Update config.json
Browse files- config.json +6 -10
config.json
CHANGED
|
@@ -1,3 +1,4 @@
|
|
|
|
|
| 1 |
{
|
| 2 |
"_name_or_path": "mbart-large-classical-chinese-after-deepl-300seq-shorter-training/checkpoint-140000",
|
| 3 |
"_num_labels": 3,
|
|
@@ -50,19 +51,14 @@
|
|
| 50 |
"pad_token_id": 1,
|
| 51 |
"scale_embedding": true,
|
| 52 |
"static_position_embeddings": false,
|
| 53 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
"tokenizer_class": "MBart50Tokenizer",
|
| 55 |
"torch_dtype": "float32",
|
| 56 |
"transformers_version": "4.21.0.dev0",
|
| 57 |
"use_cache": true,
|
| 58 |
"vocab_size": 250054
|
| 59 |
-
"task_specific_params": {
|
| 60 |
-
"translation_ch_to_en": {
|
| 61 |
-
"early_stopping": true,
|
| 62 |
-
"max_length": 300,
|
| 63 |
-
"num_beams": 5,
|
| 64 |
-
"length_penalty": 1.5,
|
| 65 |
-
"repetition_penalty" : 2.0
|
| 66 |
-
},
|
| 67 |
-
},
|
| 68 |
}
|
|
|
|
| 1 |
+
|
| 2 |
{
|
| 3 |
"_name_or_path": "mbart-large-classical-chinese-after-deepl-300seq-shorter-training/checkpoint-140000",
|
| 4 |
"_num_labels": 3,
|
|
|
|
| 51 |
"pad_token_id": 1,
|
| 52 |
"scale_embedding": true,
|
| 53 |
"static_position_embeddings": false,
|
| 54 |
+
"task_specific_params": {
|
| 55 |
+
"translation_en_to_ro": {
|
| 56 |
+
"decoder_start_token_id": 250020
|
| 57 |
+
}
|
| 58 |
+
},
|
| 59 |
"tokenizer_class": "MBart50Tokenizer",
|
| 60 |
"torch_dtype": "float32",
|
| 61 |
"transformers_version": "4.21.0.dev0",
|
| 62 |
"use_cache": true,
|
| 63 |
"vocab_size": 250054
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
}
|