Ai Tools Research
  • 🔲About Ai Tools Research
    • AI Adoption Consultation & Training Services
      • Build an Ethical Platform for AI
      • Corporate AI Safety Guidance
        • Why AI ethics?
        • Principle of Discriminatory Non-Harm
      • Training Services
        • Training Courses
          • The Full Applied GenAI Curriculum
          • Gemini Training Courses
        • Corpus Creator
    • LLM Performance Benchmarks
      • LLM Benchmarks Master List
        • LLM Benchmark Categories
      • Size, Quality and Cost of Training Data in LLM’s
      • LLM Benchmarks and Tasks
        • Chatbot Arena
          • How to use ChatBot Battle Arena
        • LLM Capabilities Test
        • Benchmarking of Large Language Models
        • F1 score (F-measure one)
        • Foundation Model Transparency Index
        • TriviaQA (5-Shot)
        • QuALITY (5-Shot)
        • Codex P@1 (0-Shot)
        • RACE-H (5-Shot)
        • ARC-Challenge (5-Shot)
        • GSM8k (0-Shot CoT)
        • MMLU (Massive Multitask Language Understanding)
          • MMLU (5-Shot CoT)
          • MMLU
        • SOTA in Artificial Intelligence
        • Context Window Size
      • Alpaca 2.0 Evall
    • Youtube Videos Directory
      • T981g - Post AGI Economics
        • Post-AGI Economics (Part 1)
        • Post-AGI Economics (Part 2)
      • T981m - Path to AGI -Debates, Definitions, and the Future Ahead
      • AI Explained: Tracing Intelligence from Mammals to Machines
      • Artificial Super Intelligence (ASI)
      • The Path From AI to ASI
        • T701 - The Stages of AI Evolution Final
        • T703 - A Decentralized Path to AGI and ASI
        • What is Artificial Intelligence (Ai) and what are Ai Neural Networks
      • AI Trends of Next Two Years
      • AI in Programming: Threat to Jobs or a Tool for Enhancement?
    • Frequently Asked Questions (FAQs)
      • A Typology of AI
        • AI Taxonomy
        • Difference between Symbolic Language and Natural Language
      • Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs) and Diffusion Models
      • What is 'Latent Space' in Image Generation?
      • What is LoRA and How does LoRA work
      • What is Gradient Descent?
      • What are Vector databases are and how they work?
      • What is 'Inpainting' & 'Outpainting' ?
      • What is 'DPO' in LLM Training?
      • What is 'One-Shot' Learning?
      • FAQs on LLM Training and Data Labelling
        • LLMs Main Concepts Explained
        • LLM Evaluation
        • Building Datasets
          • Kili Technology
          • Text Annotation
          • SmartOne Data Labeling Services
        • What is an 'Uncensored LLM'
          • Open Source AI Will Outcompete Google and OpenAI
        • What are Parameters in LLMs?
        • Parameters vs Tokens in LLMs?
        • What are Model Weights?
        • What is 'Inference Cost'?
        • Training Corpus and Datasets
          • Open-Sourced Training Datasets for LLMs
          • Datasets List from Dr. Alan Thompson
          • Corpus Used by Large Language Models (LLMs) for Different Applications
        • What are 'Tokens' ?
        • What are Token Limits?
        • What Are Context Windows?
          • Is 'Context Window' and 'Token Limit' the same?
          • LLM with Largest Context Window?
          • Tokens, Words and Pages
        • How to Fine Tune LLMs?
        • Case Study of Fine-Tuning an LLM
          • How to'Tokenize' the data?
          • What are OOV Tokens?
          • Arrays and Tensors
            • Difference between Numpy and TensorFlow
            • NumPy or TensorFlow?
          • Setting Training Parameters
          • Do you need to adjust Model Weights during training?
          • Why do some models have 'Open Weights' and others 'Closed Weights'?
          • Inference Parameters vs. Training Parameters
          • AutoTrain LLMs at HuggingFace
            • 'TheBloke' at Huggingface?
            • What are GGUF Format Model Files?
            • What does 'Mistral 7B quantized in 4-bit with AutoAWQ' mean?
            • What does a 'Quantized Version' of an LLM mean?
        • What is "RAG," (Retrieval-Augmented Generation)?
          • 𝐑𝐀𝐆 𝐅𝐫𝐨𝐦 𝐒𝐜𝐫𝐚𝐭𝐜𝐡
        • What does "Release Base, Instruct and Reward Model" mean?
    • Articles and Transcripts
      • What are AI Agents
        • Tools Developers Can Use to Build Agents
      • Fine Tuning LLMs for Use Cases
        • Fine Tuning for Risk Mgmt.
        • LoRA - Low-Rank Adaptation
      • Natural Language Processing Guides
        • Basic Guide-Natural Language Processing (NLP)
        • Comprehensive Guide to NLP [2023 Edition]
      • SLM's vs. LLM's
      • Python Tools
        • API Protocols
        • Top Python Libraries and Frameworks for Various Domains
        • 5 Layers of Software Development
        • LM Studio -Open Source LLM Installer
          • LM Studio
        • Chainlit.io
      • Deploying LLMs on Local Machine
        • Microsoft AutoGen Studio
        • Ollama
        • Gradio Web UI for LLMs
        • Pinokio -Install and Run Servers & AI Apps
      • LLM Use Case and Fine-Tuning and Training Costs
      • How to Build a LLM from Scratch
        • 1. Evaluate Financial Costs involved
        • 2. Data Curation
          • GPT-4 Tokens
        • 3. Model Architecture
        • 4. Training the Model at Scale
        • 5. Evaluating the Model
      • Databases and LLMs
        • Database-Mind
  • 🔲LARGE LANGUAGE MODELS (LLM's)
    • GOOGLE Gemini
      • Bard (Google)
      • Gemini 1.0
      • Gemini Advanced
    • META
      • LlaMA-3
      • Chameleon: Mixed-Modal Early-Fusion Foundation Models
    • Claude (Anthropic)
      • Claude 2 Parameters
    • Earnie AI (Baidu)
    • Microsoft
      • Bing Chat (Microsoft)
    • DeepMind
      • Personal AI (Pi)
    • OpenAI
      • ChatGPT-4 (OpenAi)
        • Is GPT-4 Open Source?
        • ChatGPT can now See, Hear, and Speak
      • ChatGPT-4 Turbo, Custom GPTs and AI Agents
    • Open Source LLMs
      • Open Source Image Models
        • Florence-2: Vision foundation model (Microsoft)
        • Würstchen: Fast Diffusion for Image Generation
        • Omni-Zero
      • Snowflake Arctic 128 Experts MoE
      • Mistral 7B
        • Mistral 7B Installation on Windows Locally
        • Mistral 7B Model Card-V0.1
        • Inference Parameters for Mistral 7B
      • Mixtral 8X-7B
        • Installing Mixtral 8X-7B
      • Llama 2 (Meta)
        • Installing and Using Llama 2 Locally
      • FALCON LLM
        • Deploy FALCON-180B Instantly
      • LLaVA 1.5
        • LLaVA 1.5: Best Open Source Alternative To GPT-4V
      • BLOOM
      • MPT-7B
      • Vicuna-13B
      • X-Ai's Grok
      • Qwen 2, 72B
      • DBRX (DataBricks)
      • Nemotron-4, 340B (NVidia)
        • Moshi by Kyutai
        • Performance Benchmarks
        • Nemotron 340B, Training Data
      • Florence2
  • Ai Tools Main Categories
    • 🔲TEXT & WRITING
      • Copywriting
        • Jasper
        • Rytr
        • Copy.ai
        • AI Writer
        • ShortlyAI
        • Article Forge
        • ClosersCopy
        • Typewise
        • Grammarly
        • DeepAI
        • Wordtune
      • Scriptwriting & Screenwriting
        • Celtx
        • AiScreenwriter
        • WriterDuet
        • Boords
        • StudioBinder
          • Guide to Scriptwriting
          • StudioBinder - Video
      • Story Teller / Story Writing
        • Quillbot
        • Sassbook
        • Authors.ai
        • Novel AI
        • Headlime
        • Simplified AI
        • DeepStory
        • Sudowrite
        • Compose AI (Facebook)
        • HyperWrite
        • AI Dungeon
        • Inkforall
        • Bramework
        • Plot Generator
        • StoryLab.ai
        • Charisma
        • InferKit
      • Storyboarding Tools
        • Storyboarder by Wonder Unit
        • Storyboard That
        • Krock.io (Storyboard AI)
          • Krock.io Tutorials
        • StoryboardHero AI
          • How StoryboardHero AI Works
        • FrameForge
      • Other General AI Text Tools
        • Paraphraser
        • Summarizing Tools
        • Language Translation
        • Email Assistants
          • Gappsy
      • AI Personal Assistants
        • TextCortex AI Assistant
    • 🔲AUDIO, SPEECH & MUSIC
      • SonicVisionLM
      • Text to Speech (TTS)
        • Murf AI
        • RunwayML Text-to-Speech
        • Speechify
        • Natural Readers.com
        • ElevenLabs - Prime Voice
        • ChatTTS
        • Google Text-to-Speech
        • Narakeet
          • The Narakeet story
        • Uberduck
        • TTS Technologies
      • Audio Editing
        • Audicity
      • Voice Changers / Cloners
        • Preventing the Harms of AI-enabled Voice Cloning
        • Metavoice Voice Changer
        • * OpenVoice: Open Source Voice Cloning
        • MARS5 TTS
        • RVC WebUI
          • Tutorial on How to Use RVC WebUI
        • Tortoise TTS (Text to Speech & Cloning)
          • Tutorial on How to Clone a Voice using Tortoise-TTS
        • Voice.ai
      • Video & Audio Transcription and Translation
        • Otter.ai
        • Cockatoo
        • AURIS Speech to Text
        • Whisper AI (OpenAI)
          • Command Line usage
          • Languages
          • Model Card of Whisper
        • Rask AI
        • Meta Speech-to-Speech Translation
        • Maestro AI
      • Music Generation
        • Musical Genres, Vibes and Themes
          • Musical genres
          • Most Popular Themes In Songs
          • The 13 moods of music?
        • AIVA Music Composer
        • Suno.Ai
        • Udio
        • Stable Audio
        • Google MusicFX
        • AudioCraft (Meta/Facebook)
          • Installing Audiocraft
          • How To Install Audiocraft Locally
        • Audio Cleanup - Noise Removal
          • Goyo -AI Noise Reduction Plugin
          • Goyo vs ClarityVX Audio Clean Up- comparion
        • Other AI Music Tools
      • Descript
      • Lip Sync & Animated Avatars
        • Realistic Avatars -MakeUGC
        • Headshots for Avatars and Resumes
        • Lip-Sync Articles & Resources
          • Wav2Lip Alternatives and (Jun 2023)
          • Top 10 Best Lip Sync Apps - June 2023
          • Five Lip-Sync Tools
          • 60 Best Lip Synching AI Tools
        • Hallo Portrait Animation
        • Hedra’s Character-1
        • Wav2Lip
        • AniPortrait
          • AniPortrait Tuturials
        • LivePortrait
          • LivePortrait Tutorials
        • ReadyPlayerMe (3D Models only)
        • Sync Labs (Wav2Lip compatible)
        • Lalamu Studio
          • Notes/Observations
        • TokkingHeads LipSync
        • DupDub Ai
        • D-ID
          • Notes/Comments
        • DeepFacelive
        • EMO: Emote Portrait Alive (AliBaba)
          • Lipsync AI by Emotech (3D Models Only)
        • VASA-1, AI Face Animator (Microsoft)
    • 🔲VIDEO & ANIMATION
      • Video Synthesis (Generation)
        • Top AI Video Synthesis Tools (2024)
        • Google Video
          • MAGVIT (Google)
          • Video Poet (Google)
          • Lumiere
        • Alibaba Research Lab
          • EMO: Emote Portrait Alive
          • MaTe3D
          • VividTalk
          • Outfit Anyone
          • Cloth2Tex
        • Meta
          • Make-A-Video
          • Emu Video tools
        • Apple
          • Matr-yoshka Diffusion Models (MDM)
        • Kaiber.ai
          • AI Technology used by Kaiber
          • Kaiber Tutorials
        • ⭐Runway ML
          • Technology used by Runway
          • Pricing Plans
          • Runway AI vs Pika Labs
        • Modelscope
        • ZeroScope
        • Deforum
          • Technology used by Deforum
          • How-To Videos
        • Neural Frames AI
        • ⭐Pika Labs
          • AI Technology used by Pika Labs
          • Tutorials
            • Pika Labs Motion Aware Text2Video
            • Mastering Pika 1.0
            • Pika 1.0 First Look
            • Extending Pika.AI Video Duration
        • Genmo.Ai
        • Moonvalley
        • ⭐Morph Studio - Stability AI
        • LensGo AI
          • LensGo Tutorials
        • Stable Diffusion Video
          • Stable Diffusion Video Tutorial
        • i2vGen-XL (Image to Video)
        • PromeAI
        • Assistive Video
          • Assistive Video Review
          • Assistive Video API
        • ⭐PixVerse
          • PixVerse - Next-Gen AI Video Synthesis
        • DomoAI Video Transformation
        • Artflow.ai
          • Does Artflow.Ai use LoRA to train their Models?
          • Tutorial
        • Boximator
        • AI Animation
          • Sketch.metademo
          • MagicAnimate
        • ⭐LTX Studio
          • LTX Studio Tutorial
        • ⭐Haiper AI
        • ⭐Luma Dream Machine
        • ⭐Kling.Ai
        • ⭐SORA (OpenAI)
          • SORA Technical Explanation
        • ⭐Vivago.ai
      • Video Production & Editing
        • Visla AI
        • Colossyan Creator
        • HeyGen
        • Lumen5
        • Magisto / Vimeo
        • Pictory
        • Fliki.ai
          • Fliki Review: Is it Worth it?
          • Fliki Features
        • VEED
        • AI Studios
        • Kreado.Ai
          • Ai Talking Photos
        • Synthesia
          • Synthesia AI Neural Networks for training data for Avatars
          • Review of Synthesia
          • Nvidia-backed Synthesia
        • InVideo
        • Wondershare Filmora
        • Wondershare VIBRO.AI
          • Create Realistic Ai Avatar
          • Company Information
          • Pricing
        • Clipchamp (Microsoft)
          • Auto-captions
        • DaVinci Resolve
        • Canva Video Editing
          • Canva Graphic Design
          • Using Canva
        • CapCut
        • Cloudinary
        • Neiro.Ai
        • Opus Clip
        • DupDub.ai
          • DupDup Avatars
        • Genny by LOVO
        • Descript
      • Video Enhancement & VFX
        • Topaz Video AI
        • Pixop Video Upscaler
        • Wonder Studio VFX
          • Pricing and Addition Info
        • HitPaw Video Encoder
        • Adobe Firefly
        • Vmake AI
      • 3D Asset Creation Generative AI Tools
        • Meshy Text-to-3D
        • PIFuHD
        • Sloyd.ai
        • Google DreamFusion
        • Masterpiece Studio
        • Gepetto.ai
        • Get3D by Nvidia
        • Instant NeRF (Nvidia)
        • Point-E
        • 3DFY.ai
        • Stable Diffusion for Blender
        • Reallusion Platform for Character Creators
          • iClone 8 (IC8)
          • Character Creator (CC)
          • Cartoon Animator
          • ActorCore
        • Character Creator 3
        • HeroMachine
        • Picrew
        • Storior
        • Genie 3D (Luma Labs 3D Model)
          • How to use Luma Genie 3D
        • Avatars
          • BHuman.Ai
            • BHuman.Ai Tutorials
      • 3D Virtual Worlds
        • Spline AI
        • Blockade Labs Skybox
      • Objects,Scenes & MoCap to 3D and Animation
        • Wist
        • Luma AI
        • DeepMotion Animate 3D
        • Wonder Animation
    • 🔲IMAGES, ART & DESIGN
      • Image Generation
        • Ideogram.Ai
          • Ideogram Tutorial
        • Microsoft Bing Image Creator
        • Imagen (Google)
        • ImgCreator.Ai
        • Playground.ai
        • Stable Diffusion
          • Stable Diffusion in Layman's Terms
          • Stable Diffusion and VAE
          • How-To Videos
        • RunDiffusion
        • Clipdrop by Stability.Ai
          • Full Clipdrop Toolkit
        • Invoke.ai
        • DreamStudio (StabilityAI)
        • Leonardo.Ai
        • DALL-E (OpenAI)
          • Biggest ChatGPT Update: DALL·E 3, Voice-Chat, Image Input!
        • Magnific Relight (OpenAI)
        • MidJourney
        • PicSi.ai / InsightFaceSwap
          • InsightFaceSwap
        • PicSo.ai
        • ArtSmart.ai
        • Lexica.art
        • CM3leon (Meta)
          • Meta CM3leon AI vs OpenAI Dall-E 2
        • Synthesys X
        • Starry AI
        • Blue Willow
        • DragGAN
        • NightCafe AI
        • Imagine AI Art Generator
        • EMU Image Generator (Meta)
          • EMU Image Generator News
        • Krea.AI (Pencil Sketch to Image)
          • Krea Prompts
        • SeaArt AI
        • Scenario
          • Create Consistent Characters
        • Fooocus
          • Fooocus - SDXL
          • Fooocus Training Videos
            • FaceSwap and Consistent Characters
            • Inpainting and Outpainting
            • LoRA's, Styles and Models
            • Features Tutorial
            • Fooocus Photorealistic Images
          • Fooocus Installation
            • Minimal Hardware Requirement
            • Installing Fooocus
              • Video on Fooocus
            • Default Models
            • Download
            • List of "Hidden" Tricks
            • Customization
            • All CMD Flags
            • Advanced Features
          • Running Fooocus
          • Fooocus FAQs
            • What is PyraCanny?
            • Fooocus - Face Swap
        • OmniZero for Zero-shot stylized Portrait
        • Artist
      • Image Editing Tools
        • Remove.bg
        • Adobe Photoshop with Generative AI
          • Generative AI Technologies and Models used by Adobe
          • AI Tools from Adobe for AI-Video
        • LeiaPix (Image to 3D Video)
        • Synthetik Studio Artist
        • Magnific Image Upscaler
          • Magnific.AI -vs Midjourney Upscaler Comparison
        • Photopea
        • Nero Studio AI
      • Design & Design Assistant
      • Adobe Firefly
    • 🔲PROGRAMMING & CODE
      • CODE Generators
        • GPT Pilot
        • Cursor IDE
          • Command K
          • Chat
          • Copilot++
          • @ Symbols
          • Codebase Answers
          • Docs
          • Auto-Debug
          • Fix Lints
        • AlphaDev (DeepMind)
        • AIDER
        • Locofy
        • Phind
          • Phind Technical Specs
        • GitHub Copilot
        • Tabnine
        • DeepCode
        • CodeT5
        • WPCode
        • Visual Studio IntelliCode
        • AlphaCode (DeepMind)
        • AskCodi
        • AIXcoder
        • PyCharm
        • ChatGPT (GPT-3/4)
        • Codiga
        • Smol Developer
        • Jedi
        • Cody (Sourcegraph)
        • Wing Python IDE Pro
        • Ponicode
        • Polycoder
        • DevOpsGPT
        • CodeWhisperer (Amazon)
      • Lanchain
      • VS Code - Source Code Editor
      • Low-Code/No-Code
      • Spreadsheets
      • Database Design and SQL Programming
      • Testing & QA
        • Diffblue Unit Tests for Java Apps
        • EvoSuite
    • 🟢Prompt Design and Engineering
      • IMAGE Prompts Guide
        • Image Types and Art Styles
          • Digital Art Styles
            • Digital Illustration
            • Sumic.mic
            • ArtStation
            • Minimalistic Line Art Drawing
            • Neotraditional Tattoo Design
            • Manga
            • Steampunk
          • Renowned Artists
            • Greg Rutkowski
            • Paul Zizka
            • Raffaello Ossola
            • Martin Wittfooth
            • Luigi Spano
            • Vladimir Kush
            • Martin Schoeller
            • Yoshiyuki Sadamoto
            • Yoko Taro
            • Tom Bagshaw
            • Yoshitaka Amano
            • Zdzislaw Beksinski
            • Boris Vallejo
            • Alphonse Mucha
            • Gustav Klimt
            • Hiro Oda
            • Michael Cheval
            • Tanvir Tamim
          • Art Terms you should be using in Generative Art
            • 20 Art Terms you Should be Using in Midjourney AI
            • Prompts Secrets: 20 Midjourney Art Styles to Make your AI images POP
            • 20 MORE Art Styles -More Magic Midjourney Prompts
            • 20 illustration art styles to try in Midjourney
        • Lighting and Ambience
        • Camera Instructions
        • Output Aspect Ratio
          • Complete List of Aspect Ratio's
        • Seeds, Weights, and Prompt Parameters
        • Negative Prompts
        • Help with Prompt Terms
          • Global illumination (GI) Scattering Glowing Shadows
          • Trending on Artstation, by artgerm
          • H.R Giger and Beksinski
          • Misato Katsuragi
          • Fractal Isometrics details Bioluminescence
          • Neo-noir
          • Ray tracing, ambient occlusion, txaa ssao, etc.
          • Prywinko Art
          • Studio Ghibli
          • Phrase - Visual Poetry by Yukisakura
          • Volumetric (3D Representation or Depth and Dimension)
          • Ray Tracing and Ray-Traced Lighting
          • Octane Render and Unreal Engine 5
          • Glibatree
      • IMAGE Models
        • IDEOGRAM Prompts
          • Logos
          • Banners for Social Media Pages
          • Oil Paintings
          • Fantasy
          • Posters
          • Abstract Patterns
          • Text Captions
            • Neon Lettering
            • Embroidery
          • Signs
          • Androids and Cyborgs
          • Illustrations
        • LEONARDO Prompt Guide
          • Ultimate Prompt Generating Machine
          • Male Portraits/Images
          • Female Portraits/Images
            • European Women
            • American Women
            • Elderly Women
            • Woman in stunning Thai dress
            • Beautifully designed dress adorned with golden embroidery
          • General Scenes
          • Illustrations and Creative
          • Patterns, Logos and Designs
          • Greco-Roman
          • Vikings
          • Historic Scenes
          • Islamic and Arabs
        • DALL-E3 Prompt Guide
          • Industrial Age Prompts
          • Vintage Images
          • Sci-Fi and Futuristic
            • Androids
            • Humanity in Year 2039
            • Gen-AI Revolution
              • Video Synthesis and Movie Making
              • Voice Synthesis and Cloning
            • Classrooms
            • On a Farm
            • In an Office
            • Sports
            • Festive Humans and Humanoids
          • Historic Characters and Scenes
            • Scientists
            • Mongols and Chinese
            • Greco Romans
            • Islamic History
            • Guru Nanak
          • Objects
      • TEXT Generation Prompts
        • Claude 3 Prompt Library
      • VIDEO Synthesis Prompts
        • Text to Video Prompt Techniques
        • Pixverse Examples
          • Cinematic wide angle view
          • Close-up shot
          • Image to Video Examples
        • Pika Labs Prompts
          • Pika Labs Negative Prompts
        • Runway 3
          • Camera Movement
        • Hotshot
      • IMAGE TO TEXT Prompts
        • CLIP Interrogator
        • img2prompt
        • Replicate (Image-to-Text)
      • Hallucinations in LLMs
      • Prompt Generators
        • Prompt GPTs
          • Humans and Humanoids
        • G-Prompter
        • PromptMoat
          • Blade Runner Blues
        • PromptPerfect
          • PromptPerfect Resources
        • Weloveai.ca
        • Creative fabrica
        • Dair AI Prompt Engineering Guide
        • Vercel.ai Prompt Engineering
        • Brex's Prompt Engineering Guide
        • PromptoMANIA
    • 🔲AI RESOURCES
      • Custom GPTs (Open Ai)
        • Find the perfect AI Tool or GPT for your use case.
        • OpenAi GPT Store
        • GPTs Directory
          • GPTs for Research
          • GPTs for Video
          • GPTs for Image Generation
      • Huggingface
        • Models
      • Stock Photos and Videos
        • WireStock
          • Wirestock.io news, Oct 2023
        • Envato Elements
      • Promising Projects
        • GPT4RoI
        • Simulon VFX Tool
        • IP Adapter: Text Compatible Image Prompt Adapter for Text-to-Image Diffusion Models
        • ConceptLab Creative Generation Framework
        • Dual-Stream Diffusion Net (DSDN) for Text-to-Video Generation
        • CoDeF
        • 3D Gaussian Splatting for Real-Time Radiance Field Rendering
        • MotionGPT
        • PanoHead
        • Rerender A Video
        • Picsart-AI-Research
        • Point-E (Open AI)
        • MagicAvatar
        • MagicEdit
        • Würstchen: Fast Diffusion for Image Generation
        • Zero-1-to-3 (Image to 3D)
      • Replicate.com
      • Katonic Playground
      • Vercel AI SDK
      • Cool Apps
      • Upscaling / Super-Resolution
      • Decentralized LLM's
        • PETALS
        • TAO Bittensor
        • Zarqa - Neural-Symbolic LLM
      • Singularity Twin Protocol
        • Introduction
        • Why Twin Protocol is Critically Needed
        • Corporate Case for Twin Protocol
        • Market Overview
        • Main Target Market
        • Twin Product Details
        • Twin Technology
        • Proof of Concept
        • Sources of Revenue
        • Twin Wallet
        • Twin User Data Security
        • Twin Marketplace
        • Roadmap
        • Leadership Team
        • Twin Ecosystem Economy
        • Reference
      • AI Autonomous Agents
        • AutoGen
        • MicroGPT
        • JARVIS/HuggingGPT
        • AutoGPT
        • AgentGPT
        • Crew.ai
          • Crew.Ai Tutorials
        • MetaGPT
          • How To Install MetaGPT
        • SuperAGI
          • How To Install SuperAGI
        • Agent-LLM
        • PHOTONAI (Python API)
      • Learn Generative Ai Coding
      • AI News Sources and Channels
        • weloveai.ca
        • The Birth of AI Videos
        • AI Generative Art Tools
        • 7 Underrated AI tools no one knows.. yet
    • 🔲AI HARDWARE (GPU's & TPU's) and Cloud Services
      • Hardware Specs for Stable Diffusion SDXL
      • Nvidia GPU Hardware
        • NVIDIA H100 GPU
        • Nvidia GH200
      • Echo3D Cloud Platform
      • AMD GPU Hardware
      • Monster API
      • AWS Cloud9 IDE
      • AWS Bedrock
      • Google Vertex AI
    • 🔲OTHER
      • BUSINESS & ADMIN
        • Customer Service & Support
        • Finance & Accounting
        • Human Resources
        • Legal Assistant
        • Presentations
        • Productivity
          • Notion AI
        • Tools for Startup Businesses
        • Marketing & Sales
          • TubeBuddy
          • Search Engine Optimization (SEO)
          • Social Media
          • E-commerce
          • Sales
          • Advertising and Marketing
          • Email Marketing
          • Website Generation
            • LEIA AI Website Builder
            • Braidio
            • Relume AI Site Builder
            • Pineapple AI Website Builder
            • KLEAP AI Website Builder
            • Hocoos AI Website Builder
      • LIFESTYLE & ENTERTAINMENT
        • Fashion & Grooming
        • Real Estate
        • Fun Tools
          • DAISYS.Ai
        • Gaming
        • Gift Ideas
        • Healthcare
        • Life Assistant
        • Memory, Retention and Long-Term Recollection
        • Health and Fitness
        • Travel & Vacationing
        • Dating & Companionship
        • Religion
        • AI Meme
        • Heritage, Genealogy and Family History
          • MyHeritage
        • Productivity
          • Krock.ai
      • EDUCATION / INFORMATION
        • Education Assistant
        • Tutors and Mentors
        • AI Search (Curated Research & Comparative Analysis)
          • Perplexity AI
  • 🔲SOLUTIONS & TUTORIALS
    • ALL AI TOOLS IN ONE PLACE
      • 𝐀𝐈 𝐓𝐨𝐨𝐥𝐬 𝐅𝐨𝐫 𝐄𝐯𝐞𝐫𝐲 𝐏𝐮𝐫𝐩𝐨𝐬𝐞
    • Tools Selection
      • Reface AI
      • Akool AI Swap Face Tutorial
      • Adobe Express
      • DeepFaceLab
        • Pretrained Models
      • Face Swap
      • Clone Works AI studio
      • Wav2Lip (LipSync)
      • Lalamu Studio (LipSync)
      • ROOP (Deep Fake Video)
        • ROOP Tutorial
        • ROOP Installation
      • Dataset Tag Managers
        • Booru Dataset Tag Manager
          • Why is the Booru Dataset Tag Manager called 'Booru'?
        • Image Dataset Tag Editor
    • Installing Python
      • Code in Python using AI
    • Create a News Video Channel
    • Create Animations & Videos
      • AI Animation Generator - 3D Movie with AI
      • Create Animation Videos for Kids
      • Create Animated Video from a Single Image
      • Convert Video Into Animation
    • Realistic AI Influencers
      • Hyper Realistic Aitana Lopez example
    • Talking Avatars
      • Talking Avatar with LALAMU and LEONADO
      • Talking Avatar with D-ID, Leonardo.Ai, Remove.bg and LeiaPix
      • Tokkingheads Lipsync
      • Talking Avatar using Wav2Lip
      • Talking Avatars with Stable Diffusion
      • Avatars - Lexica.art, Remove.bg and D-ID
      • Articles on Talking Avatars
        • The 10 Best Talking Avatar Software for 2023
        • 11 Best Free Talking Avatar Creators in 2023
      • Clone Yourself with Synthesia and Eleven Labs
    • Stable Diffusion Tutorials
      • Stable Diffusion Checkpoint Models
    • LoRA Guide
      • Creating Realistic AI-Models of Humans for Animation
        • Tools Used
        • Step 1: Prepare Images
        • Step 2: Training your LoRA Model Dataset
        • Step 3: Creating our Puppet
      • Fine-Tuning A Model with Kohya
        • LoRA Training Videos -Robert Jene 1
        • LoRA Training Videos -Robert Jene 2
    • Astragallery
  • 🔲AI TECHNOLOGY
    • 𝐀𝐈 𝐀𝐥𝐠𝐨𝐫𝐢𝐭𝐡𝐦𝐬 𝐒𝐢𝐦𝐩𝐥𝐢𝐟𝐢𝐞𝐝
    • Foundation Models
      • NLP, CV, Gen-AI Foundation Models
    • Artificial Neural Networks (ANN's)
      • Machine Learning (ML)
      • Deep Learning (DL)
      • Human Brain Vs. Computer Brain
    • ANN Deep Learning Architectures
      • 1. Feed-Forward Neural Networks (FFNs)
        • a) Fully Connected Networks or Multilayer Perceptrons (MLPs)
        • b) Convolutional Neural Network (CNNs)
          • (i) Residual Network (ResNet)
          • (ii) AlexNet (Convolutional Neural Network)
      • 2. Recurrent Neural Networks (RNNs)
        • a) Basic Recurrent Neural Networks
        • b) Long Short-Term Memory Networks (LSTMs)
        • c) Gated Recurrent Units (GRUs)
        • d) Bidirectional RNNs
        • e) Sequence to Sequence Models
      • 3. Transformers
        • a) Transformer (Base Model)
        • b) BERT (Bidirectional Encoder Representations from Transformers)
        • c) GPT (Generative Pretrained Transformer)
        • d) T5 (Text-to-Text Transfer Transformer)
        • e) BART (Bidirectional and Auto-Regressive Transformers)
        • f) RoBERTa, ALBERT, DistilBERT
    • Generative Ai Architectures & Models
      • Generative AI Architectures
        • a) Variational Autoencoders (VAEs)
        • b) Diffusion Models
        • c) Generative Adversarial Networks (GANs)
        • e) Transformers
        • f) Flow-based models
      • Generative AI & LLM's for Text
        • Falcon 180B
        • GPT-3 (Generative Pre-Trained Transformer-3)
        • GPT-4 (Generative Pre-Trained Transformer-4)
        • LaMDA (Language Model for Dialogue Applications)
        • LLaMA (Large Language Model Meta AI)
          • LLaMA Model Card
        • PaLM 2 ( Pathways Language Model 2)
        • Bard (Google)
        • BLOOM (Big Language Model Open for Open Research)
        • Claude 2 (Anthropic)
        • Azure OpenAI (Microsoft)
        • T5 (Text-to-Text Transfer Transformer)
        • Large Language Models (LLMs)
      • Generative AI Models For Video and Image Synthesis
        • Generative Video
        • Generative Images
          • DALL-E & DALL-E2 (OpenAi)
          • Stable Diffusion (StabilityAI)
          • StyleGAN2 (NVIDIA)
          • CLIP (OpenAI)
            • Details on CLIP
          • MIDJOURNEY
          • BigGAN (Google)
          • IMAGEN (Google)
          • GigaGAN
      • ControlNet
      • LoRA (Low-Rank Adaptation of Large Language Models)
        • Training a LoRA Model
        • Mastering LoRA
      • Pre-Training (Initializing the Network)
    • Zero-Shot Video Generation
    • Symbolic Reasoning AI
    • Neuro-Symbolic AI
    • Vision Transformer (ViT)
  • 🔲GLOSSARY OF AI TERMS
    • Keywords and Hashtags
Powered by GitBook
On this page
  1. 🔲About Ai Tools Research
  2. Frequently Asked Questions (FAQs)
  3. FAQs on LLM Training and Data Labelling

How to Fine Tune LLMs?

How to fine tune large language models (LLMs)

PreviousTokens, Words and PagesNextCase Study of Fine-Tuning an LLM

Last updated 1 year ago

CtrlK
  • How to fine tune large language models (LLMs) with Kili Technology
  • What are large language models (LLMs)?
  • Tutorial: Fine-Tuning Large Language Models with Kili Technology
  • Large Language Models fine-tuning: final thoughts
  • FAQ on Fine-Tuning LLMs

How to fine tune large language models (LLMs) with Kili Technology

Learn how to fine-tune large language models (LLMs) for specific tasks using Kili Technology. This tutorial provides a step-by-step guide and example on fine-tuning OpenAI models to categorize news articles into predefined categories.

Table of Contents

  • What are large language models (LLMs)?

    • Fine-tuning a Large Language Model

    • What is Fine-tuning?

    • What are the Most Common Fine-tuning Methods?

  • Tutorial: Fine-Tuning Large Language Models with Kili Technology

    • The data

    • The objective

  • Large Language Models fine-tuning: final thoughts

  • FAQ on Fine-Tuning LLMs

    • What are large language models?

    • What are fine-tuned models?

What are large language models (LLMs)?

Let’s start our tutorial by explaining LLMs. Large Language Models (LLMs) are a class of machine learning models that are capable of processing and generating natural language text. These models are trained on massive amounts of text data, often using unsupervised learning techniques, to learn patterns and representations of language.

Large language models are distinguished by their size and complexity, with billions of parameters, making them some of the most powerful AI systems developed to date.

Some examples of large language models include OpenAI's GPT-3, Google's T5, and Facebook's RoBERTa. These models have been shown to excel at a wide range of natural language processing tasks, including text classification, language translation, and question-answering. But as you may have experienced, these large language models may struggle with more industry-specific cases and will require additional training to be effective for more particular applications.

LLMs are typically trained using massive amounts of text data, such as web pages, books, and other sources of human-generated text. This allows the models to learn patterns and structures in language that can be applied to a wide range of tasks without needing to be retrained from scratch.

The ability to transfer knowledge from pre-training to downstream tasks is one of the critical advantages of LLMs, and it has led to significant advances in natural language processing in recent years.

Fine-tuning a Large Language Model

What is Fine-tuning?

general-to-specific-objective

Fine-tuning is a technique in machine learning used to adapt a pre-trained model to perform better on a specific task. The idea behind fine-tuning is to leverage the knowledge and representations learned by the pre-trained model and then further optimize the model's parameters for the new task.

The process of fine-tuning typically involves the following steps:

  1. Starting with a pre-trained model, often on a large dataset of data.

  2. Adding a task-specific layer on top of the pre-trained model. This layer is usually a set of trainable parameters that can be optimized for the new task.

  3. Fine-tuning the model on a smaller, task-specific dataset that is labeled with the desired output. During this stage, the parameters of the task-specific layer and the pre-trained model are updated to minimize the difference error between the predicted output and the expected output.

  4. Evaluating the performance of the fine-tuned model on a hold-out test set to ensure it is generalizing well on new examples.

Fine-tuning is a powerful technique in many areas of machine learning, including natural language processing and computer vision. By starting with a pre-trained model and only updating a small set of parameters for a specific task, fine-tuning allows for efficient use of computational resources and can often achieve state-of-the-art results.

What are the Most Common Fine-tuning Methods?

There are several common methods for fine-tuning pre-trained models in machine learning. Here are a few examples:

  1. Linear Fine-Tuning: This is the simplest and most common form of fine-tuning, where a linear layer is added on top of the pre-trained model and then trained for the specific task. In this approach, the pre-trained weights are frozen, and only the weights of the new linear layer are learned from scratch. This technique is often used for text classification and sentiment analysis tasks.

  2. Full Fine-Tuning: In this approach, all the weights of the pre-trained model are fine-tuned on the specific task, including the pre-trained layers. This approach can be more computationally expensive, but it can be effective for tasks that require more nuanced understanding of the input, such as language translation or image captioning.

  3. Gradual Unfreezing: This is a technique where the pre-trained layers are gradually unfrozen and trained on the specific task, starting from the topmost layer and gradually working down to the lower layers. This allows the model to learn task-specific features while retaining the general language or image understanding from pre-training.

  4. Adapter-Based Fine-Tuning: Adapter-based fine-tuning is a recently proposed technique that allows fine-tuning with minimal changes to the pre-trained model. This approach adds a small adapter network to the pre-trained model, which is then trained on the downstream task while the pre-trained model is frozen. This technique can reduce the computational cost and memory requirements of fine-tuning while still achieving competitive results.

These are just a few examples of the many fine-tuning techniques that exist. The choice of the best method depends on the specific task, the available computational resources, and the trade-offs between performance and efficiency.

How do we fine-tune large language models?

Fine-tuning adapts pre-trained LLMs to specific downstream tasks, such as sentiment analysis, text classification, or question-answering. The goal of fine-tuning is to leverage the knowledge and representations of natural language, code (and the list goes on) learned by the LLM during pre-training and apply them to a specific task.

The process of fine-tuning involves taking a pre-trained LLM and training it further on a smaller, task-specific dataset. During fine-tuning, the LLM's parameters are updated based on the specific task and the examples in the task-specific dataset. The model can be customized to perform well on that task by fine-tuning the LLM on the downstream task while still leveraging the representations and knowledge learned during pre-training.

The basic steps of fine-tuning a pre-trained LLM are as follows:

  1. Initialize the large language model with the pre-trained weights.

  2. Add a task-specific head to the large language model.

  3. Train the large language model on the task-specific dataset, updating the weights of both the head and the large language model.

  4. Evaluate the fine-tuned model on a validation set.

  5. Repeat steps 3-4 until the model achieves satisfactory performance.

Fine-tuning has proven to be an effective way to adapt pre-trained LLMs to a wide range of downstream tasks, often achieving state-of-the-art performance with relatively little additional training.

Learn more about how to build domain specific LLMs

Tutorial: Fine-Tuning Large Language Models with Kili Technology

In this tutorial, we will walk you through the process of fine-tuning OpenAI models using Kili Technology. We redirect you to this notebook written by technical writer. It provides a step-by-step guide to building a machine-learning model capable of categorizing short, Twitter-length news into one of eight predefined categories.

You can follow the notebook right after reading this article but for a general idea of what we've done, here's a quick description of the process and what to look out for.

The data

The data we've used is from Kaggle. It's a dataset listing HuffPost's articles published over the course of several years, with links to articles, short descriptions, authors, and dates they were published. The data here is under a 4.0 Creative Commons license which allows us to share and adapt the data however we want, so long as we give appropriate credit.

The objective

Our objective is to fine tune a large language model to assign one of 8 categories to the news items in our dataset. To speed up the process, we removed all the entries with vague categories like IMPACT or PARENTING. This is the final list:

  • MEDIA & ENTERTAINMENT

  • WORLD NEWS

  • CULTURE & ARTS

  • SCIENCE & TECHNOLOGY

  • SPORTS

  • POLITICS

  • MONEY & BUSINESS

  • STYLE & BEAUTY

To make matters as simple as possible, the assumption is that one piece of news can match only one of these categories.

OpenAI recommends having a couple of hundred training samples to fine-tune their models effectively. So we collected. To ensure we meet the requirements, we've prepared 4 sample files. Each file contains 100 labeled examples for each of the eight classes. If you need more, you can easily process the original dataset for additional samples.

The model

OpenAI has a number of models and you can find more information about their models here. When you're choosing your own model, take into consideration the costs, maximum tokens, and performance. In our use case we fell back to using Curie, which is an appropriate model that is fast, capable, and costs less than other models.

The steps

As mentioned earlier, you can follow through all the steps through our notebook. The steps outlined here is just to give you an idea of the process.

  1. Setting up - Fine-tuning your model is done via Kili's API, so if you're following along with the notebook, you'll need to have that set-up. If you need guidance, we have handy documentation here.

  2. Create a JSON file for your project's ontology - In our case, we want a classification job that splits things into various categories: media and entertainment, world news, etc. We also want to specify that it should only select one category from the list. We also added an UNABLE_TO_CLASSIFY category for situations when, for whatever reason, the model fails to do the job. Here's what it looks like for us:

interface = {
    "jobs": {
        "CLASSIFICATION_JOB": {
            "content": {
                "categories": {
                    "MEDIA_AND_ENTERTAINMENT": {
                        "children": [],
                        "name": "MEDIA AND ENTERTAINMENT",
                        "id": "category11",
                    },
                    "WORLD_NEWS": {"children": [], "name": "WORLD NEWS", "id": "category12"},
                    "CULTURE_AND_ARTS": {
                        "children": [],
                        "name": "CULTURE AND ARTS",
                        "id": "category13",
                    },
                    "SCIENCE_AND_TECHNOLOGY": {
                        "children": [],
                        "name": "SCIENCE AND TECHNOLOGY",
                        "id": "category14",
                    },
                    "SPORTS": {"children": [], "name": "SPORTS", "id": "category15"},
                    "POLITICS": {"children": [], "name": "POLITICS", "id": "category16"},
                    "MONEY_AND_BUSINESS": {
                        "children": [],
                        "name": "MONEY AND BUSINESS",
                        "id": "category17",
                    },
                    "STYLE_AND_BEAUTY": {
                        "children": [],
                        "name": "STYLE AND BEAUTY",
                        "id": "category18",
                    },
                    "UNABLE_TO_CLASSIFY": {
                        "children": [],
                        "name": "UNABLE TO CLASSIFY",
                        "id": "category19",
                    },
                },
                "input": "radio",
            },
            "instruction": "Select a matching category:",
            "mlTask": "CLASSIFICATION",
            "required": 1,
            "isChild": False,
            "isNew": False,
        }
    }
}json

3. Extract the data from the curated dataset and upload the news headlines to Kili.

4. Generate predictions using your model - In this tutorial we'll be using one of OpenAI's models. You'll need your OpenAI organization ID and OpenAI API key.

If you're used to GPT and other OpenAI tools at this point, you'll have some experience in writing a good prompt that directly states what you want your LLM to do.

In our case, we used this prompt:

Classify the text of the following message as exactly one of the following: MEDIA AND ENTERTAINMENT, WORLD NEWS, CULTURE AND ARTS, SCIENCE AND TECHNOLOGY, SPORTS, POLITICS, MONEY AND BUSINESS, STYLE AND BEAUTY.

The LLM might want to return more than one category, so make sure to filter that and remember to set the fallback of UNABLE_TO_CLASSIFY. Start with a small set for trial, and once it works, run it on the whole data set.

Here's how we did it with our code:

prompt_text = """Classify the text of the following message as exactly one of the following:
MEDIA AND ENTERTAINMENT, WORLD NEWS, CULTURE AND ARTS, SCIENCE AND TECHNOLOGY,
SPORTS, POLITICS, MONEY AND BUSINESS, STYLE AND BEAUTY.
message: .
text: """


def classify_text(content, model):
    classification = openai.Completion.create(
        model=model,
        engine="text_curie_001",
        prompt=prompt_text.replace("", content),
        max_tokens=10,
        temperature=0,
    )

    # Based on OpenAI's tokenizer, https://platform.openai.com/tokenizer,
    # our longest class is 6-tokens long.
    # Just in case, we've set the max_tokens value to 10

    returned_value = classification["choices"][0]["text"]  # type: ignore

    all_predefined_classes = [
        "POLITICS",
        "MEDIA AND ENTERTAINMENT",
        "WORLD NEWS",
        "CULTURE AND ARTS",
        "SCIENCE AND TECHNOLOGY",
        "SPORTS",
        "MONEY AND BUSINESS",
        "STYLE AND BEAUTY",
    ]

    result = "UNABLE_TO_CLASSIFY"

    # Sometimes the model returns more than one class so let's filter them out:

    index = len(returned_value)
    for predefined_class in all_predefined_classes:
        if (
            predefined_class in returned_value.upper()
            and returned_value.upper().index(predefined_class) < index
        ):
            result = predefined_class

    return resultpython
sample_message_to_classify = (
    "Golden Globes Returning To NBC In January After Year Off-Air. For the past 18 months,"
    " Hollywood has effectively boycotted the Globes after reports that the HFPA's 87 members of"
    " non-American journalists included no Black members."
)
# sample_message_to_classify_2 = "Amazon Greenlights 'Blade Runner 2099' Limited Series Produced By Ridley Scott. The director of the original 1982 film joins a writer of the 2017 sequel for the newest installment in the sci-fi franchise."
# sample_message_to_classify_3 = "Tips For Hand-Washing Clothes In The Tub. Doing laundry at home during the coronavirus pandemic? Experts share their tried-and-true ways to clean clothing by hand."

classify_text(content=sample_message_to_classify, model="curie")
# classify_text(content=sample_message_to_classify_2, model="curie")
# classify_text(content=sample_message_to_classify_3, model="curie")python

5.Do some manual human labeling In our notebook, we simulated human labeling, but in real cases manual, the objective of human labeling is to compare the model's performance and ensure the quality of the data labeling. So when putting this into practice, it's important to keep to high standards. In a real-world project where a lot is at stake, you want to avoid the situation when different labelers assign different classes to ambiguous content.

6.Use Kili's KPIs to compare the model vs. ground truth (human annotations) In our project, there are only two possible IoU scores we can have per asset: either 50% for differing results or 100% for aligned results. We just divide one by the number of selected categories for an entry. If both the LLM and the human labeler agree on the tag, your score will be 100% (one entry / one tag for this entry). If both the LLM and the human tagger have a different tag, your score will be 50% (one entry / two tags for this entry)

7.Fine-tuning the base model In our situation, the model gave us a score of 16%. Of course, we need to fine tune this. OpenAI already provides a fine tuning API. For this, you must provide a list of individual prompts and their associated output.

8.Validate the fine-tuned model - As what we've done in step 6. We will compare the model's performance by generating new predictions and benchmarking it against human labeling. After calculating the IoU, we arrive at 19%. This shows us that we're heading in the right direction, but we still need plenty of work.

9.Repeat steps 4 - 8 - As we've discussed, fine-tuning may need additional data and iteration until you are satisfied with the results. And that is how you fine-tune an large language model via Kili Technology!

Large Language Models fine-tuning: final thoughts

Large Language Models (LLMs) have become a cornerstone of modern natural language processing, enabling unprecedented performance levels across a range of language tasks. Fine-tuning pre-trained LLMs has emerged as a powerful technique for adapting these models to perform specific tasks with high accuracy, even when labeled fine-tuning datasets are small. It is clear that fine-tuning LLMs has opened up new possibilities for natural language processing and has the potential to revolutionize the way we interact with language in the years to come.

FAQ on Fine-Tuning LLMs

What are large language models?

Large Language Models (LLMs) are machine learning models that use deep neural networks to learn complex patterns in natural language. They are pre-trained on large amounts of text data and can then be fine-tuned for specific natural language processing tasks, such as language translation, text classification, or text generation. LLMs have significantly advanced natural language processing and have been widely adopted in various applications.

What are fine-tuned models?

Fine-tuned models are machine learning models that have been adapted to perform a specific task using a pre-trained model as a starting point. Fine-tuning involves adding a task-specific output layer on top of the pre-trained model and updating the weights of both the output layer and the pre-trained model on task-specific data to improve performance on the target task.

What are the most famous large language models?

Some of the most famous large language models include GPT-3 (Generative Pre-trained Transformer 3), BERT (Bidirectional Encoder Representations from Transformers), Roberta (Robustly Optimized BERT approach), and T5 (Text-to-Text Transfer Transformer). These models have significantly advanced natural language processing and have been widely adopted in various language tasks, such as text generation, classification, and language translation.

Why is it often desirable to fine-tune large pre-trained language models rather than train a new model from scratch?

Fine-tuning pre-trained language models is often desirable instead of training a new model from scratch due to the computational resources required to pre-train a model from scratch. Fine-tuning allows for faster and more efficient training, utilizing pre-learned representations that can be optimized for a specific task and achieve state-of-the-art results with less data.

How do we fine-tune large language models?
The model
The steps
What are the most famous large language models?
Why is it often desirable to fine-tune large pre-trained language models rather than train a new model from scratch?