首先当然是需要一个apikey。
import openai
openai.api_key = ""
try:
response = openai.Completion.create(
engine="text-davinci-002",
prompt="Write a tagline about artificial intelligence.",
temperature=0.9
)
print(response)
except Exception as e:
print(f"Error: {e}")
分类的时候使用关键字Classify提示，然后给他相应标签。最后的Sentiment相当于给他一个答案的开头。
import openai
openai.api_key = ""
prompt_text = ("Classify the Text's label as positive, neutral, or negative.\n"
+ "Text: I loved the new Spiderman movie!\n"
+ "Sentiment:")
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text
)
print("Prompt text:\n" + prompt_text + response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
尝试设置较高的温度，温度调节为0到1之间，越高越有创意，越大胆。
import openai
openai.api_key = ""
prompt_text = ("Classify the Text.\n"
+ "Text: I loved the new Spiderman movie!\n")
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0.9,
top_p=1
)
print("Prompt text:\n" + prompt_text + response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
解释response.choices[0].text:
response: 这是通过调用 OpenAI API 后得到的响应对象。它包含了从模型返回的所有信息，包括生成的文本、模型的置信度等。
response.choices: 这是响应中的一个属性，表示模型的选择列表。在这里，模型只返回了一个选择（最可能的生成文本），因此我们可以通过 response.choices[0] 来访问这个唯一的选择。
response.choices[0].text: 这是选择中的一个属性，表示模型生成的文本。它包含了模型认为是最适合请求的文本部分。
尝试进行多条文本分类
import openai
openai.api_key = ""
try:
prompt_text = ("Classify the sentiment in these tweets:\n"
+ "1. \"I enjoy this program\"\n"
+ "2. \"I don't like him anymore\"\n"
+ "3. \"I can't wait for Halloween!!!\"\n"
+ "4. \"I do not hate much jokers\"\n"
+ "5. \"I love eating chocolates <3!\"\n"
+ "Tweet sentiment ratings:")
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
max_tokens = 60,
temperature=0.9
)
print("Prompt text:\n" + prompt_text)
print(response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
max_tokens的默认是16，这里肯定会超出限制，所以改为60
给项目分类因为他擅长将项目分为离散类别。
import openai
openai.api_key = ""
prompt_text = "The following is a list of companies and the categories they fall into:\n\nApple, Facebook, Educative\n\nApple\nCategory:"
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0,
max_tokens=64,
top_p=1.0,
)
print("Prompt: ", prompt_text, response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
提示以生成文本
import openai
openai.api_key = ""
prompt_text = "Brainstorm some ideas combining deep learning and images:\n"
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0.75,
max_tokens=200,
top_p=1,
frequency_penalty=1,
presence_penalty=1
)
print ("Prompt text:\n" + prompt_text)
print(response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
max_tokens的最大长度就是200了。
在 OpenAI GPT 模型的生成请求中，frequency_penalty 和 presence_penalty 是两个用于调整生成文本行为的参数：
frequency_penalty：
presence_penalty：
文本翻译啦，总结文本啦，文本转换为emoji啦都是这类任务。
文本翻译
import openai
openai.api_key = ""
prompt_text = ("Translate this into 1. French, 2. Spanish and 3. Japanese.\n" +
"We provide an online learning platform made by developers, created for developers.")
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0,
max_tokens=100,
)
print("Prompt: \n" + prompt_text)
print(response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
这里温度设置为0表示不需要翻译的随机性。max_tokens的设置是因为知道会输出多长。
内容提炼
import openai
import textwrap
openai.api_key =""
prompt_text = ("Summarize this for a second-grade student:\nJupiter is the fifth planet from the "+
"Sun and the largest in the Solar System. It is a gas giant with a mass " +
"one-thousandth that of the Sun, but two-and-a-half times that of all the " +
"other planets in the Solar System combined. Jupiter is one of the brightest " +
"objects visible to the naked eye in the night sky, and has been known to ancient " +
"civilizations since before recorded history. It is named after the Roman god " +
"Jupiter. When viewed from Earth, Jupiter can be bright enough for its " +
"reflected light to cast visible shadows, and is on average the third-brightest " +
"natural object in the night sky after the Moon and Venus.")
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0.9,
max_tokens=64,
frequency_penalty=1,
presence_penalty=1
)
print("Prompt:")
print (textwrap.fill(prompt_text, width=80))
print("\nSummary:")
print (textwrap.fill(response.choices[0].text, width=80))
except Exception as e:
print(f"Error: {e}")
转换文本为emoji
import openai
openai.api_key = ""
try:
prompt_text = ("Convert movie titles into emojis.\n" +
"Back to the Future: 👨👴🚗🕒\n" +
"Batman: 🤵🦇\n" +
"Transformers: 🚗🤖\n" +
"Antman: ")
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0.9,
)
print("Prompt: \n" + prompt_text)
print(response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
给定前缀和后缀文本，进行文本插入。
import openai
import textwrap
openai.api_key = ""
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt="Educative is a leading online learning platform made by developers, created for developers.",
suffix="Educative helps software engineers grow their skill sets and reach their full potential.",
max_tokens=260
)
print("Insertion result:")
print(textwrap.fill(response.choices[0].text, width=80))
except Exception as e:
print(f"Error: {e}")
另一个例子：
import openai
import textwrap
openai.api_key = ""
prompt_prefix = "How to lose weight:\n1. Do not skip breakfast."
prompt_suffix = "12. Plan your meals"
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_prefix,
suffix=prompt_suffix,
max_tokens=400
)
print("\nPrefix:\n", prompt_prefix)
print("\nInsertion result:")
print(textwrap.fill(response.choices[0].text, width=80))
print("\nSuffix:\n", prompt_suffix)
except Exception as e:
print(f"Error: {e}")
只给一个句子开头，他会帮你完成。
import openai
openai.api_key = ""
prompt_text = "Although the tiger is a solitary beast and "
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0.20,
max_tokens=64,
top_p=1,
)
print("Prompt: " + prompt_text)
print("\nCompletion: " + response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
如果想告诉他，比如当他不知道该怎么做的时候，给出具体的回答方法。
import openai
openai.api_key = ""
prompt_text=("Q: Who is Batman?\n"+
"A: Batman is a fictional comic book character.\n"+
"Q: What is torsalplexity?\n"+
"A: ?\n"+
"Q: What is Devz9?\n"+
"A: ?\n"+
"Q: Who is George Lucas?\n"+
"A: George Lucas is American film director and producer famous for creating Star Wars.\n"+
"Q: What is the capital of California?\n"+
"A: Sacramento.\n"+
"Q: What orbits the Earth?\n"+
"A: The Moon.\n"+
"Q: Who is Fred Rickerson?\n"+
"A: ?\n"+
"Q: What is an atom?\n"+
"A: An atom is a tiny particle that makes up everything.\n"+
"Q: Who is Alvan Muntz?\n"+
"A: ?\n"+
"Q: What is Kozar-09?\n"+
"A: ?\n"+
"Q: How many moons does Mars have?\n"+
"A: Two, Phobos and Deimos.\n"+
"Q: what is the meaning of xyzz?\n"+
"A:")
try:
response = openai.Completion.create(
model="text-davinci-002",
prompt=prompt_text,
temperature=0,
max_tokens=64,
)
print("Prompt: \n" + prompt_text)
print("Response: \n" + response.choices[0].text)
except Exception as e:
print(f"Error: {e}")
response = openai.Edit.create(
model="<engine_id>",
input="<string>",
instruction="<string>",
...
)
比如语法修正
import openai
openai.api_key = ""
try:
response = openai.Edit.create(
model="text-davinci-edit-001",
input="Educative is a leading online learning platform made by developers, create for developers.",
instruction="Fix the grammar.",
temperature=0,
top_p=1
)
print(response)
except Exception as e:
print(f"Error: {e}")
语态修正
import openai
openai.api_key = ""
try:
response = openai.Edit.create(
model="text-davinci-edit-001",
input="The kangaroo carries her baby in her pouch.",
instruction="Convert sentence to passive voice",
temperature=0,
top_p=1
)
print(response)
except Exception as e:
print(f"Error: {e}")
文本相似度（回归，可视化，聚类，异常检测），文字搜索（上下文检索，搜索相关），代码搜索（搜索相关代码）
相关模块：
response = openai.Embedding.create(
input="The text whose embeddings are required",
engine="<engine_id>"
)
文本相似度
import openai
openai.api_key = ""
try:
response = openai.Embedding.create(
input="The burger was fantastic!",
engine="text-similarity-babbage-001"
)
embedding = response.data[0].embedding
print("Embedding: ", embedding)
print("Length: ", len(embedding))
except Exception as e:
print(f"Error: {e}")
response.data[0].embedding可以从响应对象中提取嵌入向量。
计算嵌入向量的余弦相似度
import openai
from numpy import dot
openai.api_key = ""
try:
response = openai.Embedding.create(
input=["The burger was fantastic", "The quick brown fox jumps over the lazy dog"],
engine="text-similarity-davinci-001")
emb_1 = response['data'][0]['embedding']
emb_2 = response['data'][1]['embedding']
# Formula to find the cosine similarity
similarity_score = dot(emb_1, emb_2)
print(similarity_score)
except Exception as e:
print(f"Error: {e}")