-
Notifications
You must be signed in to change notification settings - Fork 12
/
brainstorming.gpt4.txt
218 lines (192 loc) · 11.8 KB
/
brainstorming.gpt4.txt
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
The following command runs the described program to generate a task-oriented dialogue (TOD) dataset on the topic of hotels:
```sh
datasetGPT conversations \
--length 10 \
--agent1 "You are a receptionist at a {hotel_type} hotel called [hotel|{hotel_name}] and you are {interaction_type} with a clinet. Provide assistance and answer to their questions. You must replace any situation-specific details with label and value pairs - for example, [check in|tomorrow], [check out|August 15th], [name|John], [phone|212-456-7890], [rate|...], [total cost|...], etc." \
--agent2 "You are a client of the {hotel_type} hotel called [hotel|{hotel_name}] and you are {interaction_type} with a receptionist. Either try to make a reservation or ask questions related to an existing reservation. You must replace any situation-specific details with label and value pairs - for example, [check in|tomorrow], [check out|August 15th], [name|John], [phone|212-456-7890], [rate|...], [total cost|...], etc." \
--initial-utterance "Hello." \
--interruption "end_phrase" \
--end-phrase "Goodbye" \
--end-agent "agent1" \
--temperature 0.1 \
--temperature 0.7 \
--option interaction_type "chatting online" \
--option interaction_type "talking" \
--option hotel_type "family" \
--option hotel_type "budget" \
--option hotel_type "airport" \
--option hotel_type "boutique" \
--option hotel_type "historic" \
--option hotel_name "Four Seasons" \
--option hotel_name "The Cursed Castle" \
--num-samples 10 \
--path "tod/hotels.json" \
--single-file
```
It produces 400 conversations by first permuting all the different configurations of --option and --temperature; then it generates --num-samples number of conversations for each configuration. Rewrite the given command to produce 100 brainstorming conversations about scientific problems. Do not instruct the LLMs to use situation-specific labels.
---
```sh
datasetGPT conversations \
--length 10 \
--agent1 "You are a scientist brainstorming about scientific problems with another scientist. Discuss various scientific challenges, propose hypotheses, and explore potential research directions. Collaborate and build upon each other's ideas." \
--agent2 "You are a scientist brainstorming about scientific problems with another scientist. Discuss various scientific challenges, propose hypotheses, and explore potential research directions. Collaborate and build upon each other's ideas." \
--initial-utterance "Let's discuss some scientific problems and try to come up with potential research directions." \
--interruption "end_phrase" \
--end-phrase "This has been a fruitful discussion. Let's wrap up our brainstorming session for now." \
--end-agent "agent1" \
--temperature 0.1 \
--temperature 0.7 \
--num-samples 100 \
--path "brainstorming/scientific_problems.json" \
--single-file
```
This command will generate 100 brainstorming conversations about scientific problems between two large language models without using situation-specific labels.
---
The command will generate 200 conversations since 100 conversations will be generated for each temperature. Please correct that. Also, prompt the LLMs on a specific scientific field/problem they need to discuss and use --option to supply multiple possibilities.
---
```sh
datasetGPT conversations \
--length 10 \
--agent1 "You are a scientist brainstorming with another scientist about {scientific_field}. Discuss various challenges, propose hypotheses, and explore potential research directions within this field. Collaborate and build upon each other's ideas." \
--agent2 "You are a scientist brainstorming with another scientist about {scientific_field}. Discuss various challenges, propose hypotheses, and explore potential research directions within this field. Collaborate and build upon each other's ideas." \
--initial-utterance "Let's discuss some problems and research directions in the field of {scientific_field}." \
--interruption "end_phrase" \
--end-phrase "This has been a fruitful discussion. Let's wrap up our brainstorming session for now." \
--end-agent "agent1" \
--temperature 0.7 \
--option scientific_field "climate change" \
--option scientific_field "artificial intelligence" \
--option scientific_field "genetic engineering" \
--option scientific_field "quantum computing" \
--option scientific_field "neuroscience" \
--num-samples 20 \
--path "brainstorming/scientific_problems.json" \
--single-file
```
This command will now generate 100 brainstorming conversations about specific scientific fields or problems between two large language models, with 20 conversations per each field specified in the --option command. The temperature is set to 0.7.
---
Please bring back the temperature of 0.1
---
```sh
datasetGPT conversations \
--length 10 \
--agent1 "You are a scientist brainstorming with another scientist about {scientific_field}. Discuss various challenges, propose hypotheses, and explore potential research directions within this field. Collaborate and build upon each other's ideas." \
--agent2 "You are a scientist brainstorming with another scientist about {scientific_field}. Discuss various challenges, propose hypotheses, and explore potential research directions within this field. Collaborate and build upon each other's ideas." \
--initial-utterance "Let's discuss some problems and research directions in the field of {scientific_field}." \
--interruption "end_phrase" \
--end-phrase "This has been a fruitful discussion. Let's wrap up our brainstorming session for now." \
--end-agent "agent1" \
--temperature 0.1 \
--temperature 0.7 \
--option scientific_field "climate change" \
--option scientific_field "artificial intelligence" \
--option scientific_field "genetic engineering" \
--option scientific_field "quantum computing" \
--option scientific_field "neuroscience" \
--num-samples 10 \
--path "brainstorming/scientific_problems.json" \
--single-file
```
This command will generate 100 brainstorming conversations about specific scientific fields or problems between two large language models, with 10 conversations per each field specified in the --option command, and includes both temperature settings of 0.1 and 0.7.
---
Pretty good. The last thing to change is to bring the --num-samples down to 2 by adding more scientific_field options.
---
```sh
datasetGPT conversations \
--length 10 \
--agent1 "You are a scientist brainstorming with another scientist about {scientific_field}. Discuss various challenges, propose hypotheses, and explore potential research directions within this field. Collaborate and build upon each other's ideas." \
--agent2 "You are a scientist brainstorming with another scientist about {scientific_field}. Discuss various challenges, propose hypotheses, and explore potential research directions within this field. Collaborate and build upon each other's ideas." \
--initial-utterance "Let's discuss some problems and research directions in the field of {scientific_field}." \
--interruption "end_phrase" \
--end-phrase "This has been a fruitful discussion. Let's wrap up our brainstorming session for now." \
--end-agent "agent1" \
--temperature 0.1 \
--temperature 0.7 \
--option scientific_field "climate change" \
--option scientific_field "artificial intelligence" \
--option scientific_field "genetic engineering" \
--option scientific_field "quantum computing" \
--option scientific_field "neuroscience" \
--option scientific_field "nanotechnology" \
--option scientific_field "material science" \
--option scientific_field "renewable energy" \
--option scientific_field "astrobiology" \
--option scientific_field "particle physics" \
--num-samples 2 \
--path "brainstorming/scientific_problems.json" \
--single-file
```
This command will generate 100 brainstorming conversations about specific scientific fields or problems between two large language models, with 2 conversations per each of the 10 fields specified in the --option command, and includes both temperature settings of 0.1 and 0.7.
---
2 conversations per 10 fields with 2 temperatures make for 40 conversations, though. Please add more scientific_field options for a total of 25! List only the --option lines without the rest of the command:
---
Here are the additional --option lines to cover 25 scientific fields:
```
--option scientific_field "climate change"
--option scientific_field "artificial intelligence"
--option scientific_field "genetic engineering"
--option scientific_field "quantum computing"
--option scientific_field "neuroscience"
--option scientific_field "nanotechnology"
--option scientific_field "material science"
--option scientific_field "renewable energy"
--option scientific_field "astrobiology"
--option scientific_field "particle physics"
--option scientific_field "biomedical engineering"
--option scientific_field "robotics"
--option scientific_field "network science"
--option scientific_field "oerospace engineering"
--option scientific_field "oceanography"
--option scientific_field "ecology"
--option scientific_field "geology"
--option scientific_field "epidemiology"
--option scientific_field "cell biology"
--option scientific_field "biochemistry"
--option scientific_field "immunology"
--option scientific_field "theoretical physics"
--option scientific_field "chemistry"
--option scientific_field "cognitive science"
--option scientific_field "psychology"
```
---
Edited by the author:
```sh
datasetGPT conversations \
--length 10 \
--agent1 "You are a scientist brainstorming with another scientist about the current state of {scientific_field} and the potential research directions in the field. Engage in a conversation. Follow up on the points of your colleague and discuss them rather than just listing more ideas. Talk concisely with short messages, and do not repeat yourself." \
--agent2 "You are a scientist brainstorming with another scientist about the current state of {scientific_field} and the potential research directions in the field. Engage in a conversation. Follow up on the points of your colleague and discuss them rather than just listing more ideas. Talk concisely with short messages, and do not repeat yourself." \
--initial-utterance "Let's discuss some problems and research directions in our scientific field." \
--interruption "end_phrase" \
--end-phrase "Goodbye" \
--end-agent "both" \
--temperature 0.1 \
--temperature 0.7 \
--option scientific_field "climate change" \
--option scientific_field "artificial intelligence" \
--option scientific_field "genetic engineering" \
--option scientific_field "quantum computing" \
--option scientific_field "neuroscience" \
--option scientific_field "nanotechnology" \
--option scientific_field "material science" \
--option scientific_field "renewable energy" \
--option scientific_field "astrobiology" \
--option scientific_field "particle physics" \
--option scientific_field "biomedical engineering" \
--option scientific_field "robotics" \
--option scientific_field "network science" \
--option scientific_field "oerospace engineering" \
--option scientific_field "oceanography" \
--option scientific_field "ecology" \
--option scientific_field "geology" \
--option scientific_field "epidemiology" \
--option scientific_field "cell biology" \
--option scientific_field "biochemistry" \
--option scientific_field "immunology" \
--option scientific_field "theoretical physics" \
--option scientific_field "chemistry" \
--option scientific_field "cognitive science" \
--option scientific_field "psychology" \
--num-samples 2 \
--path "brainstorming/scientific_problems.json" \
--single-file
```