[GuideLLM Refactor] Complete CSV output

src/guidellm/benchmark/output.py

@@ -567,6 +567,13 @@ async def finalize(self, report: GenerativeBenchmarksReport) -> Path:
                 benchmark_headers: list[str] = []
                 benchmark_values: list[str | float | list[float]] = []
 
+                # Add basic run description info
+                desc_headers, desc_values = (
+                    self._get_benchmark_desc_headers_and_values(benchmark)
+                )
+                benchmark_headers.extend(desc_headers)
+                benchmark_values.extend(desc_values)
+
                 # Add status-based metrics
                 for status in StatusDistributionSummary.model_fields:
                     status_headers, status_values = (
@@ -672,6 +679,21 @@ def _get_benchmark_status_metrics_stats(
         ]
         return headers, values
 
+    def _get_benchmark_extras_headers_and_values(
+        self, benchmark: GenerativeBenchmark,
+    ) -> tuple[list[str], list[str]]:
+        headers = ["Profile", "Backend", "Generator Data"]
+        values: list[str] = [
+            benchmark.benchmarker.profile.model_dump_json(),
+            json.dumps(benchmark.benchmarker.backend),
+            json.dumps(benchmark.benchmarker.requests["attributes"]["data"]),
+        ]
+
+        if len(headers) != len(values):
+            raise ValueError("Headers and values length mismatch.")
+
+        return headers, values
+
 
 @GenerativeBenchmarkerOutput.register("html")
 class GenerativeBenchmarkerHTML(GenerativeBenchmarkerOutput):

tests/unit/benchmark/test_output.py

@@ -80,7 +80,6 @@ def test_file_yaml():
 
     mock_path.unlink()
 
-@pytest.mark.skip(reason="CSV fix not merged yet")
 @pytest.mark.asyncio
 async def test_file_csv():
     mock_benchmark = mock_generative_benchmark()
@@ -96,6 +95,7 @@ async def test_file_csv():
         rows = list(reader)
 
     assert "Type" in headers
+    assert "Profile" in headers
     assert len(rows) == 1
 
     mock_path.unlink()

tests/unit/mock_benchmark.py

@@ -76,7 +76,11 @@ def mock_generative_benchmark() -> GenerativeBenchmark:
         ),
         benchmarker=BenchmarkerDict(
             profile=SynchronousProfile.create("synchronous", rate=None),
-            requests={},
+            requests={
+                "attributes": {
+                    "data": "prompt_tokens=256,output_tokens=128",
+                },
+            },
             backend={},
             environment={},
             aggregators={},