Add project configuration manifest
Browse files- config.json +37 -0
config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"project": "Pixel Translation",
|
| 3 |
+
"task": "arabic-to-english-translation",
|
| 4 |
+
"model_families": [
|
| 5 |
+
"base",
|
| 6 |
+
"bytes",
|
| 7 |
+
"pixel",
|
| 8 |
+
"vocab"
|
| 9 |
+
],
|
| 10 |
+
"pretraining": {
|
| 11 |
+
"dataset": "msa",
|
| 12 |
+
"fractions": [
|
| 13 |
+
"5pct",
|
| 14 |
+
"10pct",
|
| 15 |
+
"25pct",
|
| 16 |
+
"50pct",
|
| 17 |
+
"full"
|
| 18 |
+
]
|
| 19 |
+
},
|
| 20 |
+
"finetuning": {
|
| 21 |
+
"curriculum": {
|
| 22 |
+
"fractions": [
|
| 23 |
+
"5pct",
|
| 24 |
+
"10pct",
|
| 25 |
+
"25pct",
|
| 26 |
+
"50pct"
|
| 27 |
+
],
|
| 28 |
+
"mixture": "msa_dialect_1x"
|
| 29 |
+
},
|
| 30 |
+
"full_data_variants": [
|
| 31 |
+
"dialect",
|
| 32 |
+
"msa_dialect_1x",
|
| 33 |
+
"msa_dialect_2x",
|
| 34 |
+
"msa_dialect_4x"
|
| 35 |
+
]
|
| 36 |
+
}
|
| 37 |
+
}
|