Results of the Quality Estimation Shared Task 2022
Note: in each table, the "Rank" column indicates the averaged ranking of participants with regards to all metrics (including memory print and number of parameters).
Task 1 -- Sentence-level Direct Assessment
(
top)
Multilingual (Full) |
Multilingual (w/o English-Yoruba) |
|
|
| |
English-Czech |
English-Japanese |
|
|
| |
English-Marathi |
Khmer-English |
|
|
| |
Pashto-English |
|
| |
Task 1 -- Sentence-level MQM
(
top)
Multilingual |
English-German |
|
|
| |
English-Russian |
Chinese-English |
|
|
| |
Task 1 -- Word-level
(
top)
Multilingual (Full) |
Multilingual (w/o English-Yoruba) |
|
|
| |
English-Czech |
English-Japanese |
|
|
| |
English-Marathi |
Khmer-English |
|
|
| |
Pashto-English |
English-German |
|
|
| |
English-Russian |
Chinese-English |
|
|
| |
Task 2 -- Explainable QE
(
top)
English-Czech |
English-Japanese |
|
|
| |
English-Marathi |
English-Russian |
|
|
| |
English-German |
English-Yoruba |
|
|
| |
Khmer-English |
Pashto-English |
|
|
| |
Chinese-English |
|
| |
Task 3 -- Critical Error Detection
(
top)
English-German (Constrained) |
English-German (UNconstrained) |
|
|
| |
Portuguese-English (Constrained) |
Portuguese-English (UNconstrained) |
|
|
| |