From 9ec42aa245ce70203eab60f94710edfc5d2cd82b Mon Sep 17 00:00:00 2001
From: Ivan Zhou <ivanz@stanford.edu>
Date: Mon, 29 Jan 2024 00:43:04 +0000
Subject: [PATCH] 0 weight on large datasets

---
 config/data/rpv1_llama.yaml | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/config/data/rpv1_llama.yaml b/config/data/rpv1_llama.yaml
index 43922ad43..74019ce7d 100644
--- a/config/data/rpv1_llama.yaml
+++ b/config/data/rpv1_llama.yaml
@@ -233,11 +233,11 @@ configs:
     train_urls:
       - gs://levanter-data/dev/redpajama/wikipedia/wiki.jsonl
 train_weights:
-  arxiv: 28
-  book: 26
-  c4: 175
-  common_crawl: 878
-  github: 59
+  arxiv: 0  # 28
+  book: 0 # 26
+  c4: 0 # 175
+  common_crawl: 0 # 878
+  github: 0 # 59
   StackExchange: 20
   wikipedia: 24
 stop_strategy: all_exhausted
\ No newline at end of file