Upgrade langchain-core (1.2.2) (#161)

* upgrade langchain-core

* adjust chunk size for embeddings generation

* docker build/push for pushes to release/* branches

Author: GFJHogue

.github/workflows/ci.yml

@@ -9,6 +9,7 @@ on:
   push:
     branches:
       - main
+      - release/*
 
 permissions:
   id-token: write
@@ -81,7 +82,7 @@ jobs:
                 path: /tmp/image.tar
 
     docker-push:
-        if: ${{ github.event_name == 'push' && github.ref == 'refs/heads/main' }}
+        if: ${{ github.event_name == 'push' && (github.ref == 'refs/heads/main' || startsWith(github.ref, 'refs/heads/release/')) }}
         needs: docker-build
         runs-on: ubuntu-latest
 

poetry.lock

@@ -53,13 +53,13 @@ def upload_to_chromadb(
     embeddings_instance: Embeddings
     if hf_model is None:  # Use OpenAI
         embeddings_instance = OpenAIEmbeddings(
-            chunk_size=500,
+            chunk_size=400,
             show_progress_bar=True,
         )
     elif hf_model.startswith("openai/text-embedding-"):
         embeddings_instance = OpenAIEmbeddings(
             model=hf_model[len("openai/") :],
-            chunk_size=500,
+            chunk_size=400,
             show_progress_bar=True,
         )
     elif "HUGGINGFACEHUB_API_TOKEN" in os.environ:

src/data_generation/reactome/__init__.py

@@ -44,13 +44,13 @@ def upload_to_chromadb(
         print("Using OpenAI embeddings")
         embeddings_instance = OpenAIEmbeddings(
             model="text-embedding-3-large",
-            chunk_size=800,
+            chunk_size=500,
             show_progress_bar=True,
         )
     elif hf_model.startswith("openai/text-embedding-"):
         embeddings_instance = OpenAIEmbeddings(
             model=hf_model[len("openai/") :],
-            chunk_size=800,
+            chunk_size=500,
             show_progress_bar=True,
         )
     elif "HUGGINGFACEHUB_API_TOKEN" in os.environ: