

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/structure_output_compliance/metric.py

@@ -69,7 +69,9 @@ class StructuredOutputCompliance(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/syc_eval/metric.py

@@ -93,7 +93,7 @@ class SycEval(base_metric.BaseMetric):
         if isinstance(model, base_model.OpikBaseModel):
             self._model = model
         else:
-            self._model = models_factory.get(model_name=model)
+            self._model = models_factory.get(model_name=model, track=self.track)
 
     def _init_rebuttal_model(
         self, rebuttal_model: Optional[Union[str, base_model.OpikBaseModel]]
@@ -101,7 +101,9 @@ class SycEval(base_metric.BaseMetric):
         if isinstance(rebuttal_model, base_model.OpikBaseModel):
             self._rebuttal_model = rebuttal_model
         else:
-            self._rebuttal_model = models_factory.get(model_name=rebuttal_model)
+            self._rebuttal_model = models_factory.get(
+                model_name=rebuttal_model, track=self.track
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/trajectory_accuracy/metric.py

@@ -84,7 +84,9 @@ class TrajectoryAccuracy(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(
         self,

{opik-1.9.54 → opik-1.9.74}/src/opik/evaluation/metrics/llm_judges/usefulness/metric.py

@@ -68,7 +68,9 @@ class Usefulness(base_metric.BaseMetric):
             if self._seed is not None:
                 model_kwargs["seed"] = self._seed
 
-            self._model = models_factory.get(model_name=model, **model_kwargs)
+            self._model = models_factory.get(
+                model_name=model, track=self.track, **model_kwargs
+            )
 
     def score(
         self, input: str, output: str, **ignored_kwargs: Any