openai

ML system design

Frame the problem as an ML task

Defining the ML objective
Specifying the system's input and output
Choosing the right ML category

Data Preparation

Data engineering
Feature engineering

Model Development

Model selection
Model training
Constructing the dataset
Choosing the loss function

Evaluation

Offline metrics
Online metrics

Serving

Prediction pipeline

Other Talking Points

Books

Table of content for volume 1:

Table of content for volume 2:

USE CASES

In the context of using a Large Language Model (LLM) like GPT-3.5 for access and identity management, there are several use cases that can provide valuable insights, enhance efficiency, and improve security. Here are some specific use cases:

Access Rights Verification: Associates can query the LLM to verify their current access rights and entitlements. For example:

"What systems can I access?"
"Do I have permission to view this document?"
"Am I authorized to modify this record?"

Access Request Guidance: When associates need to request additional access rights or permissions, the LLM can provide guidance:

"What steps do I need to follow to request access to the sales database?"
"Can you help me draft an access request for the financial reports?"

Policy Interpretation: Associates often need help understanding company policies related to access and entitlements:

"What is the policy for accessing confidential customer data?"
"Can you explain the access control policy for the HR portal?"

Policy Violation Alerts: The LLM can monitor access requests and identify potential policy violations, triggering alerts:

"Notify me if any access requests violate the principle of least privilege."
"Alert me if an employee tries to access a restricted resource."

User Training and Onboarding: During employee onboarding, the LLM can provide information about access and entitlements:

"Can you explain the different user roles and their associated permissions?"
"What training materials are available for understanding access management?"

Access Audit Trail: Associates can query the LLM to retrieve information about their past access history:

"Can you provide a log of my access activities for the last month?"
"Show me a summary of the resources I accessed last quarter."

Access Recommendations: The LLM can recommend access rights based on historical data and patterns:

"Based on your role, here are the typical access rights you should have."
"Would you like me to suggest the access permissions for this project?"

Regulatory Compliance: The LLM can assist in understanding and applying access controls to ensure compliance with industry regulations:

"What access controls are required for handling healthcare data under HIPAA?"
"Can you help me configure access rights to comply with GDPR?"

These use cases demonstrate how an LLM can play a role in enhancing access and identity management processes by providing quick and accurate information, assisting with policy compliance, and helping associates make informed decisions regarding access requests and entitlements. However, it's crucial to consider the security, privacy, and ethical implications of implementing such a system, and to ensure that it aligns with your organization's policies and regulatory requirements.

poetry new --src openai
cd openai

poetry env list
poetry env use python3
poetry config --list
poetry show


poetry lock
poetry install
poetry update
poetry run pytest
poetry export --output requirements.txt

pdoc --http HOST:PORT <filename.py>
pdoc --http localhost:8080 src/llm/test.py
pdoc --html <filename.py>
pdoc --html src/llm/test.py

git commit -m 'add pre-commit examples' --no-verify

In this template, we use five different plugins that are specified in .pre-commit-config.yaml . They are:

black — formats Python code
flake8 — checks the style and quality of your Python code
isort — automatically sorts imported libraries alphabetically and separates them into sections and types.
mypy — checks static type
nbstripout — strips output from Jupyter notebooks

To check, I use the command

cat -e -t -v makefile_name

It shows the presence of tabs with ^I and line endings with $. Both are vital to ensure that dependencies end properly and tabs mark the action for the rules so that they are easily identifiable to the make utility.

(Install a Package With Poetry)

poetry add requests
poetry add black --group dev

ChatGPT, developed by OpenAI, is a powerful tool used for various applications, including chatbots, content generation, and customer service. Its strength lies in generating human-like text based on the prompts it receives. In this tutorial, we will delve into the art and science of Prompt Engineering - crafting precise and effective prompts to get the best responses from ChatGPT. We decided to focus on ChatGPT Prompt Engineering because it's a crucial skill when working with language models. Understanding how to create effective prompts leads to more accurate, focused, and useful responses.

Setup API

https://help.openai.com/en/articles/5112595-best-practices-for-api-key-safety

Prompt Engineering

https://github.com/openai/openai-cookbook/blob/main/how_to_work_with_large_language_models.md

Rate Limits

https://github.com/openai/openai-cookbook/blob/main/examples/How_to_handle_rate_limits.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/api_request_parallel_processor.py https://github.com/openai/openai-cookbook/blob/main/examples/Unit_test_writing_using_a_multi-step_prompt.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Unit_test_writing_using_a_multi-step_prompt_with_older_completions_API.ipynb

Q&A (56)

https://github.com/openai/openai-cookbook/blob/main/examples/Question_answering_using_a_search_API.ipynb

Summarization (56)

Content Generation (53)

Semantic Search (26)

https://github.com/openai/openai-cookbook/blob/main/examples/Semantic_text_search_using_embeddings.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Question_answering_using_embeddings.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Recommendation_using_embeddings.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Code_search.ipynb

Content Extraction (21)

https://github.com/openai/openai-cookbook/blob/main/examples/Entity_extraction_for_long_documents.ipynb

Embeddings

https://github.com/openai/openai-cookbook/blob/main/examples/Customizing_embeddings.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Embedding_Wikipedia_articles_for_search.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Embedding_long_inputs.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Regression_using_embeddings.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/User_and_product_embeddings.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Zero-shot_classification_with_embeddings.ipynb

Clustering

https://github.com/openai/openai-cookbook/blob/main/examples/Clustering.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Clustering_for_transaction_classification.ipynb

Others

https://github.com/openai/openai-cookbook/blob/main/examples/Fine-tuned_classification.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/How_to_build_a_tool-using_agent_with_Langchain.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/How_to_call_functions_for_knowledge_retrieval.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/How_to_call_functions_with_chat_models.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/How_to_format_inputs_to_ChatGPT_models.ipynb https://github.com/openai/openai-cookbook/blob/main/examples/Multiclass_classification_for_transactions.ipynb https://github.com/openai/openai-python

What are embeddings?

OpenAI’s text embeddings measure the relatedness of text strings. Embeddings are commonly used for:

Search (where results are ranked by relevance to a query string)
Clustering (where text strings are grouped by similarity)
Recommendations (where items with related text strings are recommended)
Anomaly detection (where outliers with little relatedness are identified)
Diversity measurement (where similarity distributions are analyzed)
Classification (where text strings are classified by their most similar label) An embedding is a vector (list) of floating point numbers. The distance between two vectors measures their relatedness. Small distances suggest high relatedness and large distances suggest low relatedness.

References

https://github.com/dair-ai/Prompt-Engineering-Guide

https://github.com/openai/openai-cookbook/blob/main/examples/Question_answering_using_embeddings.ipynb

https://github.com/openai/openai-cookbook/blob/main/examples/vector_databases/Using_vector_databases_for_embeddings_search.ipynb

https://www.datacamp.com/tutorial/a-beginners-guide-to-chatgpt-prompt-engineering

https://learn.microsoft.com/en-us/azure/app-service/# llm

https://github.com/khuyentran1401/data-science-template/blob/dvc-poetry/README.md

https://towardsdatascience.com/how-to-structure-a-data-science-project-for-readability-and-transparency-360c6716800

https://towardsdatascience.com/4-pre-commit-plugins-to-automate-code-reviewing-and-formatting-in-python-c80c6d2e9f5

https://stanford-cs324.github.io/winter2022/lectures/

https://github.com/Ryota-Kawamura/Generative-AI-with-LLMs/tree/main

https://courses.d2l.ai/zh-v2/

https://zh-v2.d2l.ai/chapter_introduction/index.html

https://www.coursera.org/instructor/htlin

李宏毅2023春机器学习课程 https://www.bilibili.com/video/BV1TD4y137mP/?spm_id_from=333.337.search-card.all.click&vd_source=dac84efc7251c2dbebbcdf001659dc53

https://huggingface.co/getvector/earnings-transcript-summary

https://huggingface.co/soleimanian/financial-roberta-large-sentiment

https://huggingface.co/nickmuchi/quantized-optimum-finbert-tone

https://huggingface.co/sentence-transformers/all-mpnet-base-v2

https://huggingface.co/philschmid/flan-t5-base-samsum

goldin2008 / llm Goto Github PK

llm's Introduction