From 4f4d4fddc2a1a75141c7d06d968fc7e9c308b937 Mon Sep 17 00:00:00 2001
From: Tim <bleimehl@gmail.com>
Date: Tue, 18 May 2021 16:01:52 +0200
Subject: [PATCH] cache cord19 data

---
 pipeline.yaml | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/pipeline.yaml b/pipeline.yaml
index 4507317..cd17044 100644
--- a/pipeline.yaml
+++ b/pipeline.yaml
@@ -13,7 +13,8 @@ CovidGraphPipeline:
     image_repo: covidgraph/data-cord19
     env_vars:
       CONFIGS_PAPER_BATCH_SIZE: 300
-      CONFIGS_NO_OF_PROCESSES: 25
+      CONFIGS_NO_OF_PROCESSES: 12
+    volumes: { "/tmp/data/cord19": { "bind": "/app/dataset", "mode": "rw" } }
   - name: TEXT_FRAGGER
     image_repo: covidgraph/graph-processing_fragmentize_text
     dependencies: