Hello everyone,
I’m currently working on a project involving two different bots that use different vectorizers for creating and managing schemas and databases in Weaviate. However, I’m encountering an issue where only one of the bots’ vectorizers previews correctly in the Weaviate collections console. using amazon.titan-embed-text-v2:0
TLDR
The vectorizer for GPT is working correctly, but it’s not functioning for AWS Bedrock. Attempts to fix it have been unsuccessful so far.
Details of Implementations:
ChatGPT Bot
Get Client Function:
def get_client(database_name):
WCS_API_KEY, WCS_CLUSTER_URL, OPENAI_APIKEY, class_name = database_picker(database_name)
max_retries = 7
retry_delay_seconds = 2
for attempt in range(1, max_retries + 1):
try:
client = weaviate.connect_to_wcs(
cluster_url=WCS_CLUSTER_URL,
auth_credentials=weaviate.auth.AuthApiKey(WCS_API_KEY),
headers={"X-OpenAI-Api-Key": OPENAI_APIKEY},
skip_init_checks=True
)
return client
except Exception as e:
print(f"Attempt {attempt} failed with error: {str(e)}")
if attempt < max_retries:
print(f"Retrying in {retry_delay_seconds} seconds...")
time.sleep(retry_delay_seconds)
Generate Database Function:
def generate_database(class_name, client):
try:
client.collections.create(
name=class_name,
vectorizer_config=wvc.config.Configure.Vectorizer.text2vec_openai(),
generative_config=wvc.config.Configure.Generative.openai(),
properties=[
wvc.config.Property(
name=class_name,
data_type=wvc.config.DataType.TEXT
)
]
)
except Exception as e:
return {"error": str(e)}
finally:
client.close()
Bedrock Bot
using amazon.titan-embed-text-v2:0
Get Client Function:
for attempt in range(0, RETRY_AMOUNT):
try:
client = None
client = weaviate.connect_to_wcs(
cluster_url=os.getenv("WCS_CLUSTER_URL"),
auth_credentials=AuthApiKey(os.getenv("WCS_API_KEY")),
skip_init_checks=os.getenv("W_SKIP_INIT_CHECK"),
headers={
"X-AWS-Access-Key": os.getenv("ACCESS_KEY_ID"),
"X-AWS-Secret-Key": os.getenv("SECRET_ACCESS_KEY"),
},
)
while client.is_ready() == False:
pass
return client
Generate Database Function:
try:
client.collections.create(
class_name,
vectorizer_config=[
Configure.NamedVectors.text2vec_aws(
name=class_name,
region=os.getenv("WEAVIATE_REGION"),
service=os.getenv("WEAVIATE_SERVICE"),
model=os.getenv("WEAVIATE_MODEL"),
)
],
properties=[
wvc.config.Property(
name=class_name,
data_type=wvc.config.DataType.TEXT
)
],
)
return json.dumps({"result": "Database class created successfully."})
except Exception as e:
return {"error": str(e)}
Issue:
While both bots create schemas and databases successfully, only the ChatGPT bot’s vectorizer appears correctly in the Weaviate collections console. The Bedrock bot’s vectorizer does not show up as expected.
Things I’ve Tried:
- Double-check the configuration settings for both bots.
- Verified API keys and permissions.
- Ensured that both bots have the same class name and data type configurations.
Request for Help:
- Has anyone faced a similar issue with different vectorizers in Weaviate?
- Are there specific settings or configurations that I might be missing for the Bedrock bot?
- Any troubleshooting steps or advice would be greatly appreciated.
Thank you in advance for your help!
image for GPT on the console
image for the using amazon.titan-embed-text-v2:0
