Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Expose embeddings API #490

Merged
merged 5 commits into from
May 28, 2024
Merged

Expose embeddings API #490

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
00f0cc0
Extract _make_rag_request()
fantix May 1, 2024
6642a98
Expose embeddings API
fantix May 1, 2024
8d0d913
CRF: rename to `generate_embeddings()`
fantix May 1, 2024
aa3c755
CRF: unpack embedding result
fantix May 1, 2024
cc6c9a9
Unbox embedding data
fantix May 28, 2024
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
69 changes: 43 additions & 26 deletions edgedb/ai/core.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -85,6 +85,23 @@ def with_context(self, **kwargs) -> typing.Self:
rv.client = self.client
return rv

def _make_rag_request(
self,
*,
message: str,
context: typing.Optional[types.QueryContext] = None,
stream: bool,
) -> types.RAGRequest:
if context is None:
context = self.context
return types.RAGRequest(
model=self.options.model,
prompt=self.options.prompt,
context=context,
query=message,
stream=stream,
)


class EdgeDBAI(BaseEdgeDBAI):
client: httpx.Client
Expand All @@ -95,14 +112,10 @@ def _init_client(self, **kwargs):
def query_rag(
self, message: str, context: typing.Optional[types.QueryContext] = None
) -> str:
if context is None:
context = self.context
resp = self.client.post(
**types.RAGRequest(
model=self.options.model,
prompt=self.options.prompt,
**self._make_rag_request(
context=context,
query=message,
message=message,
stream=False,
).to_httpx_request()
)
Expand All @@ -111,24 +124,27 @@ def query_rag(

def stream_rag(
self, message: str, context: typing.Optional[types.QueryContext] = None
):
if context is None:
context = self.context
) -> typing.Iterator[str]:
with httpx_sse.connect_sse(
self.client,
"post",
**types.RAGRequest(
model=self.options.model,
prompt=self.options.prompt,
**self._make_rag_request(
context=context,
query=message,
message=message,
stream=True,
).to_httpx_request(),
) as event_source:
event_source.response.raise_for_status()
for sse in event_source.iter_sse():
yield sse.data

def generate_embeddings(self, *inputs: str, model: str) -> list[float]:
resp = self.client.post(
"/embeddings", json={"input": inputs, "model": model}
)
resp.raise_for_status()
return resp.json()["data"][0]["embedding"]


class AsyncEdgeDBAI(BaseEdgeDBAI):
client: httpx.AsyncClient
Expand All @@ -139,14 +155,10 @@ def _init_client(self, **kwargs):
async def query_rag(
self, message: str, context: typing.Optional[types.QueryContext] = None
) -> str:
if context is None:
context = self.context
resp = await self.client.post(
**types.RAGRequest(
model=self.options.model,
prompt=self.options.prompt,
**self._make_rag_request(
context=context,
query=message,
message=message,
stream=False,
).to_httpx_request()
)
Expand All @@ -155,20 +167,25 @@ async def query_rag(

async def stream_rag(
self, message: str, context: typing.Optional[types.QueryContext] = None
):
if context is None:
context = self.context
) -> typing.Iterator[str]:
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This returned str is a JSON string like:

{"type": "content_block_delta","index":0,"delta":{"type": "text_delta", "text": " blocking"}}

async with httpx_sse.aconnect_sse(
self.client,
"post",
**types.RAGRequest(
model=self.options.model,
prompt=self.options.prompt,
**self._make_rag_request(
context=context,
query=message,
message=message,
stream=True,
).to_httpx_request(),
) as event_source:
event_source.response.raise_for_status()
async for sse in event_source.aiter_sse():
yield sse.data

async def generate_embeddings(
self, *inputs: str, model: str
) -> list[float]:
resp = await self.client.post(
"/embeddings", json={"input": inputs, "model": model}
)
resp.raise_for_status()
return resp.json()["data"][0]["embedding"]
Loading