langchain/docs/examples/prompts/walkthrough.ipynb
Harrison Chase 0d0d3f122a cr
2022-11-14 22:54:12 -08:00

411 lines
9.5 KiB
Plaintext

{
"cells": [
{
"cell_type": "markdown",
"id": "43fb16cb",
"metadata": {},
"source": [
"# Prompt Walkthrough\n",
"\n",
"An overview of the different types of prompts in LangChain and how to use them"
]
},
{
"cell_type": "markdown",
"id": "cddb465e",
"metadata": {},
"source": [
"### Basic Prompt\n",
"\n",
"The most simple type of prompt - a string template that takes any number of input variables. The template should be formatted as a Python f-string."
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "094229f4",
"metadata": {},
"outputs": [],
"source": [
"from langchain.prompts import Prompt"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "ab46bd2a",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'Tell me a joke.'"
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# An example prompt with no input variables\n",
"no_input_prompt = Prompt(input_variables=[], template=\"Tell me a joke.\")\n",
"no_input_prompt.format()"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "c3ad0fa8",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'Tell me a funny joke.'"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# An example prompt with one input variable\n",
"no_input_prompt = Prompt(input_variables=[\"adjective\"], template=\"Tell me a {adjective} joke.\")\n",
"no_input_prompt.format(adjective=\"funny\")"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "ba577dcf",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"'Tell me a funny joke about chickens.'"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# An example prompt with multiple input variables\n",
"no_input_prompt = Prompt(input_variables=[\"adjective\", \"content\"], template=\"Tell me a {adjective} joke about {content}.\")\n",
"no_input_prompt.format(adjective=\"funny\", content=\"chickens\")"
]
},
{
"cell_type": "markdown",
"id": "d27b1824",
"metadata": {},
"source": [
"### Examples\n",
"Examples are datapoints that can be used to show the model how to produce results. They can be either strings, or dictionaries that are then turned into strings by an example prompt itself."
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "2c00e965",
"metadata": {},
"outputs": [],
"source": [
"string_examples = [\"Input: happy\\nOutput: sad\", \"Input: tall\\nOutput: short\"]\n",
"dict_examples = [{\"input\": \"happy\", \"output\": \"sad\"}, {\"input\": \"tall\", \"output\": \"short\"}]\n",
"example_prompt = Prompt(input_variables=[\"input\",\"output\"], template=\"Input: {input}\\nOutput: {output}\")"
]
},
{
"cell_type": "markdown",
"id": "1492b49d",
"metadata": {},
"source": [
"### Simple Prompt with examples\n",
"\n",
"We can then use these examples to construct prompts."
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "1a5a686d",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Give the antonym of every input\n",
"\n",
"Input: happy\n",
"Output: sad\n",
"\n",
"Input: tall\n",
"Output: short\n",
"\n",
"Input: big\n",
"Output:\n"
]
}
],
"source": [
"prompt_from_string_examples = Prompt.from_examples(\n",
" string_examples, \n",
" prefix=\"Give the antonym of every input\",\n",
" suffix=\"Input: {adjective}\\nOutput:\", \n",
" input_variables=[\"adjective\"],\n",
")\n",
"print(prompt_from_string_examples.format(adjective=\"big\"))"
]
},
{
"cell_type": "code",
"execution_count": 12,
"id": "7931e5f2",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Give the antonym of every input\n",
"\n",
"Input: happy\n",
"Output: sad\n",
"\n",
"Input: tall\n",
"Output: short\n",
"\n",
"Input: big\n",
"Output:\n"
]
}
],
"source": [
"prompt_from_string_examples = Prompt.from_structured_examples(\n",
" dict_examples,\n",
" example_prompt,\n",
" prefix=\"Give the antonym of every input\",\n",
" suffix=\"Input: {adjective}\\nOutput:\", \n",
" input_variables=[\"adjective\"],\n",
")\n",
"print(prompt_from_string_examples.format(adjective=\"big\"))"
]
},
{
"cell_type": "markdown",
"id": "861a4d1f",
"metadata": {},
"source": [
"### Dynamic Prompt\n",
"\n",
"We also do more clever things with prompts - for example, only select a certain number of examples in order to limit the size of the text passed in. This will vary with the input text size."
]
},
{
"cell_type": "code",
"execution_count": 13,
"id": "7c469c95",
"metadata": {},
"outputs": [],
"source": [
"from langchain.prompts import DynamicPrompt"
]
},
{
"cell_type": "code",
"execution_count": 17,
"id": "207e55f7",
"metadata": {},
"outputs": [],
"source": [
"dynamic_prompt = DynamicPrompt.from_structured_examples(\n",
" dict_examples,\n",
" example_prompt,\n",
" prefix=\"Give the antonym of every input\",\n",
" suffix=\"Input: {adjective}\\nOutput:\", \n",
" input_variables=[\"adjective\"],\n",
" max_length=20,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 19,
"id": "d00b4385",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Give the antonym of every input\n",
"\n",
"Input: happy\n",
"Output: sad\n",
"\n",
"Input: tall\n",
"Output: short\n",
"\n",
"Input: big\n",
"Output:\n"
]
}
],
"source": [
"# An example with small input, so it selects both examples.\n",
"print(dynamic_prompt.format(adjective=\"big\"))"
]
},
{
"cell_type": "code",
"execution_count": 20,
"id": "878bcde9",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Give the antonym of every input\n",
"\n",
"Input: happy\n",
"Output: sad\n",
"\n",
"Input: big and huge and massive\n",
"Output:\n"
]
}
],
"source": [
"# An example with long input, so it selects only one example.\n",
"print(dynamic_prompt.format(adjective=\"big and huge and massive\"))"
]
},
{
"cell_type": "markdown",
"id": "2d007b0a",
"metadata": {},
"source": [
"# Optimized Prompt\n",
"\n",
"Besides selecting a variable number of examples to show, we can also select examples that most closely match the user input. This is done by creating embeddings of the user input and comparing it embeddings of the examples."
]
},
{
"cell_type": "code",
"execution_count": 23,
"id": "241bfe80",
"metadata": {},
"outputs": [],
"source": [
"from langchain.prompts.optimized import OptimizedPrompt\n",
"from langchain.vectorstores import FAISS\n",
"from langchain.embeddings import OpenAIEmbeddings"
]
},
{
"cell_type": "code",
"execution_count": 25,
"id": "50d0a701",
"metadata": {},
"outputs": [],
"source": [
"optimized_prompt = OptimizedPrompt.from_structured_examples(\n",
" dict_examples,\n",
" example_prompt,\n",
" prefix=\"Give the antonym of every input\",\n",
" suffix=\"Input: {adjective}\\nOutput:\", \n",
" input_variables=[\"adjective\"],\n",
" embeddings=OpenAIEmbeddings(),\n",
" vectorstore_cls=FAISS\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 28,
"id": "4c8fdf45",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Give the antonym of every input\n",
"\n",
"Input: happy\n",
"Output: sad\n",
"\n",
"Input: worried\n",
"Output:\n"
]
}
],
"source": [
"# Input is a feeling, so should select the happy/sad example\n",
"print(optimized_prompt.format(adjective=\"worried\", k=1))"
]
},
{
"cell_type": "code",
"execution_count": 29,
"id": "829af21a",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Give the antonym of every input\n",
"\n",
"Input: tall\n",
"Output: short\n",
"\n",
"Input: fat\n",
"Output:\n"
]
}
],
"source": [
"# Input is a measurment, so should select the tall/short example\n",
"print(optimized_prompt.format(adjective=\"fat\", k=1))"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "76a1065d",
"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": "3.7.6"
}
},
"nbformat": 4,
"nbformat_minor": 5
}