Best of LLM — January 2025

  1. 1
    Article
    Avatar of mlmMachine Learning Mastery·1y

    7 Next-Generation Prompt Engineering Techniques

    Mastering prompt engineering is essential in optimizing large language models like ChatGPT and Gemini. Techniques such as meta prompting, least-to-most prompting, multi-task prompting, role prompting, task-specific prompting, program-aided language models, and chain-of-verification prompting can significantly enhance the performance and efficiency of LLMs. Each method has unique benefits and challenges, but collectively, they improve the accuracy and relevance of the generated content.

  2. 2
    Article
    Avatar of taiTowards AI·1y

    Lets Build Simple RAG Application

    Large Language Models (LLMs) have significantly advanced technology interactions but possess limitations like the inability to access real-time information, affecting applications requiring current data. Enhancements using techniques like in-context learning are discussed, particularly for building effective RAG applications using Langchain.

  3. 3
    Article
    Avatar of minimaxirMax Woolf's Blog·1y

    Can LLMs write better code if you keep asking them to “write better code”?

    The post explores the concept of iteratively improving code quality using LLMs like Claude 3.5 Sonnet by repeatedly asking the model to 'write better code.' Initial results of using basic prompts to refine a Python code sample show significant performance improvements, demonstrating an iterative speedup from a baseline of 657 milliseconds to code that runs in about 6 milliseconds after several iterations. Further experiments using prompt engineering, which explicitly guide the LLM with detailed instructions, yield even more optimized code. However, the process also reveals the limitations of LLMs, highlighting the necessity for human oversight to ensure correctness and manage subtle bugs.

  4. 4
    Article
    Avatar of hnHacker News·1y

    The 2025 AI Engineering Reading List

    This post presents a curated reading list for AI engineers in 2025, covering around 50 essential papers, models, and blogs across various fields in AI engineering such as LLMs, benchmarks, prompting, retrieval augmented generation (RAG), agents, code generation, vision, voice, and more. The goal is to provide practical, relevant resources for those starting from scratch. The reading list is sectioned into ten fields, with each section recommending seminal papers and providing context on their significance and application.

  5. 5
    Article
    Avatar of codingaddaCoding Adda·1y

    Create your Own AI Code Commentor and Run it on your machine using Ollama

    Learn how to create and run your own AI code comment generator using Ollama and VsCode. This project leverages NodeJs and requires a machine capable of using LLMs. Follow the setup steps to install Ollama, pull a model, clone the project repository, and generate comments for your code files with a simple command.

  6. 6
    Article
    Avatar of awegoAwesome Go·1y

    Scaling LLMs with Golang: How we serve millions of LLM requests

    Using Golang for production deployments of LLMs has proven efficient for handling millions of requests. Key benefits include its type safety, concurrency with goroutines, and composable interfaces. Despite Python's dominance in ML development, combining it with Golang allows leveraging both languages' strengths for a robust and scalable system.

  7. 7
    Article
    Avatar of milanjovanovicMilan Jovanović·1y

    Working with LLMs in .NET using Microsoft.Extensions.AI

    Learn how to integrate Large Language Models (LLMs) into .NET applications using Microsoft.Extensions.AI. This guide covers configuring Ollama for local execution, setting up dependency injection for chat completion, building interactive chats with history, summarizing articles, and applying smart categorization. It also details switching between various LLM providers like Azure OpenAI and OpenAI for flexible development and production solutions.

  8. 8
    Article
    Avatar of hnHacker News·1y

    bodo-run/yek: A fast tool to read text-based files in a repository or directory, chunk them, and serialize them for LLM consumption.

    Yek is a Rust-based tool designed to read and serialize text-based files from repositories or directories for LLM consumption. It skips files based on .gitignore rules, infers file importance from Git history, and allows content chunking by token count or byte size. Yek supports multi-directory processing, custom ignore patterns, and file priority rules via a `yek.toml` configuration file. It offers superior speed, being 230x faster than repomix in benchmarking tests.

  9. 9
    Article
    Avatar of shippingbytesshipping bytes·1y

    Stanford CS229 I Machine Learning I Building Large Language Models (LLMs)

    The post discusses the importance of learning about large language models (LLMs) and their impact on the environment. It highlights a Stanford CS229 presentation on YouTube that covers specialized training, transformation, and data quality in generative AI. The author emphasizes the ongoing effort to make these technologies more efficient and sustainable.

  10. 10
    Article
    Avatar of mlmMachine Learning Mastery·1y

    RAG Hallucination Detection Techniques

    Large language models (LLMs) can provide factually incorrect answers, often termed hallucinations. Retrieval augmented generation (RAG) mitigates this by retrieving data from a knowledge base, but hallucinations can still occur. The post discusses techniques to detect these hallucinations using metrics from the DeepEval library, the G-Eval framework, and RAG-specific metrics like faithfulness. Practical examples include the installation and usage with code snippets that evaluate the outputs for accuracy, consistency, and relevance.

  11. 11
    Article
    Avatar of taiTowards AI·1y

    Building Graph RAG for structured and unstructured data.

    RAG (Retriever-Augmented Generation) architecture helps solve the issue of missing contextualization in LLMs (Large Language Models) without the need for fine-tuning. While Vector RAGs offer some contextualization, graph-based RAGs capture more intricate relationships, making them more effective. This post discusses how to build knowledge graphs from both unstructured data (like PDFs) and structured data (like CSVs) using tools such as Langchain and Neo4j. It also outlines steps for extracting text, chunking documents, constructing graphs, and querying the graph databases using LLMs.

  12. 12
    Article
    Avatar of sebastianraschkaSebastian Raschka·1y

    Noteworthy LLM Research Papers of 2024

    The post reviews 12 influential research papers on large language models (LLMs) published throughout 2024. It covers significant advancements and methods, including the Mixture of Experts models, improvements in low-rank adaptation techniques, effective pretraining strategies, and the introduction of new scaling laws. The reviews highlight developments in LLM architectures, optimization techniques, and the use of synthetic data, emphasizing their implications for future LLM research and applications.

  13. 13
    Article
    Avatar of hnHacker News·1y

    jjcm/llmpeg: Uses an llm to generate ffmpeg commands

    The tool 'llmpeg' simplifies the usage of ffmpeg by utilizing a language model (LLM) to generate commands. For example, the command `llmpeg remove audio from exampleVid.mov` removes the audio track from a video file named exampleVid.mov.

  14. 14
    Article
    Avatar of medium_jsMedium·1y

    LangGraph AI agents : Building a Dynamic Order Management System : A Step-by-Step Tutorial

    Learn how to use LangGraph, a library designed for orchestrating complex workflows with Large Language Models (LLMs), to create a dynamic order management system. This tutorial covers setting up the environment, defining workflow nodes, integrating tools and LLMs, and visualizing and testing the workflow. By following the detailed steps, you can build a system capable of placing or canceling orders based on user queries.

  15. 15
    Article
    Avatar of dailydoseofdsDaily Dose of Data Science | Avi Chawla | Substack·1y

    [Hands-on] 100% Local RAG using DeepSeek

    Building AI applications that can work effectively with real-time web data is challenging due to the need for human-like interaction simulation, overcoming site blocks, and ensuring legal compliance. Bright Data offers infrastructure to handle these tasks at scale. Additionally, the post introduces an affordable RAG app built using DeepSeek AI's models, which offer significant cost savings compared to OpenAI. The tutorial covers the setup of the knowledge base, embedding creation, vector database indexing, and custom prompt template for LLM, concluding with a user-friendly interface and future advanced techniques to be discussed.

  16. 16
    Article
    Avatar of mlnewsMachine Learning News·1y

    Creating An AI Agent-Based System with LangGraph: A Beginner’s Guide

    An agent in AI is a system powered by a Large Language Model (LLM) capable of making decisions based on context and using external tools like web searches and databases. LangGraph is a Python library that simplifies the creation of these agents by providing tools to structure workflows with nodes, edges, and state management. This guide walks through building a basic chatbot using LangGraph, initializing an LLM, adding memory and tool integration, and enhancing functionality with a web search tool.

  17. 17
    Article
    Avatar of taiTowards AI·1y

    Scaling LLM Experimentation with SageMaker Pipelines and MLflow

    Large language models (LLMs) are revolutionizing NLP tasks across various industries and often need customization for specific domains. Amazon SageMaker and MLflow offer scalable solutions for fine-tuning and evaluating these models. The post explains how to use SageMaker JumpStart and SageMaker Clarify to evaluate models, and SageMaker Pipelines for comparison. Additionally, it covers using MLflow to track training and evaluation data, and employing Parameter-Efficient Fine-Tuning (PEFT) using the transformers library for customization.

  18. 18
    Article
    Avatar of freecodecampfreeCodeCamp·1y

    How to Use Langbase Memory Agents to Make Any LLM a Conversational AI for Your Docs

    Memory agents can securely link private data to large language models (LLMs) in real time, enabling context-aware interactions with personal documents and enhancing AI functionalities. This tutorial guides on creating AI agents using Langbase memory agents to provide context-sensitive responses while ensuring data security. Use cases include customer support, document search, education, healthcare, and legal compliance.

  19. 19
    Article
    Avatar of taiTowards AI·1y

    AI Mathematicians: How LLMs Are Redefining Mathematics

    Large Language Models (LLMs) like ChatGPT are revolutionizing mathematics by processing vast amounts of data quickly, uncovering unseen patterns, and proposing novel proofs and algorithms. They aid in prime number analysis, provide insights into the Riemann Zeta function, simulate fluid dynamics, solve symbolic mathematics problems, optimize heat transfer designs, and enhance machine learning through mathematical insights. By accelerating mathematical discoveries and expanding research horizons, LLMs are redefining the future of mathematical exploration.

  20. 20
    Article
    Avatar of devtoDEV·1y

    Getting Started with DeepSeek LLM using Ollama locally

    DeepSeek LLM, launched in early 2024, is a language model with 67 billion parameters and bilingual support for English and Chinese. DeepSeek R1, a compact AI model, is optimized for local hardware and excels in reasoning, coding, and technical tasks. Running DeepSeek R1 locally provides benefits in privacy, speed, cost, customization, and offline deployment. The post guides setting up DeepSeek R1 with Ollama, Open WebUI, and Docker, highlighting its superior reasoning capabilities and cost-efficiency.

See all LLM archives