NeMo-Guardrails
NeMo-Guardrails copied to clipboard
NVIDIA
NeMo Guardrails is an open-source toolkit for easily adding programmable guardrails to LLM-based conversational systems.
Hello, I'm currently using the latest version of the Nemo Guardrails 0.9.0 with a local LLM (Llama2) on Google Colab with an A100 GPU. I am encountering an issue where...
hi i am not able to connect my llamaguard api with nemoguardrail . below are the configs i have used . please help !!! config : ```yaml models: - type:...
Ln : 368 to 372 of colang/v1_0/runtime/runtime.py ```python if status == "failed": # TODO: make this message configurable. result = self._internal_error_action_result( "I'm sorry, an internal error has occurred." ) ```...
Tried to see if there were any questions related to my issue in this github forum but could not locate one. My question is, if anybody who works for an...
Looks like GenerateFlowFromName is creating a flow with a single quote that trips up parsing. ```log GenerateFlowFromNameActionFinished {"action_uid": "396a...", "action_name": "GenerateFlowFromNameAction", "status": "success", "is_success": true, "return_value": "flow bot execute location_api\n...
Team, Cleanlab team integrated `trustworthiness_score` as part of 3rd party API that helps identify hallucinations. It has been a month since the PR is merged to `develop` branch, followed by...
PR adding support for a .railsignore to ignore .co files in LLMRails / RailsConfig. For issue: https://github.com/NVIDIA/NeMo-Guardrails/issues/277 Tests added and passing fully. Two follow ups after this PR are: 1....
Hi, I wonder how can we load the config in dynamic way. In my case, the colang file and actions.py file stay the same among model, but I can't find...
- Added skip_user_message parameter to get_retrieved_relevant_chunks function. - Updated relevant_chunks assignment to handle cases with no user flow.
