From 2ac5b84e26b412d00985b200d3a8a35becb86532 Mon Sep 17 00:00:00 2001
From: Amnon Catav <catav.amnon1@gmail.com>
Date: Mon, 30 Oct 2023 13:45:11 +0200
Subject: [PATCH] change defaults

---
 config/config.yaml                                 | 2 +-
 src/canopy/knowledge_base/knowledge_base.py        | 2 +-
 src/canopy/knowledge_base/record_encoder/openai.py | 2 +-
 src/canopy_server/api_models.py                    | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/config/config.yaml b/config/config.yaml
index 58f63211..dba56a62 100644
--- a/config/config.yaml
+++ b/config/config.yaml
@@ -46,7 +46,7 @@ chat_engine:
         type: OpenAIRecordEncoder
         params:
           model_name: text-embedding-ada-002
-          batch_size: 100
+          batch_size: 400
 
       chunker:
         type: MarkdownChunker
diff --git a/src/canopy/knowledge_base/knowledge_base.py b/src/canopy/knowledge_base/knowledge_base.py
index 85687685..1c7e855c 100644
--- a/src/canopy/knowledge_base/knowledge_base.py
+++ b/src/canopy/knowledge_base/knowledge_base.py
@@ -490,7 +490,7 @@ def _query_index(self,
     def upsert(self,
                documents: List[Document],
                namespace: str = "",
-               batch_size: int = 100,
+               batch_size: int = 200,
                show_progress_bar: bool = False):
         """
         Upsert documents into the knowledge base.
diff --git a/src/canopy/knowledge_base/record_encoder/openai.py b/src/canopy/knowledge_base/record_encoder/openai.py
index ce18a3f8..9d4cf578 100644
--- a/src/canopy/knowledge_base/record_encoder/openai.py
+++ b/src/canopy/knowledge_base/record_encoder/openai.py
@@ -17,7 +17,7 @@ class OpenAIRecordEncoder(DenseRecordEncoder):
     def __init__(self,
                  *,
                  model_name: str = "text-embedding-ada-002",
-                 batch_size: int = 100,
+                 batch_size: int = 400,
                  **kwargs):
         encoder = OpenAIEncoder(model_name)
         super().__init__(dense_encoder=encoder, batch_size=batch_size, **kwargs)
diff --git a/src/canopy_server/api_models.py b/src/canopy_server/api_models.py
index 98ddb799..dae422f5 100644
--- a/src/canopy_server/api_models.py
+++ b/src/canopy_server/api_models.py
@@ -19,7 +19,7 @@ class ContextQueryRequest(BaseModel):
 
 class ContextUpsertRequest(BaseModel):
     documents: List[Document]
-    batch_size: int = 100
+    batch_size: int = 200
 
 
 class ContextDeleteRequest(BaseModel):