community[minor]: Add Anyscale embedding support (#17605)

**Description:** Add embedding model support for Anyscale Endpoint
**Dependencies:** openai

---------

Co-authored-by: Bagatur <baskaryan@gmail.com>

docs/docs/integrations/providers/anyscale.mdx

@@ -32,3 +32,11 @@ See a [usage example](/docs/integrations/chat/anyscale).
 ```python
 from langchain_community.chat_models.anyscale import ChatAnyscale
 ```
+
+## Embeddings
+
+See a [usage example](/docs/integrations/text_embedding/anyscale).
+
+```python
+from langchain_community.embeddings import AnyscaleEmbeddings
+```

libs/community/langchain_community/embeddings/__init__.py

@@ -18,6 +18,7 @@ from typing import Any
 _module_lookup = {
     "AlephAlphaAsymmetricSemanticEmbedding": "langchain_community.embeddings.aleph_alpha",  # noqa: E501
     "AlephAlphaSymmetricSemanticEmbedding": "langchain_community.embeddings.aleph_alpha",  # noqa: E501
+    "AnyscaleEmbeddings": "langchain_community.embeddings.anyscale",
     "AwaEmbeddings": "langchain_community.embeddings.awa",
     "AzureOpenAIEmbeddings": "langchain_community.embeddings.azure_openai",
     "BaichuanTextEmbeddings": "langchain_community.embeddings.baichuan",

libs/community/langchain_community/embeddings/anyscale.py

@@ -0,0 +1,75 @@
+"""Anyscale embeddings wrapper."""
+from __future__ import annotations
+
+from typing import Dict
+
+from langchain_core.pydantic_v1 import Field, SecretStr, root_validator
+from langchain_core.utils import convert_to_secret_str, get_from_dict_or_env
+
+from langchain_community.embeddings.openai import OpenAIEmbeddings
+from langchain_community.utils.openai import is_openai_v1
+
+DEFAULT_API_BASE = "https://api.endpoints.anyscale.com/v1"
+DEFAULT_MODEL = "thenlper/gte-large"
+
+
+class AnyscaleEmbeddings(OpenAIEmbeddings):
+    """`Anyscale` Embeddings API."""
+
+    anyscale_api_key: SecretStr = Field(default=None)
+    """AnyScale Endpoints API keys."""
+    model: str = Field(default=DEFAULT_MODEL)
+    """Model name to use."""
+    anyscale_api_base: str = Field(default=DEFAULT_API_BASE)
+    """Base URL path for API requests."""
+    tiktoken_enabled: bool = False
+    """Set this to False for non-OpenAI implementations of the embeddings API"""
+    embedding_ctx_length: int = 500
+    """The maximum number of tokens to embed at once."""
+
+    @property
+    def lc_secrets(self) -> Dict[str, str]:
+        return {
+            "anyscale_api_key": "ANYSCALE_API_KEY",
+        }
+
+    @root_validator()
+    def validate_environment(cls, values: dict) -> dict:
+        """Validate that api key and python package exists in environment."""
+        values["anyscale_api_key"] = convert_to_secret_str(
+            get_from_dict_or_env(
+                values,
+                "anyscale_api_key",
+                "ANYSCALE_API_KEY",
+            )
+        )
+        values["anyscale_api_base"] = get_from_dict_or_env(
+            values,
+            "anyscale_api_base",
+            "ANYSCALE_API_BASE",
+            default=DEFAULT_API_BASE,
+        )
+        try:
+            import openai
+
+        except ImportError:
+            raise ImportError(
+                "Could not import openai python package. "
+                "Please install it with `pip install openai`."
+            )
+        if is_openai_v1():
+            # For backwards compatibility.
+            client_params = {
+                "api_key": values["anyscale_api_key"].get_secret_value(),
+                "base_url": values["anyscale_api_base"],
+            }
+            values["client"] = openai.OpenAI(**client_params).embeddings
+        else:
+            values["openai_api_base"] = values["anyscale_api_base"]
+            values["openai_api_key"] = values["anyscale_api_key"].get_secret_value()
+            values["client"] = openai.Embedding
+        return values
+
+    @property
+    def _llm_type(self) -> str:
+        return "anyscale-embedding"

libs/community/tests/unit_tests/embeddings/test_imports.py

@@ -2,6 +2,7 @@ from langchain_community.embeddings import __all__
 
 EXPECTED_ALL = [
     "OpenAIEmbeddings",
+    "AnyscaleEmbeddings",
     "AzureOpenAIEmbeddings",
     "BaichuanTextEmbeddings",
     "ClarifaiEmbeddings",