Releases · kubeagi/arcadia

arcadia-v0.2.0

Welcome to this new release!

Images built for this release:

kubeagi/arcadia:v0.2.0
kubeagi/data-processing:v0.2.0

Breaking Changes:

None

Feature summary 🚀 🚀 🚀

For dataprocessing:

Add a new datasource Web and able to download,extract,process online url resources to enhanced QA
Deduplicate QA by similarity comparison on Q's embeddings

For model:

Support more models: mistral,qwen7b,qwen

For LLM application and agents:

Application/Agent evaluation with the help of ragas_once(a CLI of ragas) in following metrics: context_precision,context_recall,context_relevancy,answer_relevancy,answer_similarity,answer_correctness,faithfulness
Add a new CRD Agent to provides agent capabilities
Add two new tools: bing_search, weather query
Able to process documents(pdf,txts) during chat: generate embeddings and summary with a MapReduceChain
Add a new Chain ApiChain which call a open api during chat
Able to publish applications to gpt store

Changelog

New Features

0ce6076: feat: Add 1st version of agent support with a weather query tool (@nkwangleiGIT)
8514d7b: feat: Add api to support copying existing rag (@dayuy)
07a62de: feat: able to customize worker' images (@0xff-dev)
149fb59: feat: able to generate embeddings and summarizations for documents in a conversation (@bjwswang)
d3b3039: feat: able to set/update models in LLM and Embedder (@bjwswang)
c519fe5: feat: able to store chat document (@bjwswang)
540607d: feat: able to store chat document (@bjwswang)
27f9ef0: feat: able to use Web datasource in versioneddataset (@bjwswang)
e0cd235: feat: able to use openai as the embedder in knowledgebase (@bjwswang)
46f39ed: feat: add README.md for arcadia chart (@Lanture1064)
0703063: feat: add a command to download a file (@0xff-dev)
98923c3: feat: add a controller implementation of rag (@0xff-dev)
dc10c00: feat: add a new api to get rag evaluation report (@0xff-dev)
4b5e25f: feat: add a new column latency in rag eval dataset row (@0xff-dev)
89a65ae: feat: add a new mapreducechain in appruntime (@bjwswang)
181b3df: feat: add api and doc support to return to the download link (@xxxxibo)
0d6e83d: feat: add apichain (@Abirdcfly)
22ba5c4: feat: add bingsearch tools (@Abirdcfly)
5ec0900: feat: add chat latency and api: generate prompt starters (@Abirdcfly)
f7b12e2: feat: add creator for rag (@0xff-dev)
ba39369: feat: add gpt list (@Abirdcfly)
e3f036f: feat: add graphql definition of rag (@0xff-dev)
59c59c9: feat: add maxTokens in apiserver (@Abirdcfly)
fc79c1a: feat: add more info to model including MaxContextLength,HuggingfaceRepo,ModelscopeRepo (@bjwswang)
51a709b: feat: add new datasource Web (@bjwswang)
ac9e121: feat: add rag custom resource definition (@0xff-dev)
008ad77: feat: add ragas_once tool (@Lanture1064)
146af31: feat: add reocrd costtime (@0xff-dev)
736803e: feat: add some display config for app (@Abirdcfly)
9a4932e: feat: added watch on Embedder and VectorStore for KnowledgeBase reconciliation (@Abirdcfly)
6eeceee: feat: apiserver authentication support different groups (@0xff-dev)
09254c9: feat: arctl support merge files (@0xff-dev)
83e72f5: feat: bingsearch add more config (@Abirdcfly)
36cd8db: feat: change maxLength default value to 2048 (@Abirdcfly)
3dd25d3: feat: chat info store to pg (@Abirdcfly)
7c6ef7f: feat: generate evaluation test dataset (@bjwswang)
2d22e20: feat: graphql add api to create rag (@0xff-dev)
534b402: feat: graphql: add api to update rag (@dayuy)
199d0b0: feat: handle pod state Waiting in worker (@bjwswang)
b01bbdb: feat: knowledgebase support more granular update (@Abirdcfly)
4f4b17d: feat: namespace controller creates the rabc needed for the job (@0xff-dev)
dc88d12: feat: release v0.2.0 (@bjwswang)
c0ff091: feat: result & summary CSV file output (@Lanture1064)
804542d: feat: return users all conversation (@Abirdcfly)
cabcb45: feat: set to storage as a non-required field (@0xff-dev)
0dff685: feat: support for downloading data from versioneddatasets (@0xff-dev)
2377775: feat: support streaming and chain of agent (@nkwangleiGIT)
f234364: feat: support streaming in chat docs (@bjwswang)
461309d: feat: unify the implementation of list apis (@0xff-dev)
21f9aaa: feat: when a task fails, display the failure time (@0xff-dev)
a74b3d2: feat:QA remove duplicate (@)
9617971: feat:Real-time data access (@)
fe1190d: feat:Splitting website data (@wangxinbiao)
af1ee1b: feat:able to use openai in data-processing (@wangxinbiao)
af8f14c: feat:add dataprocessing task retry functionality (@wangxinbiao)
d79d56b: feat:add fields dataProcessMs and syncMsg for the data processing information and file synchronisation information respectively (@xxxxibo)
cd8550f: feat:deduplicate QA (@wangxinbiao)
77f5129: feat:details return deduplication configuration information (@wangxinbiao)
97ce84c: feat:support downloading files to specific directories (@xxxxibo)
49ed6bc: feat:when the task fails, update the 'message' information in the dataset for cr. (@wangxinbiao)

Bug Fixes

7156ee8: fix(application): add isReady check for all nodes in application (@Abirdcfly)
1ae5ff1: fix(appruntime): Add Cleanup() function to Node interface (@Abirdcfly)
a99b1b5: fix: Failed to create Versioneddataset with the inherited historical version (@dayuy)
d00c5bc: fix: able to use openai-compatible llms/embeddings in ragas_once (@bjwswang)
16dc3d5: fix: adapting ragas_once parameter changes (@0xff-dev)
e07201e: fix: add displayName in graphql TypedObjectReference (@bjwswang)
a2712ab: fix: add empty history for evaluation run (@Abirdcfly)
8230951: fix: add messageID for stream mode (@Abirdcfly)
5bf9bfc: fix: add new filed embeddderType in apiserver's knowledgebase model (@bjwswang)
8b31112: fix: add permissions for arctl (@0xff-dev)
3e12421: fix: apiserver needs access to storageclass (@0xff-dev)
51b4856: fix: application get more fields (@Abirdcfly)
875d096: fix: avoid the pointer to loop variables (@xxxxibo)
ba85408: fix: avoid to update resource status with explictly status check (@bjwswang)
bcf9cfe: fix: datasource and dataset are using the wrong bucket in command arctl (@0xff-dev)
9a70257: fix: delete the rag and clean up the minio files at the same time (@0xff-dev)
0fc1427: fix: doc score need update (@Abirdcfly)
602efc0: fix: gpt hot use storage interface (@Abirdcfly)
2699be2: fix: handle some errors in chat when using stream mode (@Abirdcfly)
d30a3ce: fix: init a start time when chat documents (@bjwswang)
48482eb: fix: langchaingo upstream update (@Abirdcfly)
b762946: fix: pass chain call options to chat_docs (@bjwswang)
ee61601: fix: pgvector return nothing when embedding table has vector with different dimensions (@Abirdcfly)
ca249e7: fix: prompt is using the wrong group (@0xff-dev)
fc94bf1: fix: prompt-starter api cant get data when use knowledgebase (@Abirdcfly)
1e7b2b8: fix: qachain set hi...