Add documentation for custom embeddings model and small tweaks.

CHANGELOG.md

@@ -11,6 +11,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Support for [custom configuration data](./docs/user_guide/configuration-guide.md#custom-data).
 - Example for using [custom LLM and multiple KBs](./examples/multi_kb/README.md)
 - Support for [`PROMPTS_DIR`](./docs/user_guide/advanced/prompt-customization.md#prompt-configuration).
+- [#101](https://github.com/NVIDIA/NeMo-Guardrails/pull/101) Support for [using OpenAI embeddings](./docs/user_guide/configuration-guide.md#the-embeddings-model) models in addition to SentenceTransformers.
 
 ### Fixed
 

docs/user_guide/configuration-guide.md

@@ -111,6 +111,28 @@ models:
 ```
 
 
+### The Embeddings Model
+
+To configure the embeddings model that is used for the various steps in the [guardrails process](../architecture/README.md) (e.g., canonical form generation, next step generation) you can add a model configuration in the `models` key as shown below:
+
+```yaml
+models:
+ - ...
+ - type: embedding
+ engine: SentenceTransformers
+ model: all-MiniLM-L6-v2
+```
+
+The `SentenceTransformers` engine is the default one and uses the `all-MiniLM-L6-v2` model. NeMo Guardrails also supports using OpenAI models for computing the embeddings, e.g.:
+
+```yaml
+models:
+ - ...
+ - type: embedding
+ engine: openai
+ model: text-embedding-ada-002
+```
+
 ### General Instruction
 
 The general instruction (similar to a system prompt) gets appended at the beginning of every prompt, and you can configure it as shown below:

nemoguardrails/actions/llm/generation.py

@@ -67,10 +67,13 @@ def __init__(
  self.llm = llm
  self.verbose = verbose
 
- # If we have a customized embedding model, we'll use it.
+ # The default embeddings model is using SentenceTransformers
  self.embedding_model = "all-MiniLM-L6-v2"
+ self.embedding_engine = "SentenceTransformers"
+
+ # If we have a customized embedding model, we'll use it.
  for model in self.config.models:
- if model.type == "embedding":
+ if model.type == "embeddings":
  self.embedding_model = model.model
  assert model.engine in get_embedding_provider_names()
  self.embedding_engine = model.engine

nemoguardrails/rails/llm/llmrails.py

@@ -124,7 +124,7 @@ def _init_llms(self):
  # to search for the main model config.
 
  for llm_config in self.config.models:
- if llm_config.type == "embedding":
+ if llm_config.type == "embeddings":
  if llm_config.engine not in get_embedding_provider_names():
  raise Exception(f"Unknown embedding engine: {llm_config.engine}")
  else: