EmbedderAzureOpenAIConfig use langchain.embeddings.OpenAIEmbeddings

Do not use the Fake Embeddings anymore. Requires LangChain 0.0.155 because it contains langchain-ai/langchain#3711
zioproto · May 2, 2023 · 4ef1a79 · 4ef1a79
1 parent e00a8fb
commit 4ef1a79
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 10 deletions.
diff --git a/core/cat/factory/embedder.py b/core/cat/factory/embedder.py
@@ -38,15 +38,16 @@ class Config:
             "description": "Configuration for OpenAI embeddings",
         }
 
+# https://python.langchain.com/en/latest/_modules/langchain/embeddings/openai.html#OpenAIEmbeddings
 class EmbedderAzureOpenAIConfig(EmbedderSettings):
     openai_api_key: str
-    model_name: str
-    api_base: str
+    model: str
+    openai_api_base: str
     api_type: str
     api_version: str
-    deployment_name: str
+    deployment: str
 
-    _pyclass: PyObject = langchain.embeddings.FakeEmbeddings
+    _pyclass: PyObject = langchain.embeddings.OpenAIEmbeddings
 
     class Config:
         schema_extra = {

diff --git a/core/cat/factory/llm.py b/core/cat/factory/llm.py
@@ -104,8 +104,10 @@ class Config:
 # https://python.langchain.com/en/latest/modules/models/llms/integrations/azure_openai_example.html
 class LLMAzureOpenAIConfig(LLMSettings):
     openai_api_key: str
-    api_base: str
+    openai_api_base: str
     api_type: str = "azure"
+    # https://learn.microsoft.com/en-us/azure/cognitive-services/openai/reference#completions
+    # Current supported versions 2022-12-01 or 2023-03-15-preview
     api_version: str = "2022-12-01"
     deployment_name: str = "text-davinci-003"
     model_name: str = "text-davinci-003"

diff --git a/core/cat/mad_hatter/core_plugin/hooks/models.py b/core/cat/mad_hatter/core_plugin/hooks/models.py
@@ -88,10 +88,20 @@ def get_language_embedder(cat):
                 }
             )
         else:
-            print(cat.llm)
             embedder = embedders.EmbedderAzureOpenAIConfig.get_embedder_from_config(
-                {"size": 1536}  # mock openai embedding size
-
+                {
+                   "openai_api_key": openai_key,
+                   "openai_api_type": "azure",
+                   "model": "text-embedding-ada-002",
+                   # Now the only model for embeddings is text-embedding-ada-002
+                   # It is also possible to use the Azure "deployment" name that is user defined
+                   # when the model is deployed to Azure.
+                   # "deployment": "my-text-embedding-ada-002",
+                   "openai_api_base": cat.llm.openai_api_base,
+                   # https://learn.microsoft.com/en-us/azure/cognitive-services/openai/reference#embeddings
+                   # current supported versions 2022-12-01 or 2023-03-15-preview
+                   "openai_api_version": "2022-12-01",
+                }
             )
     elif "COHERE_KEY" in os.environ:
         embedder = embedders.EmbedderCohereConfig.get_embedder_from_config(

diff --git a/core/requirements.txt b/core/requirements.txt
@@ -7,8 +7,8 @@ sqlmodel==0.0.8
 pandas==1.5.3
 scikit-learn==1.2.1
 qdrant_client==1.1.2
-langchain==0.0.121
-openai==0.27.0
+langchain==0.0.155
+openai==0.27.5
 cohere==4.0.4
 huggingface-hub==0.13.2
 beautifulsoup4==4.12.0