A Reddit thread discussing preferred local Large Language Model (LLM) setups for tasks like summarizing text, coding, and general use. Users share their model choices (Gemma, Qwen, Phi, etc.) and frameworks (llama.cpp, Ollama, EXUI) along with potential issues and configurations.
| Model | Use Cases | Size (Parameters) | Approx. VRAM (Q4 Quantization) | Approx. RAM (Q4) | Notes/Requirements |
|---|---|---|---|---|---|
| Gemma 3 (Meta) | Summarization, conversational tasks, image recognition, translation, simple writing | 3B, 4B, 7B, 8B, 12B, 27B+ | 2-4GB (3B), 4-6GB (7B), 8-12GB (12B) | 4-8GB (3B), 8-12GB (7B), 16-24GB (12B) | Excellent performance for its size. Recent versions have had memory leak issues (see Reddit post – use Ollama 0.6.6 or later, but even that may not be fully fixed). QAT versions are highly recommended. |
| Qwen 2.5 (Alibaba) | Summarization, coding, reasoning, decision-making, technical material processing | 3.5B, 7B, 72B | 2-3GB (3.5B), 4-6GB (7B), 26-30GB (72B) | 4-6GB (3.5B), 8-12GB (7B), 50-60GB (72B) | Qwen models are known for strong performance. Coder versions specifically tuned for code generation. |
| Qwen3 (Alibaba - upcoming) | General purpose, likely similar to Qwen 2.5 with improvements | 70B | Estimated 25-30GB (Q4) | 50-60GB | Expected to be a strong competitor. |
| Llama 3 (Meta) | General purpose, conversation, writing, coding, reasoning | 8B, 13B, 70B+ | 4-6GB (8B), 7-9GB (13B), 25-30GB (70B) | 8-12GB (8B), 14-18GB (13B), 50-60GB (70B) | Current state-of-the-art open-source model. Excellent balance of performance and size. |
| YiXin (01.AI) | Reasoning, brainstorming | 72B | ~26-30GB (Q4) | ~50-60GB | A powerful model focused on reasoning and understanding. Similar VRAM requirements to Qwen 72B. |
| Phi-4 (Microsoft) | General purpose, writing, coding | 14B | ~7-9GB (Q4) | 14-18GB | Smaller model, good for resource-constrained environments, but may not match larger models in complexity. |
| Ling-Lite | RAG (Retrieval-Augmented Generation), fast processing, text extraction | Variable | Varies with size | Varies with size | MoE (Mixture of Experts) model known for speed. Good for RAG applications where quick responses are important. |
Key Considerations:
- Quantization: The VRAM and RAM estimates above are based on 4-bit quantization (Q4). Lower quantization (e.g., Q2) will reduce memory usage further, but may impact quality. Higher quantization (e.g., Q8, FP16) will increase quality but require significantly more memory.
- Frameworks: Popular frameworks for running these models locally include:
- llama.cpp: Highly optimized for CPU and GPU, especially on Apple Silicon.
- Ollama: Simplified setup and management of LLMs. (Be aware of the Gemma 3 memory leak issue!)
- Text Generation WebUI (oobabooga): Web-based interface with many features and customization options.
- Hardware: A dedicated GPU with sufficient VRAM is highly recommended for decent performance. CPU-only inference is possible but can be slow. More RAM is generally better, even if the model fits in VRAM.
- Context Length: The "40k" context mentioned in the Reddit post refers to the maximum number of tokens (words or sub-words) the model can process at once. Longer context lengths require more memory.
Yoyak is a CLI tool that uses LLM to summarize and translate web pages. It supports various models and provides shell completion scripts.
A tool to download, transcribe, summarize, and chat with media files like videos, audio, documents, web articles, and books, all locally and automated.
The article explores how smaller language models like the Meta 1 Billion model can be used for efficient summarization and indexing of large documents, improving the performance and scalability of Retrieval-Augmented Generation (RAG) systems.
The article discusses Google's new AI tool Gemini and its email summarization feature, which helps manage inbox anxiety by summarizing daily emails.
This project creates bulleted notes summaries of books and other long texts using Python and language models, splitting documents into chunks for more granular summaries and question-based analyses.
A tool to transcribe and summarize videos from multiple sources using AI models in Google Colab or locally.
"Generate 5 essential questions that, when answered, capture the main points and core meaning of the text. Focus on questions that:
Address the central theme or argument
Identify key supporting ideas
Highlight important facts or evidence
Reveal the author's purpose or perspective
Explore any significant implications or conclusions
Phrase the questions to encourage comprehensive yet concise answers. Present only the questions, numbered and without any additional text."
The article explains semantic text chunking, a technique for automatically grouping similar pieces of text to be used in pre-processing stages for Retrieval Augmented Generation (RAG) or similar applications. It uses visualizations to understand the chunking process and explores extensions involving clustering and LLM-powered labeling.
In this post, we'll explore how to use Hugging Face's Pipeline API to generate summaries with a zero-shot model and train a summarization model on the arXiv dataset. We'll also evaluate the trained model and compare it to the simple heuristic we developed in the previous post.
