v0.0.27

What's Changed

• [docs] Update mcp cookbook with training of smaller generator model to avoid OOM by @nerdai in #424
• [docs] add link to paper by @nerdai in #425
• [chore] Add docstrings for exceptions module by @nerdai in #427
• [docs] update paper link to point to camera-ready version by @nerdai in #429
• [Feature] Add AsyncQdrantKnowledgeStore and cookbook for using it within an AsyncRAGSystem by @nerdai in #430
• [chore] - Adding few items to api reference in docs by @nerdai in #433
• [Feature] Add batch methods for RAGSystem by @czakop in #431
• Bump the all-python-packages group with 6 updates by @dependabot in #434
• [chore] rm non working codeql by @nerdai in #435
• [Feature] Add batch_retrieve to knowledge store classes by @nerdai in #436
• [Feature] Implement batch_retrieve for Qdrant knowledge stores by @nerdai in #439
• Update default batch_retrieve for RAG system classes by @nerdai in #441
• [version] bump version to v0.0.27 by @nerdai in #442

Full Changelog: v0.0.26...v0.0.27

v0.0.26

What's Changed

• [version] bump to version v0.0.25.post1 by @nerdai in #421
• Bump the all-python-packages group with 5 updates by @dependabot in #422
• Batching and cosine changes by @amasin2111 in #366
• [version] bump version to v0.0.26 by @nerdai in #423

Full Changelog: v0.0.25...v0.0.26

v0.0.25.post1

What's Changed

• Bump the all-python-packages group across 1 directory with 4 updates by @dependabot in #407
• [docs] integrations page update by @nerdai in #408
• [docs] Add docs for LangChain Bridge by @czakop in #409
• [Feature] add complete method to BaseGenerator by @nerdai in #410
• [Feature] Add notebook utils process monitor by @nerdai in #412
• [fix] ProcessMonitor.monitor_live to not loop through keys by @nerdai in #413
• [feature] Update monitor_live to work on a list of processes and break if all have stopped by @nerdai in #414
• [docs] Add cookbook for basic federated fine-tuning of RAG system by @nerdai in #415
• [chore] Add fl notebook to docs by @nerdai in #416
• [chore] Update citation in README.md by @nerdai in #418
• Improve import time for fed_rag by lazy loading heavy modules and classes by @nerdai in #419
• [version] bump version to v0.0.25 by @nerdai in #420
• [version] bump to version v0.0.25.post1 by @nerdai in #421

Full Changelog: v0.0.24...v0.0.25.post1

v0.0.24

What's Changed

• [feature] Add NoEncodeBaseKnowledgeStore and NoEncodeAsyncBaseKnowledgeStore by @nerdai in #385
• [Feature] Adds NoEncodeRAGSystem by @nerdai in #386
• [Feature] Verify bridge framework version by @czakop in #381
• [Feature] Add AsyncNoEncodeRAGSystem and AsyncRAGSystem by @nerdai in #392
• Feature/364 onboard more benchmarks by @Izukimat in #391
• [docs fix] Incorrect filename for docs/api_reference/benchmarks/natural_questions by @nerdai in #394
• [fix] natural questions benchmark module naming by @nerdai in #395
• [Feature] Add MCPKnowledgeStore (a NoEncode Knowledge Store) by @nerdai in #393
• [Feature] Add MCPStdioKnowledgeSource and rename MCPKnowledgeSource to MCPStreamableHttpKnowledgeSource by @nerdai in #399
• [refactor] Add BaseMCPKnowledgeSource to be more dry with MCPStdioKnowledgeSource and MCPStreamableHttpKnowledgeSoruce by @nerdai in #400
• [feature] Add rerank_callback to MCPKnowledgeStore by @nerdai in #402
• [docs] New cookbook for an AsyncNoEncodeRAGSystem using MCPKnowledgeStore by @nerdai in #401
• [Feature] Add to_sync() methods for AsyncKnowledgeStore and AsyncRAGSystem and their NoEncode variants. by @nerdai in #404
• [Feature] Add LangChain bridge by @czakop in #397
• [version] bump version to v0.0.24 by @nerdai in #406

Full Changelog: v0.0.23...v0.0.24

v0.0.23

What's Changed

• [Feature] Add ability to save benchmark evaluations to a JSONL file by @nerdai in #377
• [Feature] Add BaseAsyncKnowledgeBase (#262) by @czakop in #376
• [version] bump version to v0.0.23 by @nerdai in #378
• Revert "v0.0.22 (#378)" by @nerdai in #379
• [version] bump version to v0.0.23 by @nerdai in #380

Full Changelog: v0.0.22...v0.0.23

v0.0.22

What's Changed

• [chore] docs uncomment installation instructions in unsloth cookbook by @nerdai in #362
• Passing kwargs for trainer by @amasin2111 in #365
• [Refactor] Simplify HF generator implementations by @czakop in #369
• [Fix] Make HF generators return newly generated tokens only by @czakop in #370
• Feature/348 onboard pubmedqa by @Izukimat in #363
• Update README.md by @nerdai in #371
• Update README.md by @nerdai in #372
• Bump the all-python-packages group with 4 updates by @dependabot in #373
• [version] bump version to v0.0.22 by @nerdai in #375

New Contributors

• @amasin2111 made their first contribution in #365
• @czakop made their first contribution in #369
• @Izukimat made their first contribution in #363

Full Changelog: v0.0.21...v0.0.22

v0.0.21

What's Changed

• [fix] bug for mismatch types in UnslothFastModelGenerator._get_peft_model() between LoRA adapters and base model by @nerdai in #360
• [version] bump version to v0.0.21 by @nerdai in #361

Full Changelog: v0.0.20...v0.0.21

v0.0.20

What's Changed

• Make knowledge node embedding optional and remove redundant embedding in Qdrant upserts by @nerdai in #353
• [chore] fix link for quick starts benchmarking RAG by @nerdai in #355
• [Feature] Add UnslothFastModelGenerator 🦥 by @nerdai in #356
• [Feature] Support get_peft_model in UnslothFastModelGenerator by @nerdai in #357
• [version] bump version to v0.0.20 by @nerdai in #358

Full Changelog: v0.0.19...v0.0.20

v0.0.19

What's Changed

• [docs] Small typo in qdrant example .md by @nerdai in #324
• [Feature] Add base.evals module and BaseBenchmark BaseBenchmarker classes by @nerdai in #326
• [Feature] Add ExactMatchEvaluationMetric by @nerdai in #333
• Bump datasets from 3.1.0 to 3.6.0 in the all-python-packages group by @dependabot in #335
• [Feature] Add HuggingFaceBenchmarkMixin and first HuggingFaceMMLU benchmark by @nerdai in #334
• [Feature] Enable streaming for BaseBenchmark with .as_stream() and implement with HF mixin by @nerdai in #337
• Add EvalsError and docs update by @nerdai in #339
• [Fix] Fix codification of response in HuggingFaceMMLU benchmark. by @nerdai in #340
• [fix] fix num_total_examples in Benchmarker by @nerdai in #342
• [fix] Add num_examples abstract property to BaseBenchmark by @nerdai in #344
• [docs] Notebook and quick start guide for running benchmarks by @nerdai in #343
• [chore] Add example benchmarking nb to docs by @nerdai in #347
• [docs] Add Qdrant example notebook by @nerdai in #349
• [Feature] Add support for in-memory instances of Qdrant for QdrantKnowledgeStore by @nerdai in #350
• [chore] Set WITH_DOCKER to True in qdrant cookbook by @nerdai in #351
• [version] bump version to v0.0.19 by @nerdai in #352

Full Changelog: v0.0.18...v0.0.19

v0.0.18

What's Changed

• [refactor] types to data_structures by @nerdai in #319
• Add trainer managers as a root import by @nerdai in #320
• Add fed_rag.trainers submodule public api by @nerdai in #321
• [version] bump version to v0.0.18 by @nerdai in #322

Full Changelog: v0.0.17...v0.0.18