Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | [Minor] C&P issue in last commit | alex081512 | 2025-03-14 | 1 | -2/+2 |
| | |||||
* | [Minor] Reverted commit for result history | alex081512 | 2025-03-14 | 1 | -9/+2 |
| | |||||
* | [Minor] Remove whitespace for lua check | alex081512 | 2025-03-13 | 1 | -2/+2 |
| | |||||
* | [Minor] Insert Reason to result to show gpt answer in history | alex081512 | 2025-03-13 | 1 | -2/+10 |
| | |||||
* | [Fix] Add header with reason everytime (not only for ham) and use correct ↵ | alex081512 | 2025-03-13 | 1 | -4/+4 |
| | | | | value for header | ||||
* | [Minor] Improve gpt prompt - Only ask for red flag when it's spam | alex081512 | 2025-03-13 | 1 | -2/+2 |
| | |||||
* | [Fix] initialize ollama result table | Jens Erat | 2025-03-13 | 1 | -0/+4 |
| | | | | | | similar to gpt.lua:704 for the chatgpt code path, the ollama code path needs to initialize the result table. Without, rspamd fails gpt requests with > lua_http_finish_handler: callback call failed: /usr/share/rspamd/plugins/gpt.lua:740: attempt to index a nil value | ||||
* | [Minor] Fix invocationvstakhov-cache-gpt | Vsevolod Stakhov | 2025-03-06 | 1 | -1/+1 |
| | |||||
* | [Rework] GPT: Use cache framework | Vsevolod Stakhov | 2025-03-06 | 1 | -44/+18 |
| | |||||
* | [Feature] Allow additional categories to be defined in GPTvstakhov-gpt-tunes2 | Vsevolod Stakhov | 2025-02-25 | 1 | -32/+108 |
| | |||||
* | [Minor] Tune caching | Vsevolod Stakhov | 2025-02-25 | 1 | -4/+16 |
| | |||||
* | [Minor] Tune caching | Vsevolod Stakhov | 2025-02-25 | 1 | -5/+7 |
| | |||||
* | [Feature] Cache LLM replies | Vsevolod Stakhov | 2025-02-25 | 1 | -49/+124 |
| | |||||
* | [Feature] Improve prompt and use plaintext instead of JSON | Vsevolod Stakhov | 2025-02-25 | 1 | -11/+104 |
| | |||||
* | [Minor] GPT: Allow to trigger on specific symbols | Vsevolod Stakhov | 2025-02-24 | 1 | -15/+36 |
| | |||||
* | [Minor] Allow to exclude all symbolsvstakhov-gpt-reason | Vsevolod Stakhov | 2025-02-01 | 1 | -1/+5 |
| | |||||
* | [Feature] GPT: Support reason adding | Vsevolod Stakhov | 2025-01-31 | 1 | -7/+32 |
| | |||||
* | [Feature] Support LLM models consensusvstakhov-gpt-consensus | Vsevolod Stakhov | 2025-01-27 | 1 | -91/+145 |
| | |||||
* | The gpt4all api which is openai compatible does not support response_format ↵ | John Quaglieri | 2025-01-02 | 1 | -3/+13 |
| | | | | | | | | | | | | and defaults to json_objec. This change adds the ability to add include_response_format = false; In the gpt.conf local.d settings and exclude sending response_format. No other changes are needed for gpt4all api fixing the error: gpt.lua:365: got reply: {"error":{"code":null,"message":"'response_format' is not supported","param":null,"type":"invalid_request_error"}} | ||||
* | [Feature] GPT: Add ollama supportvstakhov-gpt-ollama | Vsevolod Stakhov | 2024-12-18 | 1 | -31/+223 |
| | |||||
* | [Feature] GPT: Try harder to find JSON in NN reply | Vsevolod Stakhov | 2024-12-10 | 1 | -0/+37 |
| | |||||
* | [Minor] GPT: use lua_mime routine to select text part | Vsevolod Stakhov | 2024-12-10 | 1 | -13/+2 |
| | |||||
* | [Minor] GPT: add `allow_passthrough` and `allow_ham` settings | Vsevolod Stakhov | 2024-12-10 | 1 | -2/+8 |
| | |||||
* | [Fix] GPT: Fix bug in condition check | Vsevolod Stakhov | 2024-07-25 | 1 | -2/+2 |
| | |||||
* | [Minor] Use gpt-4o-mini by default | Vsevolod Stakhov | 2024-07-22 | 1 | -2/+2 |
| | |||||
* | [Minor] Improve prompt and add some conversion heursitics | Vsevolod Stakhov | 2024-07-22 | 1 | -2/+14 |
| | |||||
* | [Minor] Ensure some safety when checking weights | Vsevolod Stakhov | 2024-07-22 | 1 | -3/+5 |
| | | | | Issue: #5065 | ||||
* | [Minor] Set response_format | Vsevolod Stakhov | 2024-07-22 | 1 | -0/+1 |
| | |||||
* | [Minor] Remove top_p reduce temperature to 0 | Vsevolod Stakhov | 2024-07-22 | 1 | -6/+2 |
| | |||||
* | [Feature] Allow to specify minimum weight in GPT plugin | Vsevolod Stakhov | 2024-07-08 | 1 | -9/+20 |
| | |||||
* | [Project] Trying to test various things with GPT | Vsevolod Stakhov | 2024-07-01 | 1 | -9/+36 |
| | |||||
* | [Minor] Add some missing symbols | Vsevolod Stakhov | 2024-06-29 | 1 | -0/+2 |
| | |||||
* | [Minor] Add from and subject to the prompt | Vsevolod Stakhov | 2024-06-29 | 1 | -0/+9 |
| | |||||
* | [Minor] Allow autolearn | Vsevolod Stakhov | 2024-06-28 | 1 | -1/+6 |
| | |||||
* | [Minor] Apply review suggestion | Vsevolod Stakhov | 2024-06-28 | 1 | -5/+1 |
| | |||||
* | [Minor] Make except symbols configurable | Vsevolod Stakhov | 2024-06-28 | 1 | -10/+11 |
| | |||||
* | Update src/plugins/lua/gpt.lua | Vsevolod Stakhov | 2024-06-28 | 1 | -1/+1 |
| | | | Co-authored-by: Andrew Lewis <nerf@judo.za.org> | ||||
* | [Minor] Check words count properly | Vsevolod Stakhov | 2024-06-28 | 1 | -7/+9 |
| | |||||
* | [Project] Add checks to decide if we need a GPT check | Vsevolod Stakhov | 2024-06-28 | 1 | -24/+88 |
| | |||||
* | [Minor] Also use gzip compression | Vsevolod Stakhov | 2024-06-28 | 1 | -0/+1 |
| | |||||
* | [Minor] Use keep-alive and upstreams logic | Vsevolod Stakhov | 2024-06-28 | 1 | -1/+13 |
| | |||||
* | [Minor] Copy&paste fix | Vsevolod Stakhov | 2024-06-27 | 1 | -2/+2 |
| | |||||
* | [Project] Add GPT plugin | Vsevolod Stakhov | 2024-06-27 | 1 | -0/+256 |