LLMs

• L

  • Webs
    • pu is enough for this LLM? -
    • leaderboards
      • artificialanalysis.ai -
      • web.lmarena.ai w
  • summarizer
    • talk to transcript w
    • https://notegpt.io/youtube-transcript-generator

• Cloud-based services

  • Grok - good recent online research
  • Microsoft
  • Google/Gemini
  • GPT
  • llama
  • deepseek
  • Claude
  • mistral
  • phind.com
  • alibaba
  • amazon

• Models

  • types
    • Transformer-Based Models
      • preferred over contextual embeddings
      • designed to understand context and semantics better than traditional models
      • like BERT, GPT, and T5
    • Hybrid Retrieval Models
      • keyword-based search & vector-based retrieval
    • meh
      • Contextual Embeddings
        • capture contextual information by considering the surrounding text
        • like ELMo and contextual BERT variants
  • Most popular
    • foss
      • mistral
      • LLAMA - text & images
        • ollama
        • can be run even on a 4gb RAM laptop
        • binaries are being shared but not the source code, and they still call it open source… liars, but still better than ""Open""Ai
    • non foss
      • gemini
      • GPT4 - conversational
      • claude 3
      • grok-1
  • naming convention
    • math or MoE - Mixture of Experts model
    • formats
      • safetensors - secure file format to avoid malware
      • GGML - newer GGUF
        • binary
        • support diff quantization schemes running on CPU in a single file
    • Quantization methods
      • Exl2 - best optimization but only for nvidia
      • AWQ - round weights
      • GPTQ - worst one
  • HT run smoother
    • quantarization - less accuracy, but less RAM/VRAM usage
    • parallelism - share resources between GPUs and CPU
    • LocalAI