-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathtest_dialogue_react_agent.py
224 lines (174 loc) · 8.23 KB
/
test_dialogue_react_agent.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
import unittest
from unittest.mock import patch
from dialogue_react_agent import load_dialogue_react_fewshots, load_dialogue_act_taxonomy, DialogueReactAgent
from llm_engines import LLMApi
class TestDialogueReactAgent(unittest.TestCase):
def test_load_dialogue_react_fewshots(self):
file_path = "fewshots_mpc/generated_dialogue_reacts.jsonl"
prompts = load_dialogue_react_fewshots(file_path)
self.assertIsInstance(prompts, list)
self.assertGreater(len(prompts), 0)
def test_load_dialogue_act_taxonomy(self):
file_path = "prompts/dialogue_taxonomy.txt"
acts = load_dialogue_act_taxonomy(file_path)
self.assertIsInstance(acts, list)
self.assertGreater(len(acts), 0)
@patch('logging.basicConfig')
def test_main(self, mock_logging):
name = "John"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 0.5
reflections_freq = 0.3
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
self.assertEqual(agent.persona, persona)
self.assertEqual(agent.memory_freq, memory_freq)
self.assertEqual(agent.reflections_freq, reflections_freq)
self.assertEqual(agent.n_fewshots, n_fewshots)
def test_get_answer(self):
last_messages = ["Hello", "How are you?"]
extra_context = "Some extra context"
agent_list = ["Agent1", "Agent2"]
n_agents = 2
name = "John"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 0.5
reflections_freq = 0.3
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
answer = agent.get_answer(last_messages, extra_context=extra_context, agent_list=agent_list, n_agents=n_agents, turn_count=4)
self.assertIsInstance(answer, str)
self.assertNotEqual(answer, "")
# Add more assertions as needed
def test_gen_memories(self):
last_messages = [
(1, "Mark", "I like discussing about 18th century german philosophy."),
(2, "Alex", "Oh really? I really think that in his seminal work, Kant was really onto something."),
(3, "Mark", "I disagree. I think that Hegel's dialectical method is more relevant than ever.")
]
n_memories = 3
agent_list = ["Mark", "Alex"]
n_agents = 2
turn_count = 4
name = "Mark"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 0.5
reflections_freq = 0.3
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
memories = agent.gen_memories(last_messages, n_memories, agent_list=agent_list, n_agents=n_agents, turn_count=turn_count)
self.assertIsInstance(memories, list)
self.assertEqual(len(memories), n_memories)
for i, memory in enumerate(memories):
self.assertIsInstance(memory, dict)
self.assertIn("turn", memory)
self.assertIn("text", memory)
self.assertIn("memory_n", memory)
# check that the memory has the correct index
self.assertEqual(memory["memory_n"], i+1)
def test_gen_reflection(self):
last_messages = [
(1, "Mark", "I like discussing about 18th century german philosophy."),
(2, "Alex", "Oh really? I really think that in his seminal work, Kant was really onto something"),
(3, "Mark", "I disagree. I think that Hegel's dialectical method is more relevant than ever.")
]
n_memories = 3
n_reflections = 5
agent_list = ["Mark", "Alex"]
n_agents = 2
turn_count = 4
name = "Mark"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 5
reflections_freq = 10
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
memories=agent.gen_memories(last_messages, n_memories, agent_list=agent_list, n_agents=n_agents, turn_count=turn_count)
agent.save_observations(memories)
reflections = agent.gen_reflection(n_memories, n_reflections, agent_list=agent_list, n_agents=n_agents, turn_count=turn_count)
self.assertIsInstance(reflections, list)
self.assertEqual(len(reflections), n_reflections)
for i, reflection in enumerate(reflections):
self.assertIsInstance(reflection, dict)
self.assertIn("memory_n", reflection)
self.assertIn("turn", reflection)
self.assertIn("text", reflection)
# check that the reflection has the correct index
self.assertEqual(reflection["memory_n"], agent.memory.count() + i + 1)
def test_save_observations(self):
observations_list = [
{"id": 1, "text": "Observation 1"},
{"id": 2, "text": "Observation 2"},
{"id": 3, "text": "Observation 3"}
]
name = "John"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 5
reflections_freq = 10
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
agent.save_observations(observations_list)
# Verify that the observations are saved in memory
self.assertEqual(agent.memory.count(), len(observations_list))
total_memories = agent.memory.search("select id, text from txtai")
for observation in observations_list:
for memory in total_memories:
if memory["text"] == observation["text"]:
break
else:
self.fail(f"Observation {observation} not found in memory")
def test_generate_with_memory(self):
last_messages = [
(1, "Mark", "I like discussing about 18th century german philosophy."),
(2, "Alex", "Oh really? I really think that in his seminal work, Kant was really onto something"),
(3, "Mark", "I disagree. I think that Hegel's dialectical method is more relevant than ever.")
]
extra_context = "Some extra context"
agent_list = ["Mark", "Alex"]
n_agents = 2
name = "Mark"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 5
reflections_freq = 10
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
## generate the memory
memories = agent.gen_memories(last_messages, n_memories=5, agent_list=agent_list, n_agents=n_agents, turn_count=4)
## save the memory
agent.save_observations(memories)
answer = agent.get_answer(last_messages, extra_context=extra_context, agent_list=agent_list, n_agents=n_agents, turn_count=4)
self.assertIsInstance(answer, str)
self.assertNotEqual(answer, "")
def test_run_routines(self):
turn_count = 5
chat_history = [
(1, "Mark", "Hello"),
(2, "Alex", "Hi Mark, how are you?"),
(3, "Mark", "I'm good, thanks for asking.")
]
name = "John"
llm = LLMApi()
persona = "I am a persona"
memory_freq = 5
reflections_freq = 10
n_fewshots = 5
agent = DialogueReactAgent(name, llm, persona, memory_freq, reflections_freq, n_fewshots)
agent.run_routines(turn_count, chat_history, n_agents=2, agent_list=["Mark", "Alex"])
# Verify that save_observations is called when turn_count is divisible by memory_freq
if turn_count % memory_freq == 0:
assert agent.memory.count()==5
## turn count is divisible by reflections_freq
turn_count = 20
print(f"N memory objects before: {agent.memory.count()}")
agent.run_routines(turn_count, chat_history, n_agents=2, agent_list=["Mark", "Alex"])
print(f"N memory objects after: {agent.memory.count()}")
if turn_count % reflections_freq == 0:
assert agent.memory.count()==15
if __name__ == '__main__':
unittest.main()