From b43a514f960e1fe58ad17ef2fbc63b32baef0f3d Mon Sep 17 00:00:00 2001
From: Yang Yang <yyy@meta.com>
Date: Fri, 14 Mar 2025 11:34:09 -0700
Subject: [PATCH] fix the split of docVQA dataset

---
 llama_stack/templates/open-benchmark/open_benchmark.py | 2 +-
 llama_stack/templates/open-benchmark/run.yaml          | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/llama_stack/templates/open-benchmark/open_benchmark.py b/llama_stack/templates/open-benchmark/open_benchmark.py
index bbcda106e..6dbc65495 100644
--- a/llama_stack/templates/open-benchmark/open_benchmark.py
+++ b/llama_stack/templates/open-benchmark/open_benchmark.py
@@ -209,7 +209,7 @@ def get_distribution_template() -> DistributionTemplate:
             url=URL(uri="https://huggingface.co/datasets/llamastack/docvqa"),
             metadata={
                 "path": "llamastack/docvqa",
-                "split": "train",
+                "split": "val",
             },
             dataset_schema={
                 "chat_completion_input": {"type": "string"},
diff --git a/llama_stack/templates/open-benchmark/run.yaml b/llama_stack/templates/open-benchmark/run.yaml
index 3cf3af45d..29c2ffb4f 100644
--- a/llama_stack/templates/open-benchmark/run.yaml
+++ b/llama_stack/templates/open-benchmark/run.yaml
@@ -200,7 +200,7 @@ datasets:
     uri: https://huggingface.co/datasets/llamastack/docvqa
   metadata:
     path: llamastack/docvqa
-    split: train
+    split: val
   dataset_id: docvqa
   provider_id: huggingface
 >>>>>>> 81df5a4e (add DocVQA scoring)