-
Notifications
You must be signed in to change notification settings - Fork 0
/
model.py
343 lines (240 loc) · 10.1 KB
/
model.py
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
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
import util
import openai
import time
class EvaluateLLM():
def __init__(self, model_name, temperature, max_tokens, api_key, method):
self.model_name = model_name
self.llm = None
if(self.model_name.startswith("gpt-")):
self.llm = openai.OpenAI(base_url= "https://api.zhiyunai168.com/v1",
api_key = api_key)
self.temperature = temperature
self.method = method
self.max_tokens = max_tokens
def generate(self, model_name, prompt, system_message=None):
response = None
if(model_name.startswith("gpt-")):
while True:
try:
response = self.llm.chat.completions.create(
messages=[
{
"role": "system",
"content": system_message
},
{
"role": "user",
"content": prompt
}
],
model=model_name, temperature=self.temperature,
max_tokens=self.max_tokens, n=1
)
break
except (openai.APIError, openai.RateLimitError) as e:
print("Error: {}".format(e))
time.sleep(2)
continue
response = response.choices[0].message.content.strip()
return response
def predict_answer(self, story, question, task_type):
predicted_answer = None
#find personX
personX = util.find_personX(question, task_type)
if(task_type == 'forward_belief'):
#prompt percept
percept_message = util.percept_prompt.format(story=story, personX=personX,
percept_definitional_constraint=util.constraints["percept_definitional_constraint"].format(personX=personX),
causal_event_definitional_constraint=util.constraints["causal_event_definitional_constraint"])
percept_response = self.generate(model_name=self.model_name,
prompt=percept_message,
system_message=util.percept_instruction)
print('percept_response: \n')
print(percept_response)
print('\n')
percept_response = util.parse_chat_response(percept_response)
print('parsed percept_response: \n')
print(percept_response)
print('\n')
#answer forward belief question
forward_belief_task_message = util.forward_belief_task_prompt.format(story=story, personX=personX,
question=question, belief_definitional_constraint=util.constraints['belief_definitional_constraint'].format(personX=personX),
belief_dependency_constraint=util.constraints['belief_dependency_constraint'].format(personX=personX),
causal_event_definitional_constraint=util.constraints["causal_event_definitional_constraint"],
percept_response=percept_response)
predicted_answer = self.generate(model_name=self.model_name,
prompt=forward_belief_task_message,
system_message=util.tom_task_instruction)
print('predicted_answer: \n')
print(predicted_answer)
print('\n')
predicted_answer = util.parse_chat_response(predicted_answer)
print('parsed predicted_answer: \n')
print(predicted_answer)
print('\n')
elif(task_type == 'forward_action'):
#prompt percept
percept_message = util.percept_prompt.format(story=story, personX=personX,
percept_definitional_constraint=util.constraints["percept_definitional_constraint"].format(personX=personX),
causal_event_definitional_constraint=util.constraints["causal_event_definitional_constraint"])
percept_response = self.generate(model_name=self.model_name,
prompt=percept_message,
system_message=util.percept_instruction)
print('percept_response: \n')
print(percept_response)
print('\n')
percept_response = util.parse_chat_response(percept_response)
print('parsed percept_response: \n')
print(percept_response)
print('\n')
#prompt belief
belief_message = util.belief_prompt.format(story=story,personX=personX, belief_definitional_constraint=\
util.constraints["belief_definitional_constraint"].format(personX=personX),
belief_dependency_constraint=util.constraints['belief_dependency_constraint'].format(personX=personX),
causal_event_definitional_constraint=util.constraints["causal_event_definitional_constraint"],
percept_response=percept_response)
belief_response = self.generate(model_name=self.model_name,
prompt=belief_message,
system_message=util.belief_instruction)
print('belief_response: \n')
print(belief_response)
print('\n')
belief_response = util.parse_chat_response(belief_response)
print('parsed belief_response: \n')
print(belief_response)
print('\n')
#prompt desire
desire_message = util.desire_prompt.format(story=story, personX=personX,
desire_definitional_constraint=util.constraints["desire_definitional_constraint"].format(personX=personX))
desire_response = self.generate(model_name=self.model_name,
prompt=desire_message,
system_message=util.desire_instruction)
print('desire_response: \n')
print(desire_response)
print('\n')
desire_response = util.parse_chat_response(desire_response)
print('parsed desire_response: \n')
print(desire_response)
print('\n')
#answer forward action question
forward_action_task_message = util.forward_action_task_prompt.format(story=story, personX=personX,
question=question, action_definitional_constraint_2=util.constraints['action_definitional_constraint_2'].format(personX=personX),
causal_event_definitional_constraint=util.constraints['causal_event_definitional_constraint'],
belief_response=belief_response, desire_response=desire_response,
action_dependency_constraint=util.constraints['action_dependency_constraint'].format(personX=personX))
predicted_answer = self.generate(model_name=self.model_name,
prompt=forward_action_task_message,
system_message=util.tom_task_instruction)
print('predicted_answer: \n')
print(predicted_answer)
print('\n')
predicted_answer = util.parse_chat_response(predicted_answer)
print('parsed predicted_answer: \n')
print(predicted_answer)
print('\n')
elif(task_type == 'backward_belief'):
#prompt desire
desire_message = util.desire_prompt.format(story=story, personX=personX,
desire_definitional_constraint=util.constraints["desire_definitional_constraint"].format(personX=personX))
desire_response = self.generate(model_name=self.model_name,
prompt=desire_message,
system_message=util.desire_instruction)
print('desire_response: \n')
print(desire_response)
print('\n')
desire_response = util.parse_chat_response(desire_response)
print('parsed desire_response: \n')
print(desire_response)
print('\n')
#prompt action
action_message = util.action_prompt_2.format(story=story, personX=personX,
action_definitional_constraint_1=util.constraints["action_definitional_constraint_1"].format(personX=personX),
causal_event_definitional_constraint=util.constraints["causal_event_definitional_constraint"])
action_response = self.generate(model_name=self.model_name,
prompt=action_message,
system_message=util.action_instruction_2)
print('action_response: \n')
print(action_response)
print('\n')
action_response = util.parse_chat_response(action_response)
print('parsed action_response: \n')
print(action_response)
print('\n')
#answer backward belief question
backward_belief_task_message = util.backward_belief_task_prompt.format(story=story, personX=personX,
question=question, belief_definitional_constraint=\
util.constraints['belief_definitional_constraint'].format(personX=personX),
desire_response=desire_response, action_response=action_response,
action_dependency_constraint=util.constraints['action_dependency_constraint'].format(personX=personX))
predicted_answer = self.generate(model_name=self.model_name,
prompt=backward_belief_task_message,
system_message=util.tom_task_instruction)
print('predicted_answer: \n')
print(predicted_answer)
print('\n')
predicted_answer = util.parse_chat_response(predicted_answer)
print('parsed predicted_answer: \n')
print(predicted_answer)
print('\n')
else:
print('Unsupported task type!')
sys.exit(1)
return predicted_answer
def grade_answer(self, query, predicted_answer, true_answer, wrong_answer):
prompt = util.grade_prompt.format(query=query, wrong_answer=wrong_answer, predicted_answer=predicted_answer, true_answer=true_answer)
response = self.generate(model_name=self.model_name,
prompt=prompt, system_message=util.grade_instruction)
response = util.parse_chat_response(response)
if(response[-1] == '.'): response = response[:-1]
return response
class EvaluateLLM_cot():
def __init__(self, model_name, temperature, max_tokens, api_key, method):
self.model_name = model_name
self.llm = None
if(self.model_name.startswith("gpt-")):
self.llm = openai.OpenAI(base_url= "https://api.zhiyungpt.com/v1",
api_key = api_key)
self.temperature = temperature
self.method = method
self.max_tokens = max_tokens
def generate(self, model_name, prompt, system_message=None):
response = None
if(model_name.startswith("gpt-")):
while True:
try:
response = self.llm.chat.completions.create(
messages=[
{
"role": "system",
"content": system_message
},
{
"role": "user",
"content": prompt
}
],
model=model_name, temperature=self.temperature,
max_tokens=self.max_tokens, n=1
)
break
except (openai.APIError, openai.RateLimitError) as e:
print("Error: {}".format(e))
time.sleep(2)
continue
response = response.choices[0].message.content.strip()
return response
def predict_answer(self, story, question, task_type):
predicted_answer = None
prompt = """Story: {story}\nQuestion: {question}""".format(
story=story, question=question)
predicted_answer = self.generate(model_name=self.model_name,
prompt=prompt,
system_message=util.cot_instruction)
print('predicted_answer: \n')
print(predicted_answer)
print('\n')
predicted_answer = util.parse_chat_response(predicted_answer)
print('parsed predicted_answer: \n')
print(predicted_answer)
print('\n')
return predicted_answer