This is an old revision of the document!
Large Language Model (LLM)
A Large Language Model (LLM) is an artificial intelligence model trained on vast amounts of text data to understand and generate human language in a coherent and context-aware way.
What is an LLM?
An LLM uses deep learning, particularly the transformer architecture, to process and predict language. It excels at predicting the next word in a sentence, enabling it to perform tasks such as:
- Text generation
- Translation
- Question answering
- Summarization
- Code completion
Training
LLMs are trained on large-scale datasets (books, articles, web pages, code) using supervised and unsupervised methods. Training requires immense computational resources.
Parameter Scale
Models are measured by the number of parameters (internal adjustable weights). Examples:
- GPT-3 → 175 billion parameters
- GPT-4 → unknown exact size (larger and multimodal)
- LLaMA 2 → 7B to 65B parameters
Key Examples
| Model | Developer | Year | Notes |
|---|---|---|---|
| GPT-3/4 | OpenAI | 2020–25 | Powers ChatGPT |
| Claude | Anthropic | 2023–25 | Emphasis on alignment and safety |
| Gemini | Google DeepMind | 2023–25 | Integrated with Google products |
| LLaMA | Meta | 2023–25 | Open-source, academic use |
| Mistral | Mistral.ai | 2023–25 | Efficient, performant smaller models |
Strengths
- Versatile language tasks
- Learns from context
- Multilingual support
- Can be fine-tuned for specific domains
Limitations
- May produce incorrect or biased outputs
- No true understanding (statistical patterns only)
- High computational and energy cost
- Requires oversight in critical applications
Applications
- Chatbots and virtual assistants
- Automated writing and summarization
- Legal and medical drafting support
- Programming assistance
- Research and data extraction