Merge pull request #647 from NVIDIA/docs/doc-updates-0.9.1

Documentation updates for 0.9.1.

README.md

@@ -100,7 +100,7 @@ The input and output format for the `generate` method is similar to the [Chat Co
 
 #### Async API
 
-NeMo Guardrails is an async-first toolkit, i.e., the core mechanics are implemented using the Python async model. The public methods have both a sync and an async version (e.g., `LLMRails.generate` and `LLMRails.generate_async`).
+NeMo Guardrails is an async-first toolkit, which means that the core mechanics are implemented using the Python async model. The public methods have both a sync and an async version, such as `LLMRails.generate` and `LLMRails.generate_async`.
 
 ### Supported LLMs
 
@@ -116,7 +116,7 @@ NeMo Guardrails supports five main types of guardrails:
 
 1. **Input rails**: applied to the input from the user; an input rail can reject the input, stopping any additional processing, or alter the input (e.g., to mask potentially sensitive data, to rephrase).
 
-2. **Dialog rails**: influence how the LLM is prompted; dialog rails operate on canonical form messages (more details [here](https://docs.nvidia.com/nemo/guardrails/user_guides/colang-language-syntax-guide.html)) and determine if an action should be executed, if the LLM should be invoked to generate the next step or a response, if a predefined response should be used instead, etc.
+2. **Dialog rails**: influence how the LLM is prompted; dialog rails operate on canonical form messages for details see [Colang Guide](https://docs.nvidia.com/nemo/guardrails/user_guides/colang-language-syntax-guide.html)) and determine if an action should be executed, if the LLM should be invoked to generate the next step or a response, if a predefined response should be used instead, etc.
 
 3. **Retrieval rails**: applied to the retrieved chunks in the case of a RAG (Retrieval Augmented Generation) scenario; a retrieval rail can reject a chunk, preventing it from being used to prompt the LLM, or alter the relevant chunks (e.g., to mask potentially sensitive data).
 
@@ -140,7 +140,7 @@ The standard structure for a guardrails configuration folder looks like this:
 │   ├── ...
 ```
 
-The `config.yml` contains all the general configuration options (e.g., LLM models, active rails, custom configuration data), the `config.py` contains any custom initialization code and the `actions.py` contains any custom python actions. For a complete overview, check out the [Configuration Guide](https://docs.nvidia.com/nemo/guardrails/user_guides/configuration-guide.html).
+The `config.yml` contains all the general configuration options, such as LLM models, active rails, and custom configuration data". The `config.py` file contains any custom initialization code and the `actions.py` contains any custom python actions. For a complete overview, see the [Configuration Guide](https://docs.nvidia.com/nemo/guardrails/user_guides/configuration-guide.html).
 
 Below is an example `config.yml`:
 
@@ -208,19 +208,25 @@ define flow
 
 To configure and implement various types of guardrails, this toolkit introduces **Colang**, a modeling language specifically created for designing flexible, yet controllable, dialogue flows. Colang has a python-like syntax and is designed to be simple and intuitive, especially for developers.
 
-**NOTE**: Currently two versions of Colang are supported (1.0 and 2.0-beta) and Colang 1.0 is the default. Versions 0.1.0 up to 0.7.1 of NeMo Guardrails used Colang 1.0 exclusively. Versions 0.8.0 introduced Colang 2.0-alpha and version 0.9.0 introduced Colang 2.0-beta. We expect Colang 2.0 to go out of Beta and replace 1.0 as the default option in NeMo Guardrails version 0.11.0.
+**NOTE**: Currently two versions of Colang, 1.0 and 2.0, are supported and Colang 1.0 is the default. Versions 0.1.0 up to 0.7.1 of NeMo Guardrails used Colang 1.0 exclusively. Versions 0.8.0 introduced Colang 2.0-alpha and version 0.9.0 introduced Colang 2.0-beta. We expect Colang 2.0 to go out of Beta and replace 1.0 as the default option in NeMo Guardrails version 0.11.0.
 
-For a brief introduction to the Colang 1.0 syntax, check out the [Colang 1.0 Language Syntax Guide](https://docs.nvidia.com/nemo/guardrails/user_guides/colang-language-syntax-guide.html).
+For a brief introduction to the Colang 1.0 syntax, see the [Colang 1.0 Language Syntax Guide](https://docs.nvidia.com/nemo/guardrails/user_guides/colang-language-syntax-guide.html).
 
-To get started with Colang 2.0, check out the [Colang 2.0 Documentation](https://docs.nvidia.com/nemo/guardrails/colang_2/overview.html).
+To get started with Colang 2.0, see the [Colang 2.0 Documentation](https://docs.nvidia.com/nemo/guardrails/colang_2/overview.html).
 
 ### Guardrails Library
 
 NeMo Guardrails comes with a set of [built-in guardrails](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html).
 
 > **NOTE**: The built-in guardrails are only intended to enable you to get started quickly with NeMo Guardrails. For production use cases, further development and testing of the rails are needed.
 
-Currently, the guardrails library includes guardrails for: [jailbreak detection](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#jailbreak-detection-heuristics), [output moderation](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#self-check-output), [fact-checking](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#fact-checking), [sensitive data detection](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#presidio-based-sensitive-data-detection), [hallucination detection](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#hallucination-detection), [input moderation using ActiveFence](<<https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#activefence>), [hallucination detection for RAG applications using Got It AI's TruthChecker API](docs/user_guides/guardrails-library.md#got-it-ai), and [RAG hallucination detection using Patronus Lynx](docs/user_guides/guardrails-library.md#patronus-lynx-based-rag-hallucination-detection).
+Currently, the guardrails library includes guardrails for:
+- [jailbreak detection](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#jailbreak-detection-heuristics)
+- [output moderation](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#self-check-output)
+- [fact-checking](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#fact-checking)
+- [sensitive data detection](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#presidio-based-sensitive-data-detection), [hallucination detection](https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#hallucination-detection)
+- [input moderation using ActiveFence] (https://docs.nvidia.com/nemo/guardrails/user_guides/guardrails-library.html#activefence>), [hallucination detection for RAG applications using Got It AI's TruthChecker API](docs/user_guides/guardrails-library.md#got-it-ai)
+[RAG hallucination detection using Patronus Lynx](docs/user_guides/guardrails-library.md#patronus-lynx-based-rag-hallucination-detection)
 
 ## CLI
 
@@ -266,7 +272,7 @@ Sample output:
 
 #### Docker
 
-To start a guardrails server, you can also use a Docker container. NeMo Guardrails provides a [Dockerfile](./Dockerfile) that you can use to build a `nemoguardrails` image. For more details, check out the guide for [using Docker](https://docs.nvidia.com/nemo/guardrails/user_guides/advanced/using-docker.html).
+To start a guardrails server, you can also use a Docker container. NeMo Guardrails provides a [Dockerfile](./Dockerfile) that you can use to build a `nemoguardrails` image. For further information, see the [using Docker](https://docs.nvidia.com/nemo/guardrails/user_guides/advanced/using-docker.html) section.
 
 ## Integration with LangChain
 

docs/user_guides/configuration-guide.md

@@ -86,11 +86,11 @@ The meaning of the attributes is as follows:
 You can use any LLM provider that is supported by LangChain, e.g., `ai21`, `aleph_alpha`, `anthropic`, `anyscale`, `azure`, `cohere`, `huggingface_endpoint`, `huggingface_hub`, `openai`, `self_hosted`, `self_hosted_hugging_face`. Check out the LangChain official documentation for the full list.
 
 ```{note}
-To use any of the providers, you will need to install additional packages; when you first try to use a configuration with a new provider, you will typically receive an error from LangChain that will instruct you on what packages should be installed.
+To use any of the providers, you must install additional packages; when you first try to use a configuration with a new provider, you will typically receive an error from LangChain that will instructs which packages you should install.
 ```
 
 ```{important}
-While from a technical perspective, you can instantiate any of the LLM providers above, depending on the capabilities of the model, some will work better than others with the NeMo Guardrails toolkit. The toolkit includes prompts that have been optimized for certain types of models (e.g., `openai`, `nemollm`). For others, you can optimize the prompts yourself (see the [LLM Prompts](#llm-prompts) section).
+Although you can instantiate any of the previously mentioned LLM providers, depending on the capabilities of the model, the NeMo Guardrails toolkit works better with some providers than others. The toolkit includes prompts that have been optimized for certain types of models, such as `openai` and `nemollm`. For others, you can optimize the prompts yourself following the information in the [LLM Prompts](#llm-prompts) section.
 ```
 #### NIM for LLMs
 
@@ -120,13 +120,13 @@ models:
 ```
 
 ```{important}
-To use the `nim` LLM provider, you must install the `langchain-nvidia-ai-endpoints` package (`pip install langchain-nvidia-ai-endpoints`).
+To use the `nim` LLM provider, install the `langchain-nvidia-ai-endpoints` package using the command `pip install langchain-nvidia-ai-endpoints`.
 ```
 
 
 #### NVIDIA AI Endpoints
 
-[NVIDIA AI Endpoints](https://www.nvidia.com/en-us/ai-data-science/foundation-models/) give users easy access to NVIDIA hosted API endpoints for NVIDIA AI Foundation Models like Llama 3, Mixtral 8x7B, Stable Diffusion, etc.
+[NVIDIA AI Endpoints](https://www.nvidia.com/en-us/ai-data-science/foundation-models/) give users easy access to NVIDIA hosted API endpoints for NVIDIA AI Foundation Models such as Llama 3, Mixtral 8x7B, and Stable Diffusion.
 These models, hosted on the [NVIDIA API catalog](https://build.nvidia.com/), are optimized, tested, and hosted on the NVIDIA AI platform, making them fast and easy to evaluate, further customize, and seamlessly run at peak performance on any accelerated stack.
 
 To use an LLM model through the NVIDIA AI Endpoints, use the following model configuration:
@@ -148,10 +148,10 @@ models:
 ```
 
 ```{important}
-To use the `nvidia_ai_endpoints` LLM provider, you must install the `langchain-nvidia-ai-endpoints` package (`pip install langchain-nvidia-ai-endpoints`) and configure a valid `NVIDIA_API_KEY`.
+To use the `nvidia_ai_endpoints` LLM provider, you must install the `langchain-nvidia-ai-endpoints` package using the command `pip install langchain-nvidia-ai-endpoints`, and configure a valid `NVIDIA_API_KEY`.
 ```
 
-For more details, check out this [user guide](./llm/nvidia_ai_endpoints/README.md).
+For further information, see the [user guide](./llm/nvidia_ai_endpoints/README.md).
 
 Here's an example configuration for using `llama3` model with [Ollama](https://ollama.com/):
 
@@ -257,7 +257,7 @@ models:
 
 ### The Embeddings Model
 
-To configure the embedding model 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:
+To configure the embedding model used for the various steps in the [guardrails process](../architecture/README.md), such as canonical form generation and next step generation, add a model configuration in the `models` key as shown in the following configuration file:
 
 ```yaml
 models:
@@ -279,7 +279,7 @@ models:
 
 #### Supported Embedding Providers
 
-The complete list of supported embedding providers is the following:
+The following tables lists the supported embedding providers:
 
 | Provider Name        | `engine_name`          | `model`                            |
 |----------------------|------------------------|------------------------------------|
@@ -289,7 +289,7 @@ The complete list of supported embedding providers is the following:
 | NVIDIA AI Endpoints  | `nvidia_ai_endpoints`  | `nv-embed-v1`, etc.                |
 
 ```{note}
-For any of the supported embedding providers you can use any of the supported models.
+You can use any of the supported models for any of the supported embedding providers.
 The previous table includes an example of a model that can be used.
 ```
 
@@ -298,7 +298,7 @@ The previous table includes an example of a model that can be used.
 You can also register a custom embedding provider by using the `LLMRails.register_embedding_provider` function.
 
 To register a custom LLM provider,
-you need to create a class that inherits from `EmbeddingModel` and register it in your `config.py`.
+create a class that inherits from `EmbeddingModel` and register it in your `config.py`.
 
 ```python
 from typing import List
@@ -354,7 +354,7 @@ models:
 
 ### Embedding Search Provider
 
-NeMo Guardrails uses embedding search (a.k.a. vector databases) for implementing the [guardrails process](../architecture/README.md#the-guardrails-process) and for the [knowledge base](#knowledge-base-documents) functionality. The default embedding search uses FastEmbed for computing the embeddings (the `all-MiniLM-L6-v2` model) and [Annoy](https://github.com/spotify/annoy) for performing the search. As shown in the previous section, the embeddings model supports both FastEmbed and OpenAI. SentenceTransformers is also supported.
+NeMo Guardrails uses embedding search, also called vector databases, for implementing the [guardrails process](../architecture/README.md#the-guardrails-process) and for the [knowledge base](#knowledge-base-documents) functionality. The default embedding search uses FastEmbed for computing the embeddings (the `all-MiniLM-L6-v2` model) and [Annoy](https://github.com/spotify/annoy) for performing the search. As shown in the previous section, the embeddings model supports both FastEmbed and OpenAI. SentenceTransformers is also supported.
 
 For advanced use cases or integrations with existing knowledge bases, you can [provide a custom embedding search provider](advanced/embedding-search-providers.md).
 

docs/user_guides/llm/nvidia_ai_endpoints/README.md

@@ -1,6 +1,6 @@
 # Using LLMs hosted on NVIDIA API Catalog
 
-This guide teaches you how to use NeMo Guardrails with LLMs hosted on NVIDIA API Catalog. It uses the [ABC Bot configuration](../../../../examples/bots/abc) and changes the model to `meta/llama3-70b-instruct`.
+This guide teaches you how to use NeMo Guardrails with LLMs hosted on NVIDIA API Catalog. It uses the [ABC Bot configuration](../../../../examples/bots/abc) and with the `meta/llama-3.1-70b-instruct` model. Similarly, you can use `meta/llama-3.1-405b-instruct`, `meta/llama-3.1-8b-instruct` or any other [AI Foundation Model](https://build.nvidia.com/explore/discover).
 
 ## Prerequisites
 
@@ -15,7 +15,7 @@ pip install -U --quiet langchain-nvidia-ai-endpoints
 2. An NVIDIA NGC account to access AI Foundation Models. To create a free account go to [NVIDIA NGC website](https://ngc.nvidia.com/).
 
 3. An API key from NVIDIA API Catalog:
-   - Generate an API key by navigating to the AI Foundation Models section on the NVIDIA NGC website, selecting a model with an API endpoint, and generating an API key. You can use this API key for all models available in the NVIDIA API Catalog.
+   - Generate an API key by navigating to the [AI Foundation Models](https://build.nvidia.com/explore/discover) section on the NVIDIA NGC website, selecting a model with an API endpoint, and generating an API key. You can use this API key for all models available in the NVIDIA API Catalog.
    - Export the NVIDIA API key as an environment variable:
 
 ```bash
@@ -45,7 +45,7 @@ Update the `models` section of the `config.yml` file to the desired model suppor
 models:
   - type: main
     engine: nvidia_ai_endpoints
-    model: meta/llama3-70b-instruct
+    model: meta/llama-3.1-70b-instruct
 ...
 ```
 
@@ -60,10 +60,6 @@ config = RailsConfig.from_path("./config")
 rails = LLMRails(config)
 ```
 
-```
-Fetching 7 files:   0%|          | 0/7 [00:00<?, ?it/s]
-```
-
 Test that it works:
 
 ```python
@@ -76,11 +72,11 @@ print(response['content'])
 ```
 
 ```
-According to the employee handbook, eligible employees are entitled to 20 days of paid vacation per year, accrued monthly.
+According to our company policy, you are eligible for 20 days of vacation per year, accrued monthly.
 ```
 
 You can see that the bot responds correctly.
 
 ## Conclusion
 
-In this guide, you learned how to connect a NeMo Guardrails configuration to an NVIDIA API Catalog LLM model. This guide uses `meta/llama3-70b-instruct`, however, you can connect any other model by following the same steps.
+In this guide, you learned how to connect a NeMo Guardrails configuration to an NVIDIA API Catalog LLM model. This guide uses `meta/llama-3.1-70b-instruct`, however, you can connect any other model by following the same steps.