bugfix: fixed update train

Brain/src/rising_plugin/csv_embed.py

@@ -4,6 +4,7 @@ from langchain.document_loaders.csv_loader import CSVLoader
 from langchain.embeddings.openai import OpenAIEmbeddings
 import json
 
+from Brain.src.model.req_model import ReqModel
 from ..common.utils import OPENAI_API_KEY
 
 
@@ -25,8 +26,8 @@ def csv_embed():
 """getting embed"""
 
 
-def get_embed(data: str) -> list[float]:
-    embeddings = OpenAIEmbeddings(openai_api_key=OPENAI_API_KEY)
+def get_embed(data: str, setting:ReqModel) -> list[float]:
+    embeddings = OpenAIEmbeddings(openai_api_key=setting.openai_key)
     return embeddings.embed_query(data)
 
 

Brain/src/rising_plugin/image_embedding.py

@@ -12,8 +12,8 @@ from ..model.image_model import ImageModel
 from ..model.req_model import ReqModel
 
 
-def get_embeddings():
-    return OpenAIEmbeddings(openai_api_key=OPENAI_API_KEY)
+def get_embeddings(setting: ReqModel):
+    return OpenAIEmbeddings(openai_api_key=setting.openai_key)
 
 
 # def embed_image_text(image_text: str, image_name: str, uuid: str) -> str:
@@ -24,7 +24,7 @@ def embed_image_text(image: ImageModel, setting: ReqModel) -> str:
         {image.image_text}
         """
 
-    embed_image = get_embeddings().embed_query(prompt_template)
+    embed_image = get_embeddings(setting=setting).embed_query(prompt_template)
     index = init_pinecone(index_name=PINECONE_INDEX_NAME, setting=setting)
 
     """create | update | delete in pinecone"""
@@ -48,7 +48,7 @@ def embed_image_text(image: ImageModel, setting: ReqModel) -> str:
 
 
 def query_image_text(image_content, message, setting: ReqModel):
-    embed_image = get_embeddings().embed_query(
+    embed_image = get_embeddings(setting=setting).embed_query(
         get_prompt_image_with_message(image_content, message)
     )
     index = init_pinecone(index_name=PINECONE_INDEX_NAME, setting=setting)

Brain/src/rising_plugin/risingplugin.py

@@ -64,7 +64,7 @@ def llm_rails(
 
     """step 1: handle with gpt-4"""
 
-    query_result = get_embed(query)
+    query_result = get_embed(data=query, setting=setting)
     try:
         relatedness_data = index.query(
             vector=query_result,

Brain/src/router/train_router.py

@@ -20,7 +20,7 @@ def construct_blueprint_train_api() -> APIRouter:
         status_code=200, schema={"message": "message", "result": "test_result"}
     )"""
 
-    @router.get("")
+    @router.post("")
     def read_all_documents(data: BasicReq):
         # parsing params
         try:
@@ -82,7 +82,7 @@ def construct_blueprint_train_api() -> APIRouter:
     @generator.response( status_code=200, schema={"message": "message", "result": {"document_id": "document_id", 
     "page_content":"page_content"}} )"""
 
-    @router.post("")
+    @router.post("/create")
     def create_document_train(data: Document):
         # parsing params
         try:

Brain/src/service/contact_service.py

@@ -33,7 +33,7 @@ class ContactsService:
             key = contact.contact_id
             value = f"{contact.display_name}, {contact.get_str_phones()}"
             # get vectoring data(embedding data)
-            vectoring_values = get_embed(value)
+            vectoring_values = get_embed(data=value, setting=self.setting)
             # create | update | delete pinecone
             if contact.status == ContactStatus.CREATED:
                 add_pinecone(
@@ -60,7 +60,7 @@ class ContactsService:
     response: list of contactId as index key of pinecone"""
 
     def query_contacts(self, uuid: str, search: str) -> List[str]:
-        vector_data = get_embed(search)
+        vector_data = get_embed(data=search, setting=self.setting)
         index = init_pinecone(index_name=PINECONE_INDEX_NAME, setting=self.setting)
         relatedness_data = index.query(
             vector=vector_data,

Brain/src/service/train_service.py

@@ -101,12 +101,12 @@ class TrainService:
         result = list()
         pinecone_namespace = self.get_pinecone_index_namespace()
         for item in documents:
-            query_result = get_embed(item["page_content"])
+            query_result = get_embed(data=item["page_content"], setting=self.setting)
             result.append(query_result)
             key = item["document_id"]
             value = f'{item["page_content"]}'
             # get vectoring data(embedding data)
-            vectoring_values = get_embed(value)
+            vectoring_values = get_embed(data=value, setting=self.setting)
             add_pinecone(
                 namespace=pinecone_namespace,
                 key=key,
@@ -120,12 +120,12 @@ class TrainService:
         self.init_firestore()
         pinecone_namespace = self.get_pinecone_index_namespace()
         result = list()
-        query_result = get_embed(page_content)
+        query_result = get_embed(data=page_content, setting=self.setting)
         result.append(query_result)
         key = document_id
         value = f"{page_content}, {query_result}"
         # get vectoring data(embedding data)
-        vectoring_values = get_embed(value)
+        vectoring_values = get_embed(data=value, setting=self.setting)
         add_pinecone(
             namespace=pinecone_namespace,
             key=key,

requirements.txt

@@ -68,4 +68,5 @@ wrapt==1.15.0
 yarl==1.8.2
 twilio==8.2.1
 nemoguardrails==0.2.0
-user-agents==2.2.0
+user-agents==2.2.0
+tiktoken==0.4.0