-
Notifications
You must be signed in to change notification settings - Fork 24
Expand file tree
/
Copy pathevaluation_lib.py
More file actions
228 lines (191 loc) · 7.01 KB
/
evaluation_lib.py
File metadata and controls
228 lines (191 loc) · 7.01 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
# coding=utf-8
# Copyright 2025 The Google Research Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Binary of evaluating instruction following. See README.md."""
import collections
import dataclasses
import json
from typing import Dict, Optional, Union
import instructions_registry
@dataclasses.dataclass
class InputExample:
key: int
instruction_id_list: list[str]
prompt: str
kwargs: list[Dict[str, Optional[Union[str, int]]]]
@dataclasses.dataclass
class OutputExample:
instruction_id_list: list[str]
prompt: str
response: str
follow_all_instructions: bool
follow_instruction_list: list[bool]
def read_prompt_list(input_jsonl_filename):
"""Read inputs from jsonl."""
inputs = []
with open(input_jsonl_filename, "r") as f:
for l in f:
example = json.loads(l)
inputs.append(
InputExample(key=example["key"],
instruction_id_list=example["instruction_id_list"],
prompt=example["prompt"],
kwargs=example["kwargs"]))
return inputs
def write_outputs(output_jsonl_filename, outputs):
"""Writes outputs to jsonl."""
assert outputs
with open(output_jsonl_filename, "w") as f:
for o in outputs:
f.write(
json.dumps(
{
attr_name: o.__getattribute__(attr_name)
for attr_name in [
name for name in dir(o) if not name.startswith("_")
]
}
)
)
f.write("\n")
def test_instruction_following_strict(
inp,
prompt_to_response,
):
"""Tests response to see if instrutions are followed."""
response = prompt_to_response[inp.prompt]
instruction_list = inp.instruction_id_list
is_following_list = []
for index, instruction_id in enumerate(instruction_list):
instruction_cls = instructions_registry.INSTRUCTION_DICT[instruction_id]
instruction = instruction_cls(instruction_id)
inp.kwargs[index] = {key: value for key, value in inp.kwargs[index].items() if value is not None}
instruction.build_description(**inp.kwargs[index])
args = instruction.get_instruction_args()
if args and "prompt" in args:
instruction.build_description(prompt=inp.prompt)
if response and response.strip() and instruction.check_following(response):
is_following_list.append(True)
else:
is_following_list.append(False)
return OutputExample(
instruction_id_list=inp.instruction_id_list,
prompt=inp.prompt,
response=response,
follow_all_instructions=all(is_following_list),
follow_instruction_list=is_following_list,
)
def test_instruction_following_loose(
inp,
prompt_to_response,
):
"""Tests response for an upper bound for following instructions."""
response = prompt_to_response[inp.prompt]
if response is None:
return OutputExample(
instruction_id_list=inp.instruction_id_list,
prompt=inp.prompt,
response="",
follow_all_instructions=False,
follow_instruction_list=[False] * len(inp.instruction_id_list),
)
r = response.split("\n")
response_remove_first = "\n".join(r[1:]).strip()
response_remove_last = "\n".join(r[:-1]).strip()
response_remove_both = "\n".join(r[1:-1]).strip()
revised_response = response.replace("*", "")
revised_response_remove_first = response_remove_first.replace("*", "")
revised_response_remove_last = response_remove_last.replace("*", "")
revised_response_remove_both = response_remove_both.replace("*", "")
all_responses = [
response,
revised_response,
response_remove_first,
response_remove_last,
response_remove_both,
revised_response_remove_first,
revised_response_remove_last,
revised_response_remove_both,
]
instruction_list = inp.instruction_id_list
is_following_list = []
for index, instruction_id in enumerate(instruction_list):
instruction_cls = instructions_registry.INSTRUCTION_DICT[instruction_id]
instruction = instruction_cls(instruction_id)
instruction.build_description(**inp.kwargs[index])
args = instruction.get_instruction_args()
if args and "prompt" in args:
instruction.build_description(prompt=inp.prompt)
is_following = False
for r in all_responses:
if r.strip() and instruction.check_following(r):
is_following = True
break
is_following_list.append(is_following)
return OutputExample(
instruction_id_list=inp.instruction_id_list,
prompt=inp.prompt,
response=response,
follow_all_instructions=all(is_following_list),
follow_instruction_list=is_following_list,
)
def read_prompt_to_response_dict(input_jsonl_filename):
"""Creates dictionary matching prompt and response."""
return_dict = {}
with open(input_jsonl_filename, "r") as f:
for l in f:
example = json.loads(l)
return_dict[example["prompt"]] = example["response"]
return return_dict
def print_report(outputs):
"""Prints a report on accuracy scores."""
prompt_total = 0
prompt_correct = 0
instruction_total = 0
instruction_correct = 0
tier0_total = collections.defaultdict(int)
tier0_correct = collections.defaultdict(int)
tier1_total = collections.defaultdict(int)
tier1_correct = collections.defaultdict(int)
for example in outputs:
follow_instruction_list = example.follow_instruction_list
instruction_id_list = example.instruction_id_list
prompt_total += 1
if all(follow_instruction_list):
prompt_correct += 1
instruction_total += len(instruction_id_list)
instruction_correct += sum(follow_instruction_list)
for instruction_id, followed_or_not in zip(
instruction_id_list, follow_instruction_list
):
instruction_id = instruction_id.split(":")[0]
tier0_total[instruction_id] += 1
if followed_or_not:
tier0_correct[instruction_id] += 1
for instruction_id, followed_or_not in zip(
instruction_id_list, follow_instruction_list
):
tier1_total[instruction_id] += 1
if followed_or_not:
tier1_correct[instruction_id] += 1
print(f"prompt-level: {prompt_correct / prompt_total}")
print(f"instruction-level: {instruction_correct / instruction_total}")
print()
for instruction_id in sorted(tier0_total.keys()):
accuracy = tier0_correct[instruction_id] / tier0_total[instruction_id]
print(f"{instruction_id} {accuracy}")
print()
for instruction_id in sorted(tier1_total.keys()):
accuracy = tier1_correct[instruction_id] / tier1_total[instruction_id]
print(f"{instruction_id} {accuracy}")