adjust tokenizer config
Browse files- tokenization_aria.py +13 -15
- tokenizer_config.json +2 -2
tokenization_aria.py
CHANGED
|
@@ -32,16 +32,14 @@ class AriaTokenizer(PreTrainedTokenizer):
|
|
| 32 |
|
| 33 |
def __init__(
|
| 34 |
self,
|
| 35 |
-
add_bos_token=True,
|
| 36 |
add_eos_token=True,
|
| 37 |
-
add_dim_token=
|
| 38 |
clean_up_tokenization_spaces=False,
|
| 39 |
use_default_system_prompt=False,
|
| 40 |
**kwargs,
|
| 41 |
):
|
| 42 |
self._tokenizer = AbsTokenizer()
|
| 43 |
|
| 44 |
-
self.add_bos_token = add_bos_token
|
| 45 |
self.add_eos_token = add_eos_token
|
| 46 |
self.add_dim_token = add_dim_token
|
| 47 |
self.use_default_system_prompt = use_default_system_prompt
|
|
@@ -77,35 +75,35 @@ class AriaTokenizer(PreTrainedTokenizer):
|
|
| 77 |
def tokenize(
|
| 78 |
self,
|
| 79 |
midi_dict: MidiDict,
|
| 80 |
-
|
| 81 |
-
|
| 82 |
**kwargs,
|
| 83 |
) -> List[Token]:
|
| 84 |
return self._tokenizer.tokenize(
|
| 85 |
midi_dict=midi_dict,
|
| 86 |
add_dim_tok=(
|
| 87 |
-
|
|
|
|
|
|
|
| 88 |
),
|
| 89 |
add_eos_tok=(
|
| 90 |
-
|
|
|
|
|
|
|
| 91 |
),
|
| 92 |
)
|
| 93 |
|
| 94 |
def _tokenize(
|
| 95 |
self,
|
| 96 |
midi_dict: MidiDict,
|
| 97 |
-
|
| 98 |
-
|
| 99 |
**kwargs,
|
| 100 |
) -> List[Token]:
|
| 101 |
return self._tokenizer.tokenize(
|
| 102 |
midi_dict=midi_dict,
|
| 103 |
-
add_dim_tok=
|
| 104 |
-
|
| 105 |
-
),
|
| 106 |
-
add_eos_tok=(
|
| 107 |
-
add_eos_tok if add_eos_tok is not None else self.add_eos_token
|
| 108 |
-
),
|
| 109 |
)
|
| 110 |
|
| 111 |
def __call__(
|
|
|
|
| 32 |
|
| 33 |
def __init__(
|
| 34 |
self,
|
|
|
|
| 35 |
add_eos_token=True,
|
| 36 |
+
add_dim_token=False,
|
| 37 |
clean_up_tokenization_spaces=False,
|
| 38 |
use_default_system_prompt=False,
|
| 39 |
**kwargs,
|
| 40 |
):
|
| 41 |
self._tokenizer = AbsTokenizer()
|
| 42 |
|
|
|
|
| 43 |
self.add_eos_token = add_eos_token
|
| 44 |
self.add_dim_token = add_dim_token
|
| 45 |
self.use_default_system_prompt = use_default_system_prompt
|
|
|
|
| 75 |
def tokenize(
|
| 76 |
self,
|
| 77 |
midi_dict: MidiDict,
|
| 78 |
+
add_dim_token: Optional[bool] = None,
|
| 79 |
+
add_eos_token: Optional[bool] = None,
|
| 80 |
**kwargs,
|
| 81 |
) -> List[Token]:
|
| 82 |
return self._tokenizer.tokenize(
|
| 83 |
midi_dict=midi_dict,
|
| 84 |
add_dim_tok=(
|
| 85 |
+
add_dim_token
|
| 86 |
+
if add_dim_token is not None
|
| 87 |
+
else self.add_dim_token
|
| 88 |
),
|
| 89 |
add_eos_tok=(
|
| 90 |
+
add_eos_token
|
| 91 |
+
if add_eos_token is not None
|
| 92 |
+
else self.add_eos_token
|
| 93 |
),
|
| 94 |
)
|
| 95 |
|
| 96 |
def _tokenize(
|
| 97 |
self,
|
| 98 |
midi_dict: MidiDict,
|
| 99 |
+
add_dim_token: Optional[bool] = None,
|
| 100 |
+
add_eos_token: Optional[bool] = None,
|
| 101 |
**kwargs,
|
| 102 |
) -> List[Token]:
|
| 103 |
return self._tokenizer.tokenize(
|
| 104 |
midi_dict=midi_dict,
|
| 105 |
+
add_dim_tok=add_dim_token,
|
| 106 |
+
add_eos_tok=add_eos_token,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
)
|
| 108 |
|
| 109 |
def __call__(
|
tokenizer_config.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
"auto_map": {
|
| 5 |
"AutoTokenizer": [
|
| 6 |
"tokenization_aria.AriaTokenizer",
|
|
|
|
| 1 |
{
|
| 2 |
+
"add_eos_token": true,
|
| 3 |
+
"add_dim_token": false,
|
| 4 |
"auto_map": {
|
| 5 |
"AutoTokenizer": [
|
| 6 |
"tokenization_aria.AriaTokenizer",
|