inspect-ai 0.3.85__tar.gz → 0.3.87__tar.gz

{inspect_ai-0.3.85 → inspect_ai-0.3.87}/CHANGELOG.md

@@ -1,3 +1,15 @@
+## v0.3.87 (10 April 2025)
+
+- Eval: Fix an error when attempting to display realtime metrics for an evaluation.
+- Log Viewer: Fix an error when displaying a running log with a null metric value.
+
+## v0.3.86 (09 April 2025)
+
+- Open AI: Treat `UnprocessableEntityError` as bad request so we can include the request payload in the error message.
+- Eval Retry: Correctly restore model-specific generation config on retry.
+- Inspect View: Resolve sample attachments before including in realtime event stream.
+- Bugfix: Properly handle special characters in IDs during event database cleanup.
+
 ## v0.3.85 (08 April 2025)
 
 - Remove support for `goodfire` model provider (dependency conflicts).

{inspect_ai-0.3.85 → inspect_ai-0.3.87}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: inspect_ai
-Version: 0.3.85
+Version: 0.3.87
 Summary: Framework for large language model evaluations
 Author: UK AI Security Institute
 License: MIT License

{inspect_ai-0.3.85 → inspect_ai-0.3.87}/src/inspect_ai/_display/textual/widgets/task_detail.py

@@ -221,13 +221,13 @@ class TaskMetrics(Widget):
             self.recompute_grid()
 
     def on_mount(self) -> None:
-        self.recompute_grid(True)
+        self.call_after_refresh(self.recompute_grid)
 
-    def recompute_grid(self, force: bool = False) -> None:
-        if not self.is_mounted and not force:
+    def recompute_grid(self) -> None:
+        if not self.is_mounted:
             return
-        grid = self.query_one(f"#{self.grid_id()}")
 
+        grid = self.query_one(f"#{self.grid_id()}")
         grid.remove_children()
         for metric in self.metrics:
             # Add the value static but keep it around

{inspect_ai-0.3.85 → inspect_ai-0.3.87}/src/inspect_ai/_eval/eval.py

@@ -43,7 +43,7 @@ from inspect_ai.model import (
     GenerateConfigArgs,
     Model,
 )
-from inspect_ai.model._model import init_active_model, resolve_models
+from inspect_ai.model._model import get_model, init_active_model, resolve_models
 from inspect_ai.scorer._reducer import reducer_log_names
 from inspect_ai.solver._chain import chain
 from inspect_ai.solver._solver import Solver, SolverSpec
@@ -751,10 +751,15 @@ async def eval_retry_async(
             else None
         )
 
+        # resolve the model
+        model = get_model(
+            model=eval_log.eval.model,
+            config=eval_log.eval.model_generate_config,
+            base_url=eval_log.eval.model_base_url,
+            **eval_log.eval.model_args,
+        )
+
         # collect the rest of the params we need for the eval
-        model = eval_log.eval.model
-        model_base_url = eval_log.eval.model_base_url
-        model_args = eval_log.eval.model_args
         task_args = eval_log.eval.task_args
         tags = eval_log.eval.tags
         limit = eval_log.eval.config.limit
@@ -813,8 +818,6 @@ async def eval_retry_async(
                     id=task_id, task=task, task_args=task_args, model=None, log=eval_log
                 ),
                 model=model,
-                model_base_url=model_base_url,
-                model_args=model_args,
                 task_args=task_args,
                 sandbox=eval_log.eval.sandbox,
                 sandbox_cleanup=sandbox_cleanup,

{inspect_ai-0.3.85 → inspect_ai-0.3.87}/src/inspect_ai/_eval/task/log.py

@@ -139,6 +139,7 @@ class TaskLogger:
             tags=tags,
             solver_args=solver.args if solver else None,
             model=str(ModelName(model)),
+            model_generate_config=model.config,
             model_base_url=model.api.base_url,
             dataset=EvalDataset(
                 name=dataset.name,