File size: 11,182 Bytes
ed4d993
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "9e9b7651",
   "metadata": {},
   "source": [
    "# How to use a SmartLLMChain\n",
    "\n",
    "A SmartLLMChain is a form of self-critique chain that can help you if have particularly complex questions to answer. Instead of doing a single LLM pass, it instead performs these 3 steps:\n",
    "1. Ideation: Pass the user prompt n times through the LLM to get n output proposals (called \"ideas\"), where n is a parameter you can set \n",
    "2. Critique: The LLM critiques all ideas to find possible flaws and picks the best one \n",
    "3. Resolve: The LLM tries to improve upon the best idea (as chosen in the critique step) and outputs it. This is then the final output.\n",
    "\n",
    "SmartLLMChains are based on the SmartGPT workflow proposed in https://youtu.be/wVzuvf9D9BU.\n",
    "\n",
    "Note that SmartLLMChains\n",
    "- use more LLM passes (ie n+2 instead of just 1)\n",
    "- only work then the underlying LLM has the capability for reflection, which smaller models often don't\n",
    "- only work with underlying models that return exactly 1 output, not multiple\n",
    "\n",
    "This notebook demonstrates how to use a SmartLLMChain."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "714dede0",
   "metadata": {},
   "source": [
    "##### Same LLM for all steps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "d3f7fb22",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "os.environ[\"OPENAI_API_KEY\"] = \"...\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "10e5ece6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.prompts import PromptTemplate\n",
    "from langchain_experimental.smart_llm import SmartLLMChain\n",
    "from langchain_openai import ChatOpenAI"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "1780da51",
   "metadata": {},
   "source": [
    "As example question, we will use \"I have a 12 liter jug and a 6 liter jug. I want to measure 6 liters. How do I do it?\". This is an example from the original SmartGPT video (https://youtu.be/wVzuvf9D9BU?t=384). While this seems like a very easy question, LLMs struggle do these kinds of questions that involve numbers and physical reasoning.\n",
    "\n",
    "As we will see, all 3 initial ideas are completely wrong - even though we're using GPT4! Only when using self-reflection do we get a correct answer. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "054af6b1",
   "metadata": {},
   "outputs": [],
   "source": [
    "hard_question = \"I have a 12 liter jug and a 6 liter jug. I want to measure 6 liters. How do I do it?\""
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "8049cecd",
   "metadata": {},
   "source": [
    "So, we first create an LLM and prompt template"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "811ea8e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "prompt = PromptTemplate.from_template(hard_question)\n",
    "llm = ChatOpenAI(temperature=0, model_name=\"gpt-4\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "50b602e4",
   "metadata": {},
   "source": [
    "Now we can create a SmartLLMChain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8cd49199",
   "metadata": {},
   "outputs": [],
   "source": [
    "chain = SmartLLMChain(llm=llm, prompt=prompt, n_ideas=3, verbose=True)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "6a72f276",
   "metadata": {},
   "source": [
    "Now we can use the SmartLLM as a drop-in replacement for our LLM. E.g.:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "074e5e75",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new SmartLLMChain chain...\u001b[0m\n",
      "Prompt after formatting:\n",
      "\u001b[32;1m\u001b[1;3mI have a 12 liter jug and a 6 liter jug. I want to measure 6 liters. How do I do it?\u001b[0m\n",
      "Idea 1:\n",
      "\u001b[36;1m\u001b[1;3m1. Fill the 6-liter jug completely.\n",
      "2. Pour the water from the 6-liter jug into the 12-liter jug.\n",
      "3. Fill the 6-liter jug again.\n",
      "4. Carefully pour the water from the 6-liter jug into the 12-liter jug until the 12-liter jug is full.\n",
      "5. The amount of water left in the 6-liter jug will be exactly 6 liters.\u001b[0m\n",
      "Idea 2:\n",
      "\u001b[36;1m\u001b[1;3m1. Fill the 6-liter jug completely.\n",
      "2. Pour the water from the 6-liter jug into the 12-liter jug.\n",
      "3. Fill the 6-liter jug again.\n",
      "4. Carefully pour the water from the 6-liter jug into the 12-liter jug until the 12-liter jug is full.\n",
      "5. Since the 12-liter jug is now full, there will be 2 liters of water left in the 6-liter jug.\n",
      "6. Empty the 12-liter jug.\n",
      "7. Pour the 2 liters of water from the 6-liter jug into the 12-liter jug.\n",
      "8. Fill the 6-liter jug completely again.\n",
      "9. Pour the water from the 6-liter jug into the 12-liter jug, which already has 2 liters in it.\n",
      "10. Now, the 12-liter jug will have exactly 6 liters of water (2 liters from before + 4 liters from the 6-liter jug).\u001b[0m\n",
      "Idea 3:\n",
      "\u001b[36;1m\u001b[1;3m1. Fill the 6-liter jug completely.\n",
      "2. Pour the water from the 6-liter jug into the 12-liter jug.\n",
      "3. Fill the 6-liter jug again.\n",
      "4. Carefully pour the water from the 6-liter jug into the 12-liter jug until the 12-liter jug is full.\n",
      "5. The amount of water left in the 6-liter jug will be exactly 6 liters.\u001b[0m\n",
      "Critique:\n",
      "\u001b[33;1m\u001b[1;3mIdea 1:\n",
      "1. Fill the 6-liter jug completely. (No flaw)\n",
      "2. Pour the water from the 6-liter jug into the 12-liter jug. (No flaw)\n",
      "3. Fill the 6-liter jug again. (No flaw)\n",
      "4. Carefully pour the water from the 6-liter jug into the 12-liter jug until the 12-liter jug is full. (Flaw: The 12-liter jug will never be full in this step, as it can hold 12 liters and we are only pouring 6 liters into it.)\n",
      "5. The amount of water left in the 6-liter jug will be exactly 6 liters. (Flaw: This statement is incorrect, as there will be no water left in the 6-liter jug after pouring it into the 12-liter jug.)\n",
      "\n",
      "Idea 2:\n",
      "1. Fill the 6-liter jug completely. (No flaw)\n",
      "2. Pour the water from the 6-liter jug into the 12-liter jug. (No flaw)\n",
      "3. Fill the 6-liter jug again. (No flaw)\n",
      "4. Carefully pour the water from the 6-liter jug into the 12-liter jug until the 12-liter jug is full. (Flaw: The 12-liter jug will never be full in this step, as it can hold 12 liters and we are only pouring 6 liters into it.)\n",
      "5. Since the 12-liter jug is now full, there will be 2 liters of water left in the 6-liter jug. (Flaw: This statement is incorrect, as the 12-liter jug will not be full and there will be no water left in the 6-liter jug.)\n",
      "6. Empty the 12-liter jug. (No flaw)\n",
      "7. Pour the 2 liters of water from the 6-liter jug into the 12-liter jug. (Flaw: This step is based on the incorrect assumption that there are 2 liters of water left in the 6-liter jug.)\n",
      "8. Fill the 6-liter jug completely again. (No flaw)\n",
      "9. Pour the water from the 6-liter jug into the 12-liter jug, which already has 2 liters in it. (Flaw: This step is based on the incorrect assumption that there are 2 liters of water in the 12-liter jug.)\n",
      "10. Now, the 12-liter jug will have exactly 6 liters of water (2 liters from before + 4 liters from the 6-liter jug). (Flaw: This conclusion is based on the incorrect assumptions made in the previous steps.)\n",
      "\n",
      "Idea 3:\n",
      "1. Fill the 6-liter jug completely. (No flaw)\n",
      "2. Pour the water from the 6-liter jug into the 12-liter jug. (No flaw)\n",
      "3. Fill the 6-liter jug again. (No flaw)\n",
      "4. Carefully pour the water from the 6-liter jug into the 12-liter jug until the 12-liter jug is full. (Flaw: The 12-liter jug will never be full in this step, as it can hold 12 liters and we are only pouring 6 liters into it.)\n",
      "5. The amount of water left in the 6-liter jug will be exactly 6 liters. (Flaw: This statement is incorrect, as there will be no water left in the 6-liter jug after pouring it into the 12-liter jug.)\u001b[0m\n",
      "Resolution:\n",
      "\u001b[32;1m\u001b[1;3m1. Fill the 12-liter jug completely.\n",
      "2. Pour the water from the 12-liter jug into the 6-liter jug until the 6-liter jug is full.\n",
      "3. The amount of water left in the 12-liter jug will be exactly 6 liters.\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'1. Fill the 12-liter jug completely.\\n2. Pour the water from the 12-liter jug into the 6-liter jug until the 6-liter jug is full.\\n3. The amount of water left in the 12-liter jug will be exactly 6 liters.'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chain.invoke({})"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "bbfebea1",
   "metadata": {},
   "source": [
    "##### Different LLM for different steps"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "5be6ec08",
   "metadata": {},
   "source": [
    "You can also use different LLMs for the different steps by passing `ideation_llm`, `critique_llm` and `resolve_llm`. You might want to do this to use a more creative (i.e., high-temperature) model for ideation and a more strict (i.e., low-temperature) model for critique and resolution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9c33fa19",
   "metadata": {},
   "outputs": [],
   "source": [
    "chain = SmartLLMChain(\n",
    "    ideation_llm=ChatOpenAI(temperature=0.9, model_name=\"gpt-4\"),\n",
    "    llm=ChatOpenAI(\n",
    "        temperature=0, model_name=\"gpt-4\"\n",
    "    ),  # will be used for critique and resolution as no specific llms are given\n",
    "    prompt=prompt,\n",
    "    n_ideas=3,\n",
    "    verbose=True,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "886c1cc1",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}