Commit Graph

11880 Commits

Author SHA1 Message Date
49317bbee4 Merge pull request #5625 from martin-g/faster-search-hybrid-it-tests
tests: Faster search::hybrid IT tests
2025-06-03 13:54:38 +00:00
82313a4444 Cargo fmt 2025-06-03 15:39:26 +02:00
8fdcdee0cc Do a first clippy pass 2025-06-03 15:39:26 +02:00
3c218cc3a0 Update the default chat completions prompt
Co-authored-by: Martin Grigorov <martin-g@users.noreply.github.com>
2025-06-03 15:39:26 +02:00
7d574433b6 Clean up chat completions modules a bit 2025-06-03 15:39:26 +02:00
201a808fe2 Better report errors happening with the underlying LLM 2025-06-03 15:39:26 +02:00
f827c2442c Mark tool calls to be implemented later for non-streaming 2025-06-03 15:36:35 +02:00
87d2e213f3 Update chat keys 2025-06-03 15:36:35 +02:00
3b931e75d9 Make the chats settings and chat completions route experimental 2025-06-03 15:36:35 +02:00
ae135d1d46 Implement a first version of a streamed chat API 2025-06-03 15:36:35 +02:00
0efb72fe66 Introduce the first version of the /chat route that mimics the OpenAI API 2025-06-03 15:36:35 +02:00
bed442528f Update charabia v0.9.4 2025-06-03 15:31:28 +02:00
496685fa26 Implement deserr on ChatCompletions settings structs 2025-06-03 15:31:28 +02:00
02cbcea3db Better chat completions settings management 2025-06-03 15:31:28 +02:00
0f7f5fa104 Introduce listing/getting/deleting/updating chat workspace settings 2025-06-03 15:31:28 +02:00
50fafbbc8b Implement useful conversion strategies and clean up the code 2025-06-03 15:31:28 +02:00
2821163b95 Clean up the code a bit 2025-06-03 15:31:27 +02:00
2da64e835e Factorize the code a bit more and support reporting errors 2025-06-03 15:31:27 +02:00
420c6e1932 Report the sources 2025-06-03 15:31:27 +02:00
2a067d3327 Fix compilation error in test 2025-06-03 15:31:27 +02:00
564cad1163 Call specific tools to show progression and results. 2025-06-03 15:31:27 +02:00
33dfd422db Introduce a lot of search parameters and make Deserr happy 2025-06-03 15:31:27 +02:00
036a9d5dbc Expose a well defined set of sources 2025-06-03 15:31:26 +02:00
7b74810b03 Add the index descriptions to the function description 2025-06-03 15:31:26 +02:00
3e53527bff redact the chat settings API key 2025-06-03 15:31:26 +02:00
7929872091 Better chat settings management 2025-06-03 15:31:26 +02:00
afb43d266e Correctly list the chat settings key actions 2025-06-03 15:31:26 +02:00
05828ff2c7 Always use the frequency matching strategy 2025-06-03 15:31:26 +02:00
75c3f33478 Correctly support document templates on the chat API 2025-06-03 15:31:25 +02:00
c6930c8819 Introduce the new index chat settings 2025-06-03 15:31:25 +02:00
439146289e Make sure errorneous calls are handled and forwarded to the LLM 2025-06-03 15:31:25 +02:00
6bf214bb14 Catch invalid argument calls to search function 2025-06-03 15:31:25 +02:00
fcf694026d Support multiple indexes and not only main 2025-06-03 15:31:25 +02:00
0b675bd530 Limit the number of internal loop calls and change the function name 2025-06-03 15:31:25 +02:00
7636365a65 Correctly support tenant tokens and filters 2025-06-03 15:31:24 +02:00
46680585ae Stream errors 2025-06-03 15:31:24 +02:00
bcec8d8984 Stop the stream when the connexion stops and chnage the events 2025-06-03 15:31:24 +02:00
56c1bd3afe Generate a new default chat API key 2025-06-03 15:31:24 +02:00
1a84f00fbf Change the /chat route to /chat/completions to be OpenAI-compatible 2025-06-03 15:31:24 +02:00
39320a6fce Better stop the stream 2025-06-03 15:31:24 +02:00
1d2dbcb51f Update the streaming detection to work with Mistral 2025-06-03 15:31:23 +02:00
341183cd57 Make it compatible with the Mistral API 2025-06-03 15:31:23 +02:00
b9716ec346 Support base_api in the settings 2025-06-03 15:31:03 +02:00
564f85280c Make clippy happy 2025-06-03 15:31:03 +02:00
7fa74b4931 Display pre-query prompt in search tool response 2025-06-03 15:31:03 +02:00
7d8415448c Commit when putting stuff in LMDB 2025-06-03 15:31:03 +02:00
c7839b5a84 Remove useless function 2025-06-03 15:31:03 +02:00
a52b513023 Expose new chat settings routes 2025-06-03 15:31:02 +02:00
77e03e3f8c Factorise a bit the code 2025-06-03 15:31:02 +02:00
148816a3da Display the different tool calls we need to do 2025-06-03 15:31:02 +02:00