source-academy / backend / 9c2823cf899c9314fd4ccec9dd6c3b589d83e839 / 1
89%
master: 89%

DEFAULT BRANCH: master
Ran
Files 158
Run time 5s
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 88.716% (-0.9%) from 89.621%
9c2823cf899c9314fd4ccec9dd6c3b589d83e839.1

push

github

web-flow 
AI-powered marking (#1248)

* feat: v1 of AI-generated comments

* feat: added logging of inputs and outputs

* Update generate_ai_comments.ex

* feat: function to save outputs to database

* Format answers json before sending to LLM

* Add LLM Prompt to question params when submitting assessment xml file

* Add LLM Prompt to api response when grading view is open

* feat: added llm_prompt from qn to raw_prompt

* feat: enabling/disabling of LLM feature by course level

* feat: added llm_grading boolean field to course creation API

* feat: added api key storage in courses & edit api key/enable llm grading

* feat: encryption for llm_api_key

* feat: added final comment editing route

* feat: added logging of chosen comments

* fix: bugs when certain fields were missing

* feat: updated tests

* formatting

* fix: error handling when calling openai API

* fix: credo issues

* formatting

* Address some comments

* Fix formatting

* rm IO.inspect

* a

* Use case instead of if

* Streamlines generate_ai_comments to only send the selected question and its relevant info + use the correct llm_prompt

* Remove unncessary field

* default: false for llm_grading

* Add proper linking between ai_comments table and submissions. Return it to submission retrieval as well

* Resolve some migration comments

* Add llm_model and llm_api_url to the DB + schema

* Moves api key, api url, llm model and course prompt to course level

* Add encryption_key to env

* Do not hardcode formatting instructions

* Add Assessment level prompts to the XML

* Return some additional info for composing of prompts

* Remove un-used 'save comments'

* Fix existing assessment tests

* Fix generate_ai_comments test cases

* Fix bug preventing avengers from generating ai comments

* Fix up tests + error msgs

* Formatting

* some mix credo suggestions

* format

* Fix credo issue

* bug fix + credo fixes

* Fix tests

* format

* Modify test.exs

* Update lib/cadet_web/controllers/gener... (continued)

3758 of 4236 relevant lines covered (88.72%)

7103.93 hits per line

Source Files on job 9c2823cf899c9314fd4ccec9dd6c3b589d83e839.1