feat: Support langchain transformer on fabric #2036
Conversation
|
Hey @lhrotk 👋! We use semantic commit messages to streamline the release process. Examples of commit messages with semantic prefixes:
To test your commit locally, please follow our guild on building from source. |
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
Summary by GPT-4
This code change adds support for running the LangchainTransform on Synapse internal platform. It checks if the code is running on Synapse internal platform and sets the default URL accordingly. If the URL is not set and the code is running on Synapse internal, it initializes a personalized session using OpenAIPrerun. Otherwise, it uses the provided subscription key, URL, and API version to configure OpenAI API.
Here's a summary of the changes:
- Import
running_on_synapse_internalfromsynapse.ml.core.platform. - Check if running on Synapse internal platform using
running_on_synapse_internal(). - If running on Synapse internal platform, set the default URL using
get_fabric_env_config().fabric_env_config.ml_workload_endpoint + "cognitive/openai". - In
_transformmethod, check if running on Synapse internal platform and URL is not set. - If true, initialize a personalized session using
OpenAIPrerun(api_base=self.getUrl()).init_personalized_session(None). - Else, configure OpenAI API using provided subscription key, URL, and API version.
This change allows LangchainTransform to work seamlessly both on Synapse internal platform and other platforms by automatically configuring the appropriate settings based on the environment it's running in.
Suggestions
The changes in this PR seem to be well-implemented and do not require any suggestions for improvement.
| self.useFabricInternalEndpoints = Param(self, "useFabricInternalEndpoints", "use internal openai endpoints when on fabric") | ||
| self.running_on_synapse_internal = running_on_synapse_internal() | ||
| if running_on_synapse_internal(): | ||
| from synapse.ml.fabric.service_discovery import get_fabric_env_config | ||
| self.fabric_env_config = get_fabric_env_config() |
There was a problem hiding this comment.
is there any way to do this more implicitly without an explicit new param that users need to set? Im thinking that if a user doesent set the url and its on fabric it defaults to the fabric one if it can find one, otherwise it will throw an error
There was a problem hiding this comment.
I get your point, it is possible to pass only the openai api_base. I will make the change
lhrotk
force-pushed
the
cl/langchain-fabric
branch
from
ae0e66f
to
c3cf48e
Compare
mhamilton723
changed the title
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Codecov Report
@@ Coverage Diff @@
## master #2036 +/- ##
==========================================
- Coverage 87.03% 87.02% -0.01%
==========================================
Files 306 306
Lines 16063 16063
Branches 852 852
==========================================
- Hits 13980 13979 -1
- Misses 2083 2084 +1 |
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
* support langchain transformer on fabric * avoid addtional param * format code --------- Co-authored-by: cruise <[email protected]>
Related Issues/PRs
N/A
What changes are proposed in this pull request?
Make Langchain transformer use fabric internal endpoints by default
How is this patch tested?
Manually tested on fabric
Does this PR change any dependencies?
Does this PR add a new feature? If so, have you added samples on website?