leon-ai/leon v1.0.0-beta.7

1.0.0-beta.7 / A Much Better NLP

on GitHub

Please read this blog post to know more about all the new features and the exciting future of Leon.

BREAKING CHANGES

• remove legacy packages 07743657

Features

• create new NLP skills resolvers model + NLP global resolvers model 602604e4

• better isolate skill resolvers from global resolvers + finish up Akinator skill 905d248e

• transfer language from core to skills + support thematics on Akinator skill b35a249b

• actions on slot level + akinator skill progress 7101b8b4

• add Cartesian sample training on resolvers + enum entities 6ed88a59

• map skills resolvers intents eb5ade76

• train skills resolvers and remap as per changes 82df0a3c

• achieve Cartesian training a1e9011d

• introduce suggestions dcddacca

• communicate suggestions to the client 4b5a8835

• shared skills memory 795acc5b

• support dynamic variables on skill NLU settings for logic type 10d10a16

• tmp resolvers mapping b1a332ba

• start to map resolvers between the core and skills e88495a9

• train affirmation and denial resolver 993d52e8

• Python TCP server and Node.js TCP client for IPC (wip) 5970ec9e

• introduce spaCy for complete NER (wip) caa86fc8

• slot filling (wip) 76547d94

• share data across domains f4f9fff9

• dynamic variable binding on NLG 0367b44f

• context and slot filling preparation (wip) 975b8ebc

• annotate entities on the fly + prepare for dialog skill type and cross-domains data 4107932d

• new NLP training d8023308

• server:

  • trigger next action suggestions or current ones 244d08c0
  • introduce main NLP model and resolvers NLP model e37526d9
  • change log emojis 843bc428
  • provide nextAction even when no slot is set and clean up NLU object on context switch 8377c63d
  • report full traceback from skills execution b69b1fea
  • support on-the-fly entity annotation for built-in entities 567b030c
  • save slots within conversation context fce47cdb
  • resolve resolvers tmp ceea47ff
  • prepare the next action on non-slot-filled skills 0acb31a9
  • add more affirmative utterance samples 870ab2e8
  • restart a skill with the original utterance saved in context f4446ef1
  • clean up context if the action loop does not meet the expected items 035c9d52
  • add handsigns custom entity 1529c720
  • reprocess NLU in case of slot filling interruption 9e242d77
  • handle action loop when slots have all been filled at once f8830502
  • break the action loop from the skill 27dc801c
  • stop action loop from skill to core 99681e25
  • introduce basic concept of action loop c5b38400
  • prepare action loop feature 19e1aa22
  • add current utterance entities to differentiate from the whole context 8b56a185
  • when a context is activated, pick up the most probable classification 8e186879
  • persist entities into contexts 87575773
  • forward slots to skill + add original utterance 68e40f65
  • handle case when all slots have been filled in one utterance 22e9234b
  • trigger next action once all slots have been filled 9b870010
  • complete slot filling before triggering the next action 9124687e
  • from modules to skills with type at the actions level 77ebaf4a
  • verify if all slots are filled e27c1b9c
  • context and slot filling, keep context and await for entities 25adf406
  • unstack oldest context 1ece25a4
  • context setup with slot for each conversation (wip) 8257eb87
  • resolve slots from slot filling 960a6dc7
  • pickup questions for slot filling 3bbc2f8a
  • main slots structure (wip) 1d9b1809
  • introduce resolvers for slot filling (wip) 334bf393
  • slot filling PoC (tmp wip) 95bfcfe4
  • slot filling (wip) 969a83e6
  • trigger unsupported language 1845eed7
  • context (wip) d1c2a11d
  • context (wip) a9a43ac4
  • differenciate cities from countries for location entities bf9bf231
  • auto restart the TCP server when language is switching 9be7c700
  • support multi languages on TCP server a808742c
  • add auto reconnect on TCP client cbe89ed6
  • make TCP client global 006e9fb0
  • fully implement low-level networking for IPC 8acb82da
  • more accurate NLG d5577b1e
  • unknown_answers fallback on dialog type 28efe6e7
  • deep data mapping on enum NER 3ca48265
  • NLG and entities mapping 8f2f935b
  • bootstrap skill structure fe90c68e
  • on-the-fly language switching f24513a2
  • new NLP containers 34b2aa56
  • (WIP) NLU refactoring ca3f5f42
  • add skills domains cf2a28aa

• skill/akinator:

  • finish up 79e7df02
  • finished main business logic 76cae42f
  • backbone 02a2f714

• skill/birthday:
  remove birthday skill be0b345d

• skill/color:
  introduce color skill ce00989b

• skill/guess_the_number:
  introduce the Guess the Number skill fba80966

• skill/introduction:

  • add one utterance sample af0fdd1e
  • ask about owner info if necessary c5cc9bdd

• skill/mbti:

  • complete form resolver aad9f3f1
  • finish business logic 99a3f103
  • questions mapping ae4f69f7
  • complete questionnaire 7f1f8871
  • main logic backbone 33109a4c
  • main NLU structure [skip ci] 86d5040a

• skill/rochambeau:

  • add start answers 192dd0a8
  • introduce paper scissors rock 57370470
  • init 7f5e30ac

• web app:

  • join us on Discord 141c89ec
  • wait for TCP client to be connected first bc228a68

  Bug Fixes

• check script with new intent-object format fdf0a389

• check new resolvers paths cfd8f7cb

• use ports.ubuntu.com mirror for the offline TTS [skip ci] 3dd90396

• set skill config only when a bridge is set 7513aa7d

• only set skill config when it is a logic type 9ce9a8bc

• docker:

  • usage of Ubuntu base image with pyenv and nvm (#408) f507f6f7
  • check should not allocate a pseudo-TTY (#359) 4372b45f

• server:

  • make leon handle multiple socket.io-client instances 6e7c0aac
  • fallback on global resolver during resolver classification ec77dd0f
  • make use of current entities to match global entities a8d82050
  • multiple slots filling 2ac1bc63
  • context switching on action loop 6712ae55
  • check one-shot slot filling case causing infinite loop 782a3aaa
  • clean up active context after all slots have been filled faabc2c7
  • correctly extract all spaCy entities 6aa60bfb
  • intent not found 8280c658
  • fallback due to modules to skills refactoring ef0c54b2
  • NER due to modules to skills refactoring e4d3904c
  • skill/akinator:
    remove direct end on guess action f6461f73
  • skill/color:
    more appropriate answer cb18ed63
  • skill/rochambeau:
    final logic 0ebc0518

  Performance Improvements

• check Pipfile instead of Pipfile.lock to judge whether Python packages must be installed afdb71f7

Documentation Changes

• change newsletter link 4bf2a9af
• README - Edited sentence for clarity (#389) e83a1c42
• edit newsletter link fa558a44
• update sponsor [skip ci] f30ddb6b
• remove sponsor [skip ci] 5dbc010f
• remove sponsor [skip ci] f36dd20f
• remove sponsor [skip ci] 5ee57ddf
• add @ant-media sponsor [skip ci] b47cbc3a
• add long dev notice to README [skip ci] 499be77d
• move sponsor to new section [skip ci] 8825d687
• change Twitter handle [skip ci] c1afc11c
• remove sponsor [skip ci] 99b401a6
• add new sponsor self-hosted img [skip ci] 238d928c
• remove new sponsor link (broken) [skip ci] 254f2848
• in GitHub BUG.md from modules to skills 4a5480a3
• change @FluxIndustries sponsorship [skip ci] 1a118b71
• add @FluxIndustries sponsor [skip ci] 9a604d7c
• new #LeonAI link [skip ci] a0107d62
• changelog new version diff link fix [skip ci] e14c2498