From 11b6bcd94997f36fea32fb63d6d8262294d219c2 Mon Sep 17 00:00:00 2001
From: Barrett Pyke <barrett.pyke@braintrustdata.com>
Date: Fri, 12 Jun 2026 11:41:09 -0500
Subject: [PATCH] fix(experiments): pass base_experiment_id to summarize

---
 py/src/braintrust/framework.py      |  5 ++-
 py/src/braintrust/logger.py         |  6 +++
 py/src/braintrust/test_framework.py | 58 ++++++++++++++++++++++++++++-
 3 files changed, 67 insertions(+), 2 deletions(-)

diff --git a/py/src/braintrust/framework.py b/py/src/braintrust/framework.py
index 49eee975..a56a818a 100644
--- a/py/src/braintrust/framework.py
+++ b/py/src/braintrust/framework.py
@@ -1367,7 +1367,10 @@ async def run_evaluator(
     )
 
     if experiment:
-        summary = experiment.summarize(summarize_scores=evaluator.summarize_scores)
+        summary = experiment.summarize(
+            summarize_scores=evaluator.summarize_scores,
+            comparison_experiment_id=evaluator.base_experiment_id,
+        )
     else:
         summary = build_local_summary(evaluator, results)
 
diff --git a/py/src/braintrust/logger.py b/py/src/braintrust/logger.py
index 460f9232..dea4b504 100644
--- a/py/src/braintrust/logger.py
+++ b/py/src/braintrust/logger.py
@@ -3969,6 +3969,12 @@ def summarize(
                 if base_experiment:
                     comparison_experiment_id = base_experiment.id
                     comparison_experiment_name = base_experiment.name
+            else:
+                try:
+                    comparison_experiment = state.api_conn().get_json(f"v1/experiment/{comparison_experiment_id}")
+                    comparison_experiment_name = comparison_experiment.get("name")
+                except Exception:
+                    pass
 
             try:
                 summary_items = state.api_conn().get_json(
diff --git a/py/src/braintrust/test_framework.py b/py/src/braintrust/test_framework.py
index 71168a38..4ac32c81 100644
--- a/py/src/braintrust/test_framework.py
+++ b/py/src/braintrust/test_framework.py
@@ -1,7 +1,7 @@
 import importlib.util
 import re
 import sys
-from unittest.mock import MagicMock
+from unittest.mock import MagicMock, patch
 
 import pytest
 from braintrust.logger import BraintrustState
@@ -78,6 +78,62 @@ def exact_match(input_value, output, expected):
     assert result.summary.scores["exact_match"].score == 1.0
 
 
+@pytest.mark.asyncio
+async def test_run_evaluator_forwards_base_experiment_id_to_summary(with_memory_logger, with_simulate_login):
+    def exact_match(input_value, output, expected):
+        return 1.0 if output == expected else 0.0
+
+    evaluator = Evaluator(
+        project_name="test-project",
+        eval_name="test-evaluator",
+        data=[EvalCase(input=1, expected=1)],
+        task=lambda input_value: input_value,
+        scores=[exact_match],
+        experiment_name=None,
+        metadata=None,
+        base_experiment_id="base-exp-id",
+    )
+
+    exp = init_test_exp("test-evaluator", "test-project")
+    expected_summary = MagicMock()
+    exp.summarize = MagicMock(return_value=expected_summary)
+
+    result = await run_evaluator(experiment=exp, evaluator=evaluator, position=None, filters=[])
+
+    assert result.summary is expected_summary
+    exp.summarize.assert_called_once_with(
+        summarize_scores=True,
+        comparison_experiment_id="base-exp-id",
+    )
+
+
+def test_experiment_summarize_resolves_explicit_comparison_name(with_memory_logger, with_simulate_login):
+    exp = init_test_exp("test-evaluator", "test-project")
+    mock_conn = MagicMock()
+
+    def get_json(path, args=None):
+        if path == "v1/experiment/base-exp-id":
+            return {"name": "base-exp"}
+        if path == "experiment-comparison2":
+            return {"scores": {}, "metrics": {}}
+        raise AssertionError(f"Unexpected get_json call: {path}, {args}")
+
+    mock_conn.get_json.side_effect = get_json
+
+    with patch.object(exp.state, "api_conn", return_value=mock_conn):
+        summary = exp.summarize(comparison_experiment_id="base-exp-id")
+
+    assert summary.comparison_experiment_name == "base-exp"
+    mock_conn.get_json.assert_any_call("v1/experiment/base-exp-id")
+    mock_conn.get_json.assert_any_call(
+        "experiment-comparison2",
+        args={
+            "experiment_id": "test-evaluator",
+            "base_experiment_id": "base-exp-id",
+        },
+    )
+
+
 @pytest.mark.asyncio
 @pytest.mark.skipif(not HAS_PYDANTIC, reason="pydantic not installed")
 async def test_run_evaluator_exposes_validated_parameter_values_to_hooks():