v24.03.00

🚨 Breaking Changes

• Updating nlohman_json to 3.11 to match MRC (#1596) @mdemoret-nv
• Add retry logic and proxy support to the NeMo LLM Service (#1544) @mdemoret-nv
• Upgrade openai version to 1.13 and langchain to version 0.1.9 (#1529) @mdemoret-nv
• Make start_async() available to source stages (#1523) @efajardo-nv
• RAPIDS 24.02 Upgrade (#1468) @cwharris
• Decouple TritonInferenceStage from pipeline mode (#1402) @dagardner-nv

🐛 Bug Fixes

• Serialize datetime objects into the module config (#1592) @dagardner-nv
• Remove the defaults channel from dependencies.yml (#1584) @mdemoret-nv
• Fix iso_date_regex_pattern config in file_batcher module and allow override (#1580) @efajardo-nv
• Update DFP MLflow ModelManager to handle model retrieval using file URI (#1578) @efajardo-nv
• Fix configure_logging in DFP benchmarks (#1553) @efajardo-nv
• Catch langchain agent errors (#1539) @dagardner-nv
• Adding missing dependency on pydantic (#1535) @yuchenz427
• Fix memory leak in the mutable dataframe checkout/checkin code (#1534) @dagardner-nv
• Fix pathlib.Path support for FileSourceStage (#1531) @dagardner-nv
• Make start_async() available to source stages (#1523) @efajardo-nv
• Update CI Containers (#1521) @cwharris
• Fix intermittent segfault on interpreter shutdown (#1513) @dagardner-nv
• Adopt updated builds of CI runners (#1503) @dagardner-nv
• Update mlflow plugin version for deployments fix (#1499) @pdmack
• Add runtime environment output to fix building the release container (#1496) @cwharris
• Fix logging of sleep time (#1493) @dagardner-nv
• Pin pytest to <8 (#1485) @dagardner-nv
• Improve pipeline stop logic to ensure join is called exactly once for all stages (#1479) @efajardo-nv
• Fix expected JSON config file extension in logger (#1471) @efajardo-nv
• Fix Loss Function to Improve Model Convergence for AutoEncoder (#1460) @hsin-c
• GNN fraud detection notebook fix (#1450) @efajardo-nv
• Eliminate Redundant Fetches in RSS Controller (#1442) @bsuryadevara
• Updating the workspace settings to remove deprecated python options (#1440) @mdemoret-nv
• Improve camouflage startup issues (#1436) @dagardner-nv
• Fixes to modular DFP examples and benchmarks (#1429) @efajardo-nv

📖 Documentation

• Update minimum compute requirements to Volta (#1594) @dagardner-nv
• Fix broken link in getting started with Morpheus doc (#1494) @edknv
• Update abp-model-card.md (#1439) @drobison00
• Update gnn-fsi-model-card.md (#1438) @drobison00
• Update phishing-model-card.md (#1437) @drobison00
• Document incompatible mlflow models issue (#1434) @dagardner-nv

🚀 New Features

• Adding retry logic to the TritonInferenceStage to allow recovering from errors (#1548) @cwharris
• Create a base mixin class for ingress & egress stages (#1473) @dagardner-nv
• RAPIDS 24.02 Upgrade (#1468) @cwharris
• Install headers & morpheus-config.cmake (#1448) @dagardner-nv

🛠️ Improvements

• Updating nlohman_json to 3.11 to match MRC (#1596) @mdemoret-nv
• DOCA 2.6 from public repo (#1588) @e-ago
• Support ControlMessage for PreProcessNLPStage PreProcessFILStage AddScoreStageBase (#1573) @yuchenz427
• Update MLflow in Production DFP example to use Python 3.10 (#1572) @efajardo-nv
• Fix environment yaml paths (#1551) @efajardo-nv
• Add retry logic and proxy support to the NeMo LLM Service (#1544) @mdemoret-nv
• Update to match new MRC function sig for AsyncioRunnable::on_data (#1541) @dagardner-nv
• Expose max_retries parameter to OpenAIChatService & OpenAIChatClient (#1536) @dagardner-nv
• Upgrade openai version to 1.13 and langchain to version 0.1.9 (#1529) @mdemoret-nv
• Update ops-bot.yaml (#1528) @AyodeAwe
• Add the ability to attach Tensor objects and timestamps to ControlMessage (#1511) @drobison00
• Fix or silence warnings emitted during tests (#1501) @dagardner-nv
• Support ControlMessage output in the C++ impl of DeserializeStage (#1478) @dagardner-nv
• DOCA Source Stage improvements (#1475) @e-ago
• Update copyright headers for 2024 (#1474) @efajardo-nv
• Add conda builds to CI (#1466) @dagardner-nv
• Grafana log monitoring and error alerting example (#1463) @efajardo-nv
• Misc Conda Improvements (#1462) @dagardner-nv
• Simplification of the streaming RAG ingest example to improve usability (#1454) @drobison00
• Replace GPUtil with pynvml for benchmark reports (#1451) @efajardo-nv
• Misc test improvements (#1447) @dagardner-nv
• Add a --manual_seed flag to the CLI (#1445) @dagardner-nv
• Optionally skip ci based on a label in the pr (#1444) @dagardner-nv
• Refactor verification of optional dependencies (#1443) @dagardner-nv
• Use dependencies.yaml as source-of-truth for environment files. (#1441) @cwharris
• Add mocked test & benchmark for LLM agents pipeline (#1424) @dagardner-nv
• Add benchmarks for stand-alone RAG & vdb upload pipelines (#1421) @dagardner-nv
• Add benchmark for completion pipeline (#1414) @dagardner-nv
• Decouple TritonInferenceStage from pipeline mode (#1402) @dagardner-nv

Full Changelog: v23.11.01...v24.03.00