Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions eval_protocol/pytest/remote_rollout_processor.py
Original file line number Diff line number Diff line change
Expand Up @@ -122,9 +122,20 @@ async def _process_row(row: EvaluationRow) -> EvaluationRow:
status_logs.append(log)

if status_logs:
if len(status_logs) > 1:
logger.warning(
"Found %s status logs for rollout %s; expected at most 1. Using the first one: %s",
len(status_logs),
row.execution_metadata.rollout_id,
status_logs[0],
)
# Use the first log with status information
status_log = status_logs[0]
status_dict = status_log.get("status")
raw_extras = status_log.get("extras") or {}
status_extras = {
k: v for k, v in raw_extras.items() if k not in ("logger_name", "level", "timestamp")
}

logger.info(
f"Found status log for rollout {row.execution_metadata.rollout_id}: {status_log.get('message', '')}"
Expand All @@ -149,6 +160,11 @@ async def _process_row(row: EvaluationRow) -> EvaluationRow:
details=status_details,
)

if row.execution_metadata.extra:
row.execution_metadata.extra.update(status_extras)
else:
row.execution_metadata.extra = status_extras

logger.info("Stopping polling for rollout %s", row.execution_metadata.rollout_id)
break

Expand Down
13 changes: 12 additions & 1 deletion eval_protocol/pytest/tracing_utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -179,7 +179,18 @@ def update_row_with_remote_trace(
if k not in row.input_metadata.dataset_info:
row.input_metadata.dataset_info[k] = v

row.execution_metadata = remote_row.execution_metadata
preserved_extra = row.execution_metadata.extra

row.execution_metadata = remote_row.execution_metadata.model_copy(deep=True)

if preserved_extra:
if row.execution_metadata.extra:
# Merge remote and local extras; local takes precedence on conflicts
merged = row.execution_metadata.extra or {}
merged.update(preserved_extra)
row.execution_metadata.extra = merged
else:
row.execution_metadata.extra = preserved_extra
return None
else:
raise ValueError("Output data loader should return exactly one row.")
137 changes: 137 additions & 0 deletions vite-app/dist/assets/index-10cZ11iB.js

Large diffs are not rendered by default.

1 change: 1 addition & 0 deletions vite-app/dist/assets/index-10cZ11iB.js.map

Large diffs are not rendered by default.

46 changes: 0 additions & 46 deletions vite-app/dist/assets/index-CuQbfdPD.js

This file was deleted.

1 change: 0 additions & 1 deletion vite-app/dist/assets/index-CuQbfdPD.js.map

This file was deleted.

1 change: 1 addition & 0 deletions vite-app/dist/assets/index-DOD73Wyg.css

Large diffs are not rendered by default.

1 change: 0 additions & 1 deletion vite-app/dist/assets/index-iZp_HgyW.css

This file was deleted.

4 changes: 2 additions & 2 deletions vite-app/dist/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -5,8 +5,8 @@
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>EP | Log Viewer</title>
<link rel="icon" href="/assets/favicon-BkAAWQga.png" />
<script type="module" crossorigin src="/assets/index-CuQbfdPD.js"></script>
<link rel="stylesheet" crossorigin href="/assets/index-iZp_HgyW.css">
<script type="module" crossorigin src="/assets/index-10cZ11iB.js"></script>
<link rel="stylesheet" crossorigin href="/assets/index-DOD73Wyg.css">
</head>
<body>
<div id="root"></div>
Expand Down
2 changes: 1 addition & 1 deletion vite-app/src/types/eval-protocol.ts
Original file line number Diff line number Diff line change
Expand Up @@ -351,7 +351,7 @@ export const ExecutionMetadataSchema = z.object({
.nullable()
.optional()
.describe("Processing duration in seconds for an entire experiment."),
});
}).loose();

export const EvaluationRowSchema = z.object({
messages: z
Expand Down
Loading