Model options allow configuration of the specific model parameters that are available in Feedback Aide Models.
Different models and different rubric types will accept different options and silently ignore options that not relevant to it.
In the JavaScript API, the model options should be added when initializing a feedback session with the feedbackApp.feedbackSession() method. If using the RESTFul API, it can be passed via the options parameter on the evaluate endpoints.
Language
language configuration options for outputs from models.
| Property | Details |
|
(string) |
Set the language which feedback will be returned in. Default: |
Languages are input as 2 letter ISO 639-1 codes.
Currently supported languages for feedback are:
-
enoren-usEnglish US -
fr: French -
es: Spanish -
pt: Portuguese -
it: Italian -
de: German -
zh: Chinese -
id: Indonesian -
jp: Japanese -
kr: Korean
Preview support languages
These are allowed options, but have not been tested and results may vary in quality based on language.
-
ar: Arabic -
en-gb: English with United Kingdom spelling -
fi: Finnish -
sv: Swedish -
no: Norwegian -
da: Danish -
nl: Dutch -
pl: Polish -
pt-br: Portuguese (Brazil)
See also the article on Internationalization for Feedback aide to customize the interface elements.
Grading
grading configuration options to provide context to the model about the learner level and the marking style required for this evaluation.
| Property | Details |
|
(string) |
Learner level - provides hints to the model about the expected level of the learner response which can improve accuracy
Default: |
|
(string) |
Default: |
Available on standard-essay-2025-03-14 and advanced-essay-2025-03-14 and newer models.
Thinking
thinking configuration options for enabling the model to return an explanation of its decisions.
| Property | Details |
|
(boolean) |
Enable or disable the return of the models combined 'thinking'. In its current form this is a raw unprocessed return. Default: |
Available on 2025-03-14 and newer models.
Moderation (Essay only)
moderation configuration options for moderation - see Content moderation with Feedback Aide
| Property | Details |
|
(boolean) |
Enable or disable content moderation checking. Default: |
Note: Moderation is available on advanced-essay models only.
Topic Checking (Essay only)
topic_check Check if the essay is on a topic related to the question.
| Property | Details |
|
(boolean) |
Enable or disable topic checking. Default: |
|
(integer) |
Relatedness level from Note: setting this to Default: |
Available on 2024-11-22 and newer models.
Structure Checking (Essay only)
structure_check Check if the essay matches the required structure - checks for the minimum number of paragraphs.
| Property | Details |
|
(boolean) |
Enable or disable structure checking. Default: |
|
(integer) |
Check the minimum number of paragraphs. If there are less than this number then it will be flagged as Default: |
|
(integer) |
Set the minimum number of words required for a 'paragraph' to be counted as a paragraph. Default: |
|
(string) |
Mode of checking paragraphs.
Default: |
Available on 2024-11-22 and newer models.
Citation Checking (Essay only)
citation_check Check if the essay has the required number of citations.
| Property | Details |
|
(boolean) |
Enable or disable citation checking. Default: |
|
(integer) |
Check the number of in-text citations. If there are less than this number then it will be flagged as Default: |
|
(integer) |
Check the number of citations in the bibliography / works cited section. If there are less than this number then it will be flagged as Default: |
Available on 2025-03-14 and newer models.