Nemoguardrails

Changed

- [288](https://github.com/NVIDIA/NeMo-Guardrails/pull/288) Replace SentenceTransformers with FastEmbed.

Added

- [254](https://github.com/NVIDIA/NeMo-Guardrails/pull/254) Support for [Llama Guard input and output content moderation](./docs/user_guides/guardrails-library.md#llama-guard-based-content-moderation).
- [253](https://github.com/NVIDIA/NeMo-Guardrails/pull/253) Support for [server-side threads](./docs/user_guides/server-guide.md#threads).
- [235](https://github.com/NVIDIA/NeMo-Guardrails/pull/235) Improved [LangChain integration](docs/user_guides/langchain/langchain-integration.md) through `RunnableRails`.
- [190](https://github.com/NVIDIA/NeMo-Guardrails/pull/190) Add [example](./examples/notebooks/generate_events_and_streaming.ipynb) for using `generate_events_async` with streaming.
- Support for Python 3.11.

Changed

- [240](https://github.com/NVIDIA/NeMo-Guardrails/pull/240) Switch to pyproject.
- [276](https://github.com/NVIDIA/NeMo-Guardrails/pull/276) Upgraded Typer to 0.9.

Fixed

- [286](https://github.com/NVIDIA/NeMo-Guardrails/pull/286) Fixed not having the same evaluation set given a random seed for topical rails.
- [239](https://github.com/NVIDIA/NeMo-Guardrails/pull/239) Fixed logging issue where `verbose=true` flag did not trigger expected log output.
- [228](https://github.com/NVIDIA/NeMo-Guardrails/pull/228) Fix docstrings for various functions.
- [242](https://github.com/NVIDIA/NeMo-Guardrails/pull/242) Fix Azure LLM support.
- [225](https://github.com/NVIDIA/NeMo-Guardrails/pull/225) Fix annoy import, to allow using without.
- [209](https://github.com/NVIDIA/NeMo-Guardrails/pull/209) Fix user messages missing from prompt.
- [261](https://github.com/NVIDIA/NeMo-Guardrails/pull/261) Fix small bug in `print_llm_calls_summary`.
- [252](https://github.com/NVIDIA/NeMo-Guardrails/pull/252) Fixed duplicate loading for the default config.
- Fixed the dependencies pinning, allowing a wider range of dependencies versions.
- Fixed sever security issues related to uncontrolled data used in path expression and information exposure through an exception.

Added

- Support for `--version` flag in the CLI.

Changed

- Upgraded `langchain` to `0.0.352`.
- Upgraded `httpx` to `0.24.1`.
- Replaced deprecated `text-davinci-003` model with `gpt-3.5-turbo-instruct`.

Fixed

- [191](https://github.com/NVIDIA/NeMo-Guardrails/pull/191): Fix chat generation chunk issue.

Added

- Support for [explicit definition](./docs/user_guides/configuration-guide.mdguardrails-definitions) of input/output/retrieval rails.
- Support for [custom tasks and their prompts](docs/user_guides/advanced/prompt-customization.mdcustom-tasks-and-prompts).
- Support for fact-checking [using AlignScore](./docs/user_guides/guardrails-library.mdalignscore-based-fact-checking).
- Support for [NeMo LLM Service](./docs/user_guides/configuration-guide.mdnemo-llm-service) as an LLM provider.
- Support for making a single LLM call for both the guardrails process and generating the response (by setting `rails.dialog.single_call.enabled` to `True`).
- Support for [sensitive data detection](./docs/user_guides/guardrails-library.mdpresidio-based-sensitive-data-detection) guardrails using Presidio.
- [Example](./examples/configs/llm/hf_pipeline_llama2) using NeMo Guardrails with the LLaMa2-13B model.
- [Dockerfile](./Dockerfile) for building a Docker image.
- Support for [prompting modes](./docs/user_guides/advanced/prompt-customization.md) using `prompting_mode`.
- Support for [TRT-LLM](./docs/user_guides/configuration-guide.mdtrt-llm) as an LLM provider.
- Support for [streaming](./docs/user_guides/advanced/streaming.md) the LLM responses when no output rails are used.
- [Integration](./docs/user_guides/guardrails-library.mdactive-fence) of ActiveFence ActiveScore API as an input rail.
- Support for `--prefix` and `--auto-reload` in the [guardrails server](./docs/user_guides/server-guide.md).
- Example [authentication dialog flow](./examples/configs/auth).
- Example [RAG using Pinecone](./examples/configs/rag/pinecone).
- Support for loading a configuration from dictionary, i.e. `RailsConfig.from_content(config=...)`.
- Guidance on [LLM support](./docs/user_guides/llm-support.md).
- Support for `LLMRails.explain()` (see the [Getting Started](./docs/getting-started) guide for sample usage).

Changed

- Allow context data directly in the `/v1/chat/completion` using messages with the type `"role"`.
- Allow calling a subflow whose name is in a variable, e.g. `do $some_name`.
- Allow using actions which are not `async` functions.
- Disabled pretty exceptions in CLI.
- Upgraded dependencies.
- Updated the [Getting Started Guide](./docs/getting-started).
- Main [README](./README.md) now provides more details.
- Merged original examples into a single [ABC Bot](./examples/bots/abc) and removed the original ones.
- Documentation improvements.

Fixed

- Fix going over the maximum prompt length using the `max_length` attribute in [Prompt Templates](./docs/user_guides/advanced/prompt-customization.mdprompt-templates).
- Fixed problem with `nest_asyncio` initialization.
- [144](https://github.com/NVIDIA/NeMo-Guardrails/pull/144) Fixed TypeError in logging call.
- [121](https://github.com/NVIDIA/NeMo-Guardrails/pull/109) Detect chat model using openai engine.
- [109](https://github.com/NVIDIA/NeMo-Guardrails/pull/109) Fixed minor logging issue.
- Parallel flow support.
- Fix `HuggingFacePipeline` bug related to LangChain version upgrade.

Added

- Support for [custom configuration data](docs/user_guides/configuration-guide.mdcustom-data).
- Example for using [custom LLM and multiple KBs](examples/configs/rag/multi_kb/README.md)
- Support for [`PROMPTS_DIR`](docs/user_guides/advanced/prompt-customization.mdprompt-configuration).
- [101](https://github.com/NVIDIA/NeMo-Guardrails/pull/101) Support for [using OpenAI embeddings](docs/user_guides/configuration-guide.md#the-embeddings-model) models in addition to SentenceTransformers.
- First set of end-to-end QA tests for the example configurations.
- Support for configurable [embedding search providers](docs/user_guides/advanced/embedding-search-providers.md)

Changed

- Moved to using `nest_asyncio` for [implementing the blocking API](docs/user_guides/advanced/nested-async-loop.md). Fixes [3](https://github.com/NVIDIA/NeMo-Guardrails/issues/3) and [#32](https://github.com/NVIDIA/NeMo-Guardrails/issues/32).
- Improved event property validation in `new_event_dict`.
- Refactored imports to allow installing from source without Annoy/SentenceTransformers (would need a custom embedding search provider to work).

Fixed

- Fixed when the `init` function from `config.py` is called to allow custom LLM providers to be registered inside.
- [93](https://github.com/NVIDIA/NeMo-Guardrails/pull/93): Removed redundant `hasattr` check in `nemoguardrails/llm/params.py`.
- [91](https://github.com/NVIDIA/NeMo-Guardrails/issues/91): Fixed how default context variables are initialized.

Added

- [Event-based API](docs/user_guides/advanced/event-based-api.md) for guardrails.
- Support for message with type "event" in [`LLMRails.generate_async`](./docs/api/nemoguardrails.rails.llm.llmrails.mdmethod-llmrailsgenerate_async).
- Support for [bot message instructions](docs/user_guides/advanced/bot-message-instructions.md).
- Support for [using variables inside bot message definitions](docs/user_guides/colang-language-syntax-guide.mdbot-messages-with-variables).
- Support for `vicuna-7b-v1.3` and `mpt-7b-instruct`.
- Topical evaluation results for `vicuna-7b-v1.3` and `mpt-7b-instruct`.
- Support to use different models for different LLM tasks.
- Support for [red-teaming](docs/user_guides/advanced/red-teaming.md) using challenges.
- Support to disable the Chat UI when running the server using `--disable-chat-ui`.
- Support for accessing the API request headers in server mode.
- Support to [enable CORS settings](docs/user_guides/server-guide.mdcors) for the guardrails server.

Changed

- Changed the naming of the internal events to align to the upcoming UMIM spec (Unified Multimodal Interaction Management).
- If there are no user message examples, the bot messages examples lookup is disabled as well.

Fixed

- [58](https://github.com/NVIDIA/NeMo-Guardrails/issues/58): Fix install on Mac OS 13.
- [55](https://github.com/NVIDIA/NeMo-Guardrails/issues/55): Fix bug in example causing config.py to crash on computers with no CUDA-enabled GPUs.
- Fixed the model name initialization for LLMs that use the `model` kwarg.
- Fixed the Cohere prompt templates.
- [55](https://github.com/NVIDIA/NeMo-Guardrails/issues/83): Fix bug related to LangChain callbacks initialization.
- Fixed generation of "..." on value generation.
- Fixed the parameters type conversion when invoking actions from Colang (previously everything was string).
- Fixed `model_kwargs` property for the `WrapperLLM`.
- Fixed bug when `stop` was used inside flows.
- Fixed Chat UI bug when an invalid guardrails configuration was used.