Abbreviations¶
This page lists high-frequency technical abbreviations used throughout the book for quick lookup across parts and chapters.
General Abbreviations¶
Table FM-1: General Abbreviations
| Abbreviation | Full Name | Description | Main Locations |
|---|---|---|---|
| A100 | NVIDIA A100 GPU | NVIDIA A100 accelerator | Part I, Part X, Part XI |
| AI | Artificial Intelligence | Artificial intelligence | Whole book |
| AGI | Artificial General Intelligence | Artificial general intelligence | Part I, Part XI |
| API | Application Programming Interface | Application programming interface | Part I, Part IV, Part X, Part XI |
| ANN | Approximate Nearest Neighbor | Approximate nearest-neighbor retrieval | Part I, Part VII |
| ASR | Automatic Speech Recognition | Automatic speech recognition | Part III |
| BM25 | Best Matching 25 | Classic sparse retrieval ranking method | Part I, Part VII |
| CI/CD | Continuous Integration / Continuous Deployment | Continuous integration and continuous deployment | Part I, Part II, Part VIII |
| CPU | Central Processing Unit | Central processing unit | Part I, Part II |
| CSV | Comma-Separated Values | Comma-separated text format | Part I |
| ETL | Extract, Transform, Load | Extract, transform, and load | Part I, Part VIII |
| GDPR | General Data Protection Regulation | European Union data-protection regulation | Part I, Part XI, Part XIV |
| GPU | Graphics Processing Unit | Graphics processing unit | Part I, Part III, Part X, Part XI |
| GUID | Globally Unique Identifier | Globally unique identifier | Part II |
| HDFS | Hadoop Distributed File System | Hadoop distributed file system | Part I, Part X |
| H100 | NVIDIA H100 GPU | NVIDIA H100 accelerator | Part I |
| I/O | Input/Output | Input and output operations | Part I, Part II |
| JSON | JavaScript Object Notation | Structured data interchange format | Part I, Part III, Part IV, Part X, Part XI |
| JSONL | JSON Lines | Line-delimited JSON text format | Part I, Part III, Part IV, Part X |
| KPI | Key Performance Indicator | Key performance indicator | Part IV, Part VIII |
| LLM | Large Language Model | Large language model | Whole book |
| MLOps | Machine Learning Operations | Machine-learning engineering and operations system | Part I, Part VIII |
| NER | Named Entity Recognition | Named-entity recognition | Part I, Part II |
| NIST AI RMF | NIST Artificial Intelligence Risk Management Framework | NIST framework for managing artificial-intelligence risks | Part IV, Part XIV, Appendix B |
| NLP | Natural Language Processing | Natural-language processing | Part III, Part IV, Part V |
| Portable Document Format | Portable document format | Part I, Part III, Part VII, Part X, Part XI | |
| PII | Personally Identifiable Information | Personally identifiable information | Part II, Part IX, Part X |
| ROI | Return on Investment | Return on investment | Part I, Part VIII |
| SLA | Service Level Agreement | Service-level agreement | Part I, Part IV, Part VIII |
| SOPs | Standard Operating Procedures | Standard operating procedures | Part IV, Part VIII |
| SQL | Structured Query Language | Structured query language | Part I, Part IV, Part XI |
| TPU | Tensor Processing Unit | Tensor processing unit | Part I |
| UTF-8 | 8-bit Unicode Transformation Format | 8-bit Unicode transformation format | Part II |
Data Engineering and Platforms¶
Table FM-2: Data Engineering and Platforms
| Abbreviation | Full Name | Description | Main Locations |
|---|---|---|---|
| DataOps | Data Operations | Data operations and data-engineering operations system | Part II, Part VIII, Part X |
| BPE | Byte Pair Encoding | Subword tokenization algorithm | Part I, Part II, Part XIV |
| DOM | Document Object Model | Document object model | Part III, Part XI |
| DVC | Data Version Control | Data version control tool or method | Part I, Part II, Part VIII |
| FAISS | Facebook AI Similarity Search | Vector similarity search library | Part VII |
| FastText | FastText | Lightweight text representation and classification tool | Part II |
| LakeFS | LakeFS | Version management system for data lakes | Part I, Part VIII |
| LSH | Locality-Sensitive Hashing | Approximate-similarity hashing method | Part I, Part II |
| MATTR | Moving-Average Type-Token Ratio | Moving-average type-token ratio | Part II |
| MFU | Model FLOPs Utilization | Model FLOPs utilization | Part II |
| MinHash | Min-wise Independent Permutations Hashing | Approximate deduplication method based on min-wise hashing | Part I, Part II, Part XI |
| OOM | Out Of Memory | Out-of-memory error | Part I, Part II |
| PPL | Perplexity | Perplexity metric | Part I, Part II, Part IV |
| RDMA | Remote Direct Memory Access | Remote direct memory access | Part I |
| ReDoS | Regular Expression Denial of Service | Regular-expression denial-of-service risk | Part II |
| TTR | Type-Token Ratio | Type-token ratio, a diversity metric | Part II |
| WARC | Web ARChive | Web archive format | Part II |
| WebDataset | WebDataset | Data packaging format and tool for large-scale training | Part II, Part III |
Training, Alignment, and Reasoning¶
Table FM-3: Training, Alignment, and Reasoning
| Abbreviation | Full Name | Description | Main Locations |
|---|---|---|---|
| CoT | Chain-of-Thought | Chain-of-thought reasoning | Part VI, Part X, Part XI |
| DPO | Direct Preference Optimization | Direct preference optimization | Part IV, Part XI |
| LoRA | Low-Rank Adaptation | Low-rank adaptation fine-tuning method | Part XI |
| PPO | Proximal Policy Optimization | Proximal policy optimization | Part IV, Part XI |
| PEFT | Parameter-Efficient Fine-Tuning | Parameter-efficient fine-tuning methods | Part XI |
| PRM | Process Reward Model | Process reward model | Part IV, Part VI, Part X, Part XI |
| QA | Quality Assurance / Question Answering | Quality assurance or question answering, according to context | Part IV, Part VII, Part X |
| RAG | Retrieval-Augmented Generation | Retrieval-augmented generation | Part VII, Part X |
| RL | Reinforcement Learning | Reinforcement learning | Part IV, Part XI |
| RLAIF | Reinforcement Learning from AI Feedback | Reinforcement learning from AI feedback | Part IV |
| RLHF | Reinforcement Learning from Human Feedback | Reinforcement learning from human feedback | Part IV, Part XI |
| RM | Reward Model | Reward model | Part IV |
| ROUGE-L | Recall-Oriented Understudy for Gisting Evaluation - Longest Common Subsequence | Text-similarity metric based on the longest common subsequence | Part II, Part IV |
| SFT | Supervised Fine-Tuning | Supervised fine-tuning | Part IV, Part X, Part XI |
Multimodality and Vision¶
Table FM-4: Multimodality and Vision
| Abbreviation | Full Name | Description | Main Locations |
|---|---|---|---|
| BBox | Bounding Box | Bounding box | Part III, Part X, Part XI |
| ChartQA | Chart Question Answering | Chart question-answering task or dataset | Part III, Part XI |
| CLIP | Contrastive Language-Image Pre-training | Contrastive image-text pre-training model | Part III, Part XI |
| CLIP-Score | CLIP Score | Image-text relevance score based on CLIP | Part XI |
| COCO | Common Objects in Context | General object-detection and image-captioning dataset | Part III, Part X |
| DINO | DEtection TRansformer with Improved deNoising anchOr boxes | Detection model family, often used in Grounding DINO contexts | Part III, Part XI |
| DocVQA | Document Visual Question Answering | Document visual question-answering task or dataset | Part XI |
| Grounding | Visual Grounding | Visual grounding or alignment task | Part III, Part X, Part XI |
| IoU | Intersection over Union | Object-detection overlap metric | Part III |
| LLaVA | Large Language and Vision Assistant | Multimodal large model and data format name | Part X, Part XI |
| OCR | Optical Character Recognition | Optical character recognition | Part III, Part VII, Part X, Part XI |
| OCR-Rich | OCR-Rich Data | Image or document data rich in OCR information | Part XI |
| SSIM | Structural Similarity Index Measure | Structural similarity metric | Part XI |
| ViT | Vision Transformer | Vision Transformer encoder | Part XI |
| VLM | Vision-Language Model | Vision-language model | Part III, Part XI |
| VQA | Visual Question Answering | Visual question answering | Part III, Part XI |
| XML | eXtensible Markup Language | Extensible markup language | Part I, Part III |
| YOLO | You Only Look Once | Object-detection model family | Part III |
Evaluation, Compliance, and Governance¶
Table FM-5: Evaluation, Compliance, and Governance
| Abbreviation | Full Name | Description | Main Locations |
|---|---|---|---|
| AGI-Eval | AGI Evaluation | Evaluation benchmark for general-intelligence capabilities | Part XI |
| DPIA | Data Protection Impact Assessment | Data protection impact assessment | Part IX |
| GSM8K | Grade School Math 8K | Grade-school math reasoning benchmark | Part I, Part XI |
| MCTS | Monte Carlo Tree Search | Monte Carlo tree search | Part XI |
| MMLU | Massive Multitask Language Understanding | Massive multitask language-understanding benchmark | Part I, Part XI |
| MMMU | Massive Multi-discipline Multimodal Understanding and Reasoning | Multidiscipline multimodal understanding and reasoning benchmark | Part XI |
| NSFW | Not Safe For Work | Content unsuitable for public or workplace contexts | Part II |
| P99 | 99th Percentile | 99th percentile metric | Part II |
| P99.9 | 99.9th Percentile | 99.9th percentile metric | Part II |
| RoPA | Record of Processing Activities | Record of processing activities | Part IX |
| ABAC | Attribute-Based Access Control | Attribute-based access control | Part IX, Part XI |
| MCP | Model Context Protocol | Protocol for connecting models to tools and contextual resources | Part X, Part XIV, Appendix G |
| NL2SQL | Natural Language to SQL | Translation from natural-language requests to SQL | Part X, Part XIV, Appendix G |
| RBAC | Role-Based Access Control | Role-based access control | Part IX, Part XI |