Update model
Browse files- config.json +2 -52
- configuration_greedy.py +11 -9
config.json
CHANGED
|
@@ -8,58 +8,8 @@
|
|
| 8 |
},
|
| 9 |
"eos_token_id": 8,
|
| 10 |
"pad_token_id": 9,
|
| 11 |
-
"reciprocals":
|
| 12 |
-
|
| 13 |
-
4,
|
| 14 |
-
3
|
| 15 |
-
],
|
| 16 |
-
[
|
| 17 |
-
5,
|
| 18 |
-
2
|
| 19 |
-
],
|
| 20 |
-
[
|
| 21 |
-
6,
|
| 22 |
-
1
|
| 23 |
-
]
|
| 24 |
-
],
|
| 25 |
-
"reducables": [
|
| 26 |
-
[
|
| 27 |
-
[
|
| 28 |
-
4
|
| 29 |
-
],
|
| 30 |
-
[
|
| 31 |
-
3
|
| 32 |
-
]
|
| 33 |
-
],
|
| 34 |
-
[
|
| 35 |
-
[
|
| 36 |
-
5
|
| 37 |
-
],
|
| 38 |
-
[
|
| 39 |
-
2
|
| 40 |
-
]
|
| 41 |
-
],
|
| 42 |
-
[
|
| 43 |
-
[
|
| 44 |
-
6
|
| 45 |
-
],
|
| 46 |
-
[
|
| 47 |
-
1
|
| 48 |
-
]
|
| 49 |
-
],
|
| 50 |
-
[
|
| 51 |
-
[
|
| 52 |
-
4,
|
| 53 |
-
5,
|
| 54 |
-
6
|
| 55 |
-
],
|
| 56 |
-
[
|
| 57 |
-
1,
|
| 58 |
-
2,
|
| 59 |
-
3
|
| 60 |
-
]
|
| 61 |
-
]
|
| 62 |
-
],
|
| 63 |
"torch_dtype": "float32",
|
| 64 |
"transformers_version": "4.21.1",
|
| 65 |
"vocab_size": 10
|
|
|
|
| 8 |
},
|
| 9 |
"eos_token_id": 8,
|
| 10 |
"pad_token_id": 9,
|
| 11 |
+
"reciprocals": null,
|
| 12 |
+
"reducables": null,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
"torch_dtype": "float32",
|
| 14 |
"transformers_version": "4.21.1",
|
| 15 |
"vocab_size": 10
|
configuration_greedy.py
CHANGED
|
@@ -5,6 +5,16 @@ class GreedyConfig(PretrainedConfig):
|
|
| 5 |
|
| 6 |
@classmethod
|
| 7 |
def from_tokenizer(cls, freegroup_dimension, tokenizer: PreTrainedTokenizerBase, **kwargs):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
freegroup_generators = list(range(1, freegroup_dimension + 1))
|
| 10 |
|
|
@@ -17,15 +27,7 @@ class GreedyConfig(PretrainedConfig):
|
|
| 17 |
for reducable, closure_generator in zip(reducables, [[x] for x in freegroup_generators] + [freegroup_generators[::]]):
|
| 18 |
reducable.append(tokenizer.convert_tokens_to_ids(list(map(str, closure_generator))))
|
| 19 |
reducable.append(tokenizer.convert_tokens_to_ids(list(map(str, tools.reciprocal(closure_generator)))))
|
| 20 |
-
|
| 21 |
-
return cls(
|
| 22 |
-
reciprocals = reciprocals,
|
| 23 |
-
reducables = reducables,
|
| 24 |
-
vocab_size = len(tokenizer),
|
| 25 |
-
eos_token_id = tokenizer.eos_token_id,
|
| 26 |
-
pad_token_id = tokenizer.pad_token_id,
|
| 27 |
-
**kwargs
|
| 28 |
-
)
|
| 29 |
|
| 30 |
def __init__(self, **kwargs):
|
| 31 |
# reciporcals: List[List[int]]: i.e. ['x', 'X'], ...
|
|
|
|
| 5 |
|
| 6 |
@classmethod
|
| 7 |
def from_tokenizer(cls, freegroup_dimension, tokenizer: PreTrainedTokenizerBase, **kwargs):
|
| 8 |
+
config = cls(
|
| 9 |
+
vocab_size = len(tokenizer),
|
| 10 |
+
eos_token_id = tokenizer.eos_token_id,
|
| 11 |
+
pad_token_id = tokenizer.pad_token_id,
|
| 12 |
+
**kwargs
|
| 13 |
+
)
|
| 14 |
+
config._from_tokenizer(freegroup_dimension, tokenizer)
|
| 15 |
+
return config
|
| 16 |
+
|
| 17 |
+
def _from_tokenizer(self, freegroup_dimension, tokenizer):
|
| 18 |
|
| 19 |
freegroup_generators = list(range(1, freegroup_dimension + 1))
|
| 20 |
|
|
|
|
| 27 |
for reducable, closure_generator in zip(reducables, [[x] for x in freegroup_generators] + [freegroup_generators[::]]):
|
| 28 |
reducable.append(tokenizer.convert_tokens_to_ids(list(map(str, closure_generator))))
|
| 29 |
reducable.append(tokenizer.convert_tokens_to_ids(list(map(str, tools.reciprocal(closure_generator)))))
|
| 30 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
|
| 32 |
def __init__(self, **kwargs):
|
| 33 |
# reciporcals: List[List[int]]: i.e. ['x', 'X'], ...
|