 Dustin Miller
|
7b85bad267 |
Cleaned up Agent descriptions: consistent format for their beginning. |
9 年 前 |
 Andrew Cantino
|
eb52eeeeea |
experimental branch to support using fewer gems for saving RAM |
10 年 前 |
|
Andrew Cantino
|
2c820e8af2 |
allow separate events for each answer in the HumanTaskAgent |
10 年 前 |
|
Andrew Cantino
|
073df69df7 |
take dsander's advice and use interpolated instead of interpolated_options |
10 年 前 |
|
Andrew Cantino
|
a8f32c5e85 |
Add memoized interpolated_options and used it everywhere. |
10 年 前 |
 Dominik Sander
|
56324c87e2 |
Migrated HumanTask, Trigger and PeakDetectorAgent to Liquid |
10 年 前 |
|
Andrew Cantino
|
73e8406d4f |
Merge branch 'master' into do_not_symbolize |
11 年 前 |
|
Andrew Cantino
|
a408ae48ea |
globally avoid using symbols since we're moving to json storage |
11 年 前 |
|
Andrew Cantino
|
77882908c4 |
in order to avoid leaking memory, remove all symbolization |
11 年 前 |
|
Andrew Cantino
|
b32cef09c9 |
doc |
11 年 前 |
|
Andrew Cantino
|
0eb956e110 |
HumanTaskAgent can now accept `combination_mode` and `poll_options` which are used to generate a poll from the HITs answers. The poll asks additional Mechanical Turk users to rate the responses, and then the best one is returned. This allows for an additional step of validation on user generated content / data. |
11 年 前 |
|
Andrew Cantino
|
9fb53c4f64 |
silence a log |
11 年 前 |
|
Andrew Cantino
|
22e950214b |
cleanup HITs when they have been reviewed |
11 年 前 |
|
Andrew Cantino
|
c9b4d1b1f0 |
allow the HITs lifetime to be set |
11 年 前 |
|
Andrew Cantino
|
82156f3c75 |
move RTurk setup to an initializer |
11 年 前 |
|
Andrew Cantino
|
31cdf610ac |
Rename max_assignments to assignments; add lots of validations and tests |
11 年 前 |
|
Andrew Cantino
|
e284782363 |
Calculate average answers when all questions are numeric |
11 年 前 |
|
Andrew Cantino
|
1e8dcc65f0 |
Majority voting appears to work |
11 年 前 |
|
Andrew Cantino
|
0120ccb9c1 |
Basic HuamnTaskAgent with specs |
11 年 前 |
Dustin Miller
Andrew Cantino
Dominik Sander