Jon Gauthier
commited on
Commit
·
2b3c98b
1
Parent(s):
3dad049
type hint improvements
Browse files- syntaxgym.py +11 -3
syntaxgym.py
CHANGED
@@ -10,7 +10,8 @@ from copy import deepcopy
|
|
10 |
import json
|
11 |
from pathlib import Path
|
12 |
import re
|
13 |
-
from typing import List, Tuple
|
|
|
14 |
|
15 |
import datasets
|
16 |
from datasets import logging
|
@@ -126,6 +127,11 @@ class SyntaxGym(datasets.GeneratorBasedBuilder):
|
|
126 |
yield item["item_number"], item
|
127 |
|
128 |
|
|
|
|
|
|
|
|
|
|
|
129 |
class SyntaxGymMetric(datasets.Metric):
|
130 |
"""
|
131 |
SyntaxGym prediction evaluation metric.
|
@@ -144,7 +150,8 @@ class SyntaxGymMetric(datasets.Metric):
|
|
144 |
)
|
145 |
|
146 |
def _compute(self, suite, model_id, batch_size: int = 16,
|
147 |
-
add_start_token=True, device=None
|
|
|
148 |
if device is not None:
|
149 |
assert device in ["gpu", "cpu", "cuda"]
|
150 |
if device == "gpu":
|
@@ -258,7 +265,8 @@ class SyntaxGymMetric(datasets.Metric):
|
|
258 |
return ret
|
259 |
|
260 |
def compute_region_token_mapping(self, item, input_ids: torch.LongTensor,
|
261 |
-
offset_mapping: List[Tuple[int, int]]
|
|
|
262 |
# input_ids: B * T
|
263 |
# offset_mapping: B * T * 2
|
264 |
# assumes batch is sorted according to item's condition_name order
|
|
|
10 |
import json
|
11 |
from pathlib import Path
|
12 |
import re
|
13 |
+
from typing import List, Dict, Tuple
|
14 |
+
from typing_extensions import TypedDict
|
15 |
|
16 |
import datasets
|
17 |
from datasets import logging
|
|
|
127 |
yield item["item_number"], item
|
128 |
|
129 |
|
130 |
+
class SyntaxGymMetricResult(TypedDict):
|
131 |
+
prediction_results: List[List[bool]]
|
132 |
+
region_totals: List[Dict[Tuple[str, int], float]]
|
133 |
+
|
134 |
+
|
135 |
class SyntaxGymMetric(datasets.Metric):
|
136 |
"""
|
137 |
SyntaxGym prediction evaluation metric.
|
|
|
150 |
)
|
151 |
|
152 |
def _compute(self, suite, model_id, batch_size: int = 16,
|
153 |
+
add_start_token=True, device=None
|
154 |
+
) -> SyntaxGymMetricResult:
|
155 |
if device is not None:
|
156 |
assert device in ["gpu", "cpu", "cuda"]
|
157 |
if device == "gpu":
|
|
|
265 |
return ret
|
266 |
|
267 |
def compute_region_token_mapping(self, item, input_ids: torch.LongTensor,
|
268 |
+
offset_mapping: List[Tuple[int, int]]
|
269 |
+
) -> Dict[str, Dict[int, List[int]]]:
|
270 |
# input_ids: B * T
|
271 |
# offset_mapping: B * T * 2
|
272 |
# assumes batch is sorted according to item's condition_name order
|