v0.2.6

This is the final release (maybe). We do not have plan to improve RAGchain anymore.

What's Changed

• HotFix json linker fatal bug by @vkehfdl1 in #452
• Refactoring SearchQA benchmark. by @minsing-jin in #448
• Remove beir benchmark because it is useless. by @minsing-jin in #457
• Add Runnable Functionality to RAGchain modules by @vkehfdl1 in #442
• Transfer Ownership from NomaDamas to Marker-Inc by @vkehfdl1 in #458
• change insert_one() to insert_many() in mongo_db save function. by @bwook00 in #455
• Chunk data to put more than 25 elements at once in DynamoDB by @bwook00 in #460
• Feature/#450 by @vkehfdl1 in #461
• dump version V0.2.6 by @vkehfdl1 in #462

Full Changelog: v0.2.5...v0.2.6

v0.2.5

What's Changed

• Update README.md by @bwook00 in #427
• Fix error at WeightedTimeReranker by @vkehfdl1 in #433
• implement EM answer metric by @Eastsidegunn in #437
• [Hotfix] Json Linker can get db_path properly when use UUID id by @bwook00 in #430
• delete all load_dotenv in the project by @vkehfdl1 in #439
• [HotFix] EM name duplicated with retrieval_gt and answer_gt by @Eastsidegunn in #443
• dump version v0.2.5 by @vkehfdl1 in #444

Full Changelog: v0.2.4...v0.2.5

v0.2.4

What's Changed

• add new importance field at Passage by @vkehfdl1 in #410
• make SimpleImportanceReranker & WeightedImportanceReranker by @vkehfdl1 in #417
• Create google search feature and google search pipeline by @bwook00 in #419
• Create LLMLinguaCompressor for prompt compression by @vkehfdl1 in #420
• Create ELI5 benchmark by @minsing-jin in #414
• add rem loader for processing rem ingested texts by @vkehfdl1 in #423
• Create BingSearch by @bwook00 in #424
• dump version 0.2.4 by @vkehfdl1 in #425

Full Changelog: v0.2.3...v0.2.4

v0.2.3

What's Changed

• Create ASQA benchmark by @minsing-jin in #378
• Add content_datetime at Passage schema & filter Passage by time range by @vkehfdl1 in #381
• Feature/#385 & Feature/#382 by @vkehfdl1 in #386
• delete rerank_sliding_window at BaseReranker and keep that only in LL… by @vkehfdl1 in #391
• Make SimpleTimeReranker by @vkehfdl1 in #392
• Implement DSTC benchmark by @minsing-jin in #395
• add METEOR answer metric by @Eastsidegunn in #393
• add ROUGE answer metric by @Eastsidegunn in #397
• Create new linker(DynamoDB) and change linker to global variable(in init.py) by @bwook00 in #396
• make semantic clustering and time compressor using semantic clustering by @vkehfdl1 in #399
• Add local linker(Use JSON file) by @bwook00 in #403
• add WeightedTimeReranker by @vkehfdl1 in #404
• [Hotfix]Add jsonLinker selector + change class name in code comment by @bwook00 in #407
• dump version v0.2.3 by @vkehfdl1 in #408

Full Changelog: v0.2.2...v0.2.3

v0.2.2

What's Changed

• Create Trivia QA benchmark by @minsing-jin in #353
• Modify triviaQA benchmark by @minsing-jin in #363
• Create search_qa.py and test_search_qa.py. by @minsing-jin in #358
• Create nfcorpus benchmark by @minsing-jin in #369
• Create ANTIQUE benchmark by @minsing-jin in #366
• Fix PickleDB and test files for passing pytest at full run of test files. by @vkehfdl1 in #377
• Add hwp rust loader by @Hyunwook-Kwon in #282
• dump version v0.2.2 by @vkehfdl1 in #380

Full Changelog: v0.2.1...v0.2.2

v0.2.1

What's Changed

• Create DeepdoctectionPDFLoader by @bwook00 in #325
• remove no-package python files & update README.md by @vkehfdl1 in #337
• Modify mr_tydi.py some else cases. by @minsing-jin in #345
• Create beir datasets benchmarks by @minsing-jin in #333
• fix error when using context_recall metric but don't validate passages by @vkehfdl1 in #348
• Add answer gt at msmarco.py. by @minsing-jin in #349
• �New REDE Search Detector (I'm a few-shot learner~) by @vkehfdl1 in #351
• Create natural_qa.py and test file. by @minsing-jin in #350
• dump version v0.2.1 by @vkehfdl1 in #352

Full Changelog: v0.2.0...v0.2.1

v0.2.0

What's Changed

• new Pipeline structure & use LCEL instead of RAGchain LLM by @vkehfdl1 in #323
• fix missing parameter at default QueryDecomposition by @vkehfdl1 in #328
• dump version v0.2.0 by @vkehfdl1 in #329

Full Changelog: v0.1.5...v0.2.0

v0.1.5

What's Changed

• langchain and chromadb version update by @vkehfdl1 in #314
• Add BLEU score + refactor __calculate_metrics by @vkehfdl1 in #315
• import ragas when ragas metrics are used. by @vkehfdl1 in #317
• Add KF1 Metric by @vkehfdl1 in #318
• Fix error at retrieval gt calculation by @vkehfdl1 in #321
• Create Mr.tydi benchmark and test code. by @minsing-jin in #322
• dump version v0.1.5 by @vkehfdl1 in #324

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

• Hot fix/#303 by @vkehfdl1 in #304
• Create msmarco dataset evaluation code and test code. by @minsing-jin in #301
• Add code version optional and change msmarco test data to train data. by @minsing-jin in #308
• dump version V0.1.4 by @vkehfdl1 in #311

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• Create html_text_splitter by @minsing-jin in #285
• Create code splitter by @minsing-jin in #288
• delete deprecated evaluate files about ko-strategyQA by @vkehfdl1 in #289
• Create token_splitter and test file. by @minsing-jin in #290
• Remove inaccurate annotation. by @minsing-jin in #296
• add random_state option at qasper evaluator for constant benchmarking by @vkehfdl1 in #291
• independent work between LLM and retrieval by @vkehfdl1 in #292
• version update and fix langchain VectorStore location for proper use by @vkehfdl1 in #294
• dump version v0.1.3 by @vkehfdl1 in #297

Full Changelog: v0.1.2...v0.1.3