The errata list is a list of errors and their corrections that were found after the product was released.
The following errata were submitted by our customers and have not yet been approved or disproved by the author or editor. They solely represent the opinion of the customer.
Color Key: Serious technical mistake Minor technical mistake Language or formatting error Typo Question Note Update
| Version | Location | Description | Submitted by | Date submitted |
|---|---|---|---|---|
| Printed | Page 9 output of the example |
Output doesn't follow the specified format. Should include "Product description:" and "Product names:" labels as shown in the examples, not just the comma-separated names. Not sure if this is an error or indication that the model did not follow instruction. |
Anonymous | Feb 12, 2026 |
| Printed | Page 42 6th paragraph |
Vectors are described as "multi-dimensional arrays." The intended meaning is likely "high-dimensional vectors". |
Anonymous | Feb 12, 2026 |
| Printed | Page 44 diagram |
Diagram shows BERT encoder and GPT decoder as connected, implying they work together, which is misleading. The caption should say this is the original Transformer architecture that BERT and GPT are derived from — BERT uses only the encoder stack (left), GPT uses only the decoder stack (right), and they operate independently, not together. |
Anonymous | Feb 12, 2026 |
| Printed | Page 49 5th paragraph |
"A reward model is then trained to predict which outputs the labelers would prefer." implies the model compares/selects between outputs. The reward model actually scores individual responses independently — comparison happens by comparing separate scores. |
Anonymous | Feb 12, 2026 |
| Printed | Page 53 section Leveraging Quantization and LoRA |
Misleading, implies LoRA optimizes inference/running efficiency. Need clarification that LoRA makes fine-tuning efficient by training small additional matrices while freezing original weights. It doesn't modify the architecture itself or improve runtime performance. |
Anonymous | Feb 12, 2026 |
| Printed | Page 69 4th paragraph, 3rd bullet point |
"Running a serious of custom checks..." should probably be "Running a series of custom checks..." |
Velimir Graorkoski | Dec 30, 2024 |
| Printed | Page 138 1st paragraph under the LangChain Evals section, 2nd paragraph overall |
"*, which can be directly be logged in their LangSmith*" should be "*, which can be directly logged in their LangSmith*" |
Velimir Graorkoski | Jan 02, 2025 |
| Printed | Page 202 2nd paragraph, 7th point |
The last sentence is not finished: "* and indices (the indices" |
Velimir Graorkoski | Jan 06, 2025 |