Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions singlestoredb/ai/chat.py
Original file line number Diff line number Diff line change
Expand Up @@ -58,11 +58,13 @@ def SingleStoreChatFactory(
model_name=model_name,
name='',
connection_url=base_url,
internal_connection_url=base_url,
project_id='',
hosting_platform=hosting_platform,
)
if base_url is not None:
info.connection_url = base_url
info.internal_connection_url = base_url
if hosting_platform is not None:
info.hosting_platform = hosting_platform

Expand Down Expand Up @@ -98,7 +100,7 @@ def SingleStoreChatFactory(
cfg = Config(**cfg_kwargs)
client = boto3.client(
'bedrock-runtime',
endpoint_url=info.connection_url,
endpoint_url=info.internal_connection_url,
region_name='us-east-1',
aws_access_key_id='placeholder',
aws_secret_access_key='placeholder',
Expand Down Expand Up @@ -138,7 +140,7 @@ def _inject_headers(request: Any, **_ignored: Any) -> None:

return ChatBedrockConverse(
model_id=model_name,
endpoint_url=info.connection_url,
endpoint_url=info.internal_connection_url,
region_name='us-east-1',
aws_access_key_id='placeholder',
aws_secret_access_key='placeholder',
Expand All @@ -152,7 +154,7 @@ def _inject_headers(request: Any, **_ignored: Any) -> None:
token = api_key if api_key is not None else token_env

openai_kwargs = dict(
base_url=info.connection_url,
base_url=info.internal_connection_url,
api_key=token,
model=model_name,
streaming=streaming,
Expand Down
8 changes: 5 additions & 3 deletions singlestoredb/ai/embeddings.py
Original file line number Diff line number Diff line change
Expand Up @@ -57,11 +57,13 @@ def SingleStoreEmbeddingsFactory(
model_name=model_name,
name='',
connection_url=base_url,
internal_connection_url=base_url,
project_id='',
hosting_platform=hosting_platform,
)
if base_url is not None:
info.connection_url = base_url
info.internal_connection_url = base_url
if hosting_platform is not None:
info.hosting_platform = hosting_platform

Expand Down Expand Up @@ -97,7 +99,7 @@ def SingleStoreEmbeddingsFactory(
cfg = Config(**cfg_kwargs)
client = boto3.client(
'bedrock-runtime',
endpoint_url=info.connection_url,
endpoint_url=info.internal_connection_url,
region_name='us-east-1',
aws_access_key_id='placeholder',
aws_secret_access_key='placeholder',
Expand Down Expand Up @@ -129,7 +131,7 @@ def _inject_headers(request: Any, **_ignored: Any) -> None:

return BedrockEmbeddings(
model_id=model_name,
endpoint_url=info.connection_url,
endpoint_url=info.internal_connection_url,
region_name='us-east-1',
aws_access_key_id='placeholder',
aws_secret_access_key='placeholder',
Expand All @@ -142,7 +144,7 @@ def _inject_headers(request: Any, **_ignored: Any) -> None:
token = api_key if api_key is not None else token_env

openai_kwargs = dict(
base_url=info.connection_url,
base_url=info.internal_connection_url,
api_key=token,
model=model_name,
)
Expand Down
4 changes: 4 additions & 0 deletions singlestoredb/management/inference_api.py
Original file line number Diff line number Diff line change
Expand Up @@ -152,6 +152,7 @@ class InferenceAPIInfo(object):
model_name: str
name: str
connection_url: str
internal_connection_url: str
project_id: str
hosting_platform: str
_manager: Optional['InferenceAPIManager']
Expand All @@ -162,12 +163,14 @@ def __init__(
model_name: str,
name: str,
connection_url: str,
internal_connection_url: str,
project_id: str,
hosting_platform: str,
manager: Optional['InferenceAPIManager'] = None,
):
self.service_id = service_id
self.connection_url = connection_url
self.internal_connection_url = internal_connection_url
self.model_name = model_name
self.name = name
self.project_id = project_id
Expand Down Expand Up @@ -198,6 +201,7 @@ def from_dict(
model_name=obj['modelName'],
name=obj['name'],
connection_url=obj['connectionURL'],
internal_connection_url=obj['internalConnectionURL'],
hosting_platform=obj['hostingPlatform'],
)
return out
Expand Down