From 3f0cf97ecd021fe8e09b8f78f35ac61ccb547a5f Mon Sep 17 00:00:00 2001
From: Nehal Patel <1161604+habemus-papadum@users.noreply.github.com>
Date: Wed, 5 Jul 2023 09:59:29 -0400
Subject: [PATCH 1/2] Move punkt download to models.py

This allows for `punkt` to be cached in the gerev docker images along with the HF models
---
 app/models.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/app/models.py b/app/models.py
index bf27efc4..b426cb2d 100644
--- a/app/models.py
+++ b/app/models.py
@@ -1,7 +1,7 @@
 from sentence_transformers import SentenceTransformer, CrossEncoder
 from transformers import pipeline
 import torch
-
+import nltk
 
 bi_encoder = SentenceTransformer('multi-qa-MiniLM-L6-cos-v1')
 
@@ -9,3 +9,5 @@
 cross_encoder_large = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
 
 qa_model = pipeline('question-answering', model='deepset/roberta-base-squad2')
+
+nltk.download('punkt')

From dc3c1358ba67e91d945e5c5ee14509960ab07671 Mon Sep 17 00:00:00 2001
From: Nehal Patel <1161604+habemus-papadum@users.noreply.github.com>
Date: Wed, 5 Jul 2023 10:02:02 -0400
Subject: [PATCH 2/2] Location of `punkt` download has changed

---
 app/search_logic.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/app/search_logic.py b/app/search_logic.py
index 34673eaf..b1595c34 100644
--- a/app/search_logic.py
+++ b/app/search_logic.py
@@ -25,7 +25,6 @@
 BI_ENCODER_CANDIDATES = 60 if torch.cuda.is_available() else 20
 SMALL_CROSS_ENCODER_CANDIDATES = 30 if torch.cuda.is_available() else 10
 
-nltk.download('punkt')
 logger = logging.getLogger(__name__)