Learn how large language models like GPT and Gemini work under the hood in plain English.
How Large Language Models Work translates years of expert research on Large Language Models into a readable, focused introduction to working with these amazing systems. It explains clearly how LLMs function, introduces the optimization techniques to fine-tune them, and shows how to create pipelines and processes to ensure your AI applications are efficient and error-free.
In How Large Language Models Work you will learn how to:
Test and evaluate LLMs
Use human feedback, supervised fine-tuning, and Retrieval Augmented Generation (RAG)
Reducing the risk of bad outputs, high-stakes errors, and automation bias Human-computer interaction systems
Combine LLMs with traditional ML
How Large Language Models Work is authored by top machine learning researchers at Booz Allen Hamilton, including researcher Stella Biderman, Director of AI/ML Research Drew Farris, and Director of Emerging AI Edward Raff. They lay out how LLM and GPT technology works in plain language that’s accessible and engaging for all.
what's inside
Customize LLMs for specific applications
Reduce the risk of bad outputs and bias
Dispel myths about LLMs
Go beyond language processing
about the reader
No knowledge of ML or AI systems is required.
AI Reading Assistant
Whole-book reading guide from stratified index samples; jump to passages in the text
Tip the Site
Support this siteYour recognition and a small knowledge-service contribution help keep this technical work open source.Scan the WeChat Pay or Alipay code below. Logged-in and guest visitors can both tip.
WeChat Pay
Alipay
Open WeChat or Alipay and scan. No login required.
Brief outline
【One-Line Pitch】
Learn how large language models like GPT and Gemini work under the hood in plain English.
How Large Language Mode…
【Book Arc】
- **Opening (~0%–12%)**: They lay out how LLM and GPT technology works in plain language that’s accessible and engaging for all.; Sam Wood and Marija Tudor led the production of our cover, and Azra Dedic led the production of our graphics and figures.
- **Early (~12%–35%)**: uld automatically translate between languages.; years from now instead of example code that could be out of date in just a few months.
- **Middle (~35%–65%)**: For this reason, much of the work that used to go into engineering a robust normalization step has been discarded for LLMs today.; he individual digits in that number.
- **Late (~65%–88%)**: ou are using a chatbot as a digital sounding board to spark ideas.; ver, the actual count of cancer cases is constrained to the integers 1, 2, 3, .
- **Ending (~88%–100%)**: For example, we could develop a prompt for a car-selling bot as demonstrated in figure 5.2.; nce MELTON-24 does not exist.
【Key Takeaways】
- **They lay out how LLM a…** (Opening): They lay out how LLM and GPT technology works in plain language that’s accessible and engaging for all.
- **Sam Wood and Marija Tu…** (Opening): Sam Wood and Marija Tudor led the production of our cover, and Azra Dedic led the production of our graphics and figures.
- **We collect all of thes…** (Opening): We collect all of these in a references section at the end of the book, providing easy access to the entire list of resources in one place.
- **uld automatically tran…** (Early): uld automatically translate between languages.
- **years from now instead…** (Early): years from now instead of example code that could be out of date in just a few months.
- **e led to the advanceme…** (Early): e led to the advancements that make ChatGPT possible today.
【Reading Tips】
- Use Passage locations below to jump into the text and set reading anchors
- Chinese and English guides are each generated once, then served from local cache without calling the model again
【Coverage Limits】
Compressed outline without the model (~33 index chunks). Full structured guide needs AI available.
Page 8
y 9 1.6 Generative Pretrained Transformers and friends 10 1.7 Why LLMs perform so well 10 1.8 LLMs in action: The good, bad, and scary 12 2 Tokenizers: How l...
years from now instead of example code that could be out of date in just a few months. 1.3 Introducing how LLMs work Generative AI (GAI or GenAI) is poised t...
to prepare your data for the combination of model and goal. Tokenization is the feature engineering of LLMs; it is critically essential because tokens are th...
to a numeric form that deep learning algorithms can work on. Then, the LLM converts this numeric representation back into a new token. This cycle repeats ite...
ultiplication makes attention efficient when implemented on GPUs because they can perform many multiplication operations in parallel. The softmax function im...
her parameter values correspond to areas with a lower loss. Also, notice that in the second step in figure 4.5, the ball gets stuck. While it is evident that...
ge number of instructions and responses as training data. 4.5 Is bigger better? In 2019, Rich Sutton coined the term “the bitter lesson” to describe his expe...
achieve. For a chatbot like ChatGPT, the environment is the conversation with a user, and the actions are the infinite possible texts that ChatGPT might comp...
Support this siteYour recognition and a small knowledge-service contribution help keep this technical work open source.
Scan the WeChat Pay or Alipay code below. Logged-in and guest visitors can both tip.
WeChat PayAlipay
Open WeChat or Alipay and scan. No login required.
Add Tag
Enter tag name (max 50 characters)
Share E-Book
How Large Language Models Work (Edward Raff, Drew Farris, Stella Biderman)（Z-Library）
Scan QR code with your phone to access
Copy the link or scan the QR code to access this e-book on your phone
Share E-Book via Email
Please enter email address
Donation Statistics
¥.00
Total Donations
0
Donation Count
How Large Language Models Work (Edward Raff, Drew Farris, Stella Biderman)（Z-Library）
Find Your Favorite Books
Only registered users can comment after logging in. Comments need to be reviewed by administrators before being displayed
Loading comments...
Reply to Comment
Edit Comment