Spaces:
Sleeping
Sleeping
Delete create_embeddings2.py
Browse files- create_embeddings2.py +0 -58
create_embeddings2.py
DELETED
|
@@ -1,58 +0,0 @@
|
|
| 1 |
-
import pinecone
|
| 2 |
-
from langchain_google_genai import GoogleGenerativeAIEmbeddings
|
| 3 |
-
from variables import variables # Import the list of variable names
|
| 4 |
-
import uuid
|
| 5 |
-
|
| 6 |
-
# Initialize Google Embeddings
|
| 7 |
-
google_embeddings = GoogleGenerativeAIEmbeddings(
|
| 8 |
-
model="models/embedding-001", # Correct model name
|
| 9 |
-
google_api_key="AIzaSyANNRKfEb-YnVIBaSAq6hQ38XpxxGwvaws" # Your API key
|
| 10 |
-
)
|
| 11 |
-
|
| 12 |
-
# Initialize Pinecone instance
|
| 13 |
-
pc = pinecone.Pinecone(
|
| 14 |
-
api_key="4a80f293-ae6d-489a-a7d8-33ea3fcdd26b" # Your Pinecone API key
|
| 15 |
-
)
|
| 16 |
-
|
| 17 |
-
# Define the Pinecone index name (make sure it exists in your Pinecone dashboard)
|
| 18 |
-
index_name = "iocl2"
|
| 19 |
-
index = pc.Index(index_name)
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
def create_embedding(variable):
|
| 23 |
-
try:
|
| 24 |
-
content = variable.get("description", None)
|
| 25 |
-
url = variable.get("url", "")
|
| 26 |
-
tag = variable.get("tag", "")
|
| 27 |
-
updated_url = ""
|
| 28 |
-
|
| 29 |
-
if isinstance(url, list):
|
| 30 |
-
updated_url = ",".join(url)
|
| 31 |
-
else:
|
| 32 |
-
updated_url = url
|
| 33 |
-
|
| 34 |
-
embedding = google_embeddings.embed_query(content)
|
| 35 |
-
vectors = []
|
| 36 |
-
vectors.append({
|
| 37 |
-
'id': str(uuid.uuid4()),
|
| 38 |
-
'values': embedding,
|
| 39 |
-
'metadata': {
|
| 40 |
-
'chunk': content,
|
| 41 |
-
"url": updated_url,
|
| 42 |
-
"tag": tag
|
| 43 |
-
}
|
| 44 |
-
})
|
| 45 |
-
index.upsert(vectors)
|
| 46 |
-
print(f"Inserted the chunk: {updated_url}")
|
| 47 |
-
except Exception as e:
|
| 48 |
-
print(f"Error occurred: {e}")
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
# Iterate over the variable names and create embeddings
|
| 52 |
-
for variable_name in variables:
|
| 53 |
-
# Dynamically import the variable from paragraphs2
|
| 54 |
-
variable = __import__('paragraphs2', fromlist=[variable_name])
|
| 55 |
-
variable_data = getattr(variable, variable_name)
|
| 56 |
-
print(f"trying to create embedding for {variable}")
|
| 57 |
-
# Call the create_embedding function with the variable data
|
| 58 |
-
create_embedding(variable_data)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|