Releases · talkdai/dialog

What's Changed

Adds front-end to docker-compose by @vmesel in #4
Fixes database-url by @vmesel in #10
Fix/adds new questions only by @vmesel in #12
docker organize file structure by @avelino in #13
force the prompt not to say you're a model/ai by @avelino in #15
Adds timestamp through custom memory class by @vmesel in #17
llm/prompt: carregar header sub categoria by @avelino in #23
Categories conversation history in background by @avelino in #25
Cast function to load project config from toml by @walison17 in #27
parent rule when adding msg (reply) from llm by @avelino in #26
adding category and subcategory checking on PK by @vmesel in #29
Refactor chat memory usage by @walison17 in #39
Fixes load_data script by @vmesel in #41
Adds censorship to requests that don't match our message embeddings by @vmesel in #36
fixed bug: fallback prompt, does not exist by @avelino in #45
docker image by @avelino in #43
Adds virtualenvs.create as False - installing on root Python by @vmesel in #50
current stop words from nltk by @ArthurMor4is in #57
Update Makefiles by @lgabs in #56
fix file name in command by @ArthurMor4is in #58
add temperature option to toml by @lgabs in #59
Allow setting model_name in the prompt's TOML by @cosmastech in #51
add simple answer time logging by @lgabs in #62
remove duplicated line by @lgabs in #61
memory/llm: optional append in prompt by @avelino in #65
Bump aiohttp from 3.8.6 to 3.9.0 by @dependabot in #70
devcontainer support by @avelino in #68
refactor llm configs by @lgabs in #66
Whatsapp Stable Consumption by @vmesel in #71
Pinned python in the versions >=3.11 by @leogregianin in #75
Improvement load csv by @leogregianin in #78
Modularized LLM model and changes from PR #76 by @vmesel in #80
Removes webhooks from the main project by @vmesel in #82
Support multiple datasets in a single load_csv script by @vmesel in #86
Fix run.sh conditional sintax by @walison17 in #94
Fix PLUGINS default config by @walison17 in #92
Customizable CORS config by @walison17 in #95
Fix Dockerfile by @walison17 in #88
fixed load csv if empty db by @avelino in #100
Bump fastapi from 0.104.1 to 0.109.1 by @dependabot in #101
Bump the pip group across 1 directories with 1 update by @dependabot in #103
upgrade langchain to 0.1.7 by @avelino in #104
External plugins by @vmesel in #107
Adds dataset information, fallback on extension creation and fix OpenAI Embeddings by @vmesel in #109
Allow custom session identifier by @walison17 in #110
Update README.md by @vmesel in #111
Dont run docker build job on forks by @walison17 in #112
settings: llm config via config by @avelino in #113
Copy static folder by @walison17 in #121
add samples for langsmith logging by @lgabs in #123
Change document retrieval to cosine distances by @schumannc in #116
Make session_id optional by @schumannc in #115
Fix and tweak db service healthcheck by @walison17 in #129
Fix install extension pgvector via migration by @walison17 in #130
fix typo in COSINE_SIMILARITY_THRESHOLD by @schumannc in #127
Fix pandas warning by @lgabs in #132
Use pgvector image from pgvector org on dockerhub by @walison17 in #125
fix session creation by @lgabs in #137
fix fallback usage by @lgabs in #139
Better plugins installation by @walison17 in #142
load_csv: clear all table rows by @avelino in #141
not having a recommended question in the know base by @avelino in #144
load_csv/feat: support multi csv field by @avelino in #147
concat cols correctly for embedding generation by @lgabs in #153
Moves models to another file, adds description to views and adds basic LCEL support by @vmesel in #148