Odkazy z přednášky na TopMonks Caffè - AI Prakticky

topmonks-ai-sources.md

Prezentace na google drive

https://docs.google.com/presentation/d/16OHXPpd_Eca05VDJaDjNxiaIX32NsKyM/edit?usp=sharing&ouid=102632222370728552977&rtpof=true&sd=true

Zdroje a materiály

• awesome llms https://github.com/Hannibal046/Awesome-LLM
• https://github.com/Mooler0410/LLMsPracticalGuide
• počty parametrů https://informationisbeautiful.net/visualizations/the-rise-of-generative-ai-large-language-models-llms-like-chatgpt/
• pytorch tf https://www.assemblyai.com/blog/pytorch-vs-tensorflow-in-2023/

formáty

• https://pytorch.org/tutorials/beginner/saving_loading_models.html

• https://huggingface.co/docs/diffusers/using-diffusers/using_safetensors

• https://huggingface.co/docs/safetensors/index

• https://medium.com/@sandyeep70/ggml-to-gguf-a-leap-in-language-model-file-formats-cd5d3a6058f9

• https://medium.com/@phillipgimmi/what-is-gguf-and-ggml-e364834d241c

• https://deci.ai/blog/ggml-vs-gguf-comparing-formats-amp-top-5-methods-for-running-gguf-files/

• https://github.com/ggerganov/ggml

• https://ggml.ai/

• gpt4 parameters https://the-decoder.com/gpt-4-has-a-trillion-parameters/

• llama

  • https://the-decoder.com/metas-llama-language-model-shows-that-parameters-are-not-everything/
  • llama stable diffusion moment https://simonwillison.net/2023/Mar/11/llama/
  • gpu requirements https://www.reddit.com/r/LocalLLaMA/comments/11o6o3f/how_to_install_llama_8bit_and_4bit/

• přehled potřebné RAM podle velikosti modelu https://rentry.co/pygmalion-ai

• mixtral

  • https://huggingface.co/TheBloke/Mixtral-8x7B-v0.1-GGUF

kvantizace

• k-quants ggerganov/llama.cpp#1684 (comment)
• https://medium.com/@siddharth.vij10/llm-quantization-gptq-qat-awq-gguf-ggml-ptq-2e172cd1b3b5
• https://maartengrootendorst.substack.com/p/which-quantization-method-is-right

backendy, knihovny, frameworky

• https://pytorch.org/
• https://github.com/ggerganov/llama.cpp
• https://github.com/ollama/ollama
• https://github.com/Mozilla-Ocho/llamafile
• https://github.com/huggingface/text-generation-inference
• https://github.com/vllm-project/vllm
• https://lmstudio.ai/
• https://github.com/imartinez/privateGPT
• https://github.com/PygmalionAI/aphrodite-engine
• https://github.com/oobabooga/text-generation-webui

srovnání https://betterprogramming.pub/frameworks-for-serving-llms-60b7f7b23407 Speeding up inference https://betterprogramming.pub/speed-up-llm-inference-83653aa24c47 TogetherAI inference server: muc faster than TGI or vllm: https://medium.com/@datadrifters/the-worlds-fastest-llm-inference-engine-3x-faster-than-vllm-and-tgi-a2ed9e33c55f

silly tavern

• https://github.com/SillyTavern/SillyTavern
• kontext https://wikia.schneedc.com/bot-creation/trappu/creation

videa o tvorbě "AI waifu" https://www.youtube.com/watch?v=dKFnJCtcfMk&t=339s&ab_channel=SchizoDev

Relevant stuff: STT: whisper https://github.com/openai/whisper, can do STT while translating Whisper demo with microphone https://github.com/mallorbc/whisper_mic LLM: only gpt3 or character AI, it's old video Scraped gawr gura voice Developer: https://github.com/SchizoDev nothing interesting there TTS, training seems to be done by https://colab.research.google.com/drive/1jF-Otw2_ssEcus4ISaIZu3QDmtifUvyY I copied it to my drive Looks like project https://uberduck-guide.gitbook.io/uberduck-written-tutorial/ Talking head list: https://github.com/harlanhong/awesome-talking-head-generation Not mentioned what used exactl,y based on screenshot looks like https://github.com/Chrisyk/AI-Assistant or https://github.com/Rubiksman78/RenAI-Chat <- this one is STT+LLM+TTS in renpy, nice He uses https://github.com/pkhungurn/talking-head-anime-3-demo/tree/main to make moving head from single image https://pkhungurn.github.io/ has several interesting repos on this very topic

https://github.com/Koischizo/AI-Vtuber reads chat from youtube, responds using gpt-3, ElevenLabs TTS, also contains some model

Continuation: https://www.youtube.com/watch?v=g0KMPpakuJc&ab_channel=SchizoDev

Mentioned https://github.com/NVIDIA/tacotron2 for TTS - 4 years old Other https://github.com/neonbjb/tortoise-tts younger Integrated to coqui https://docs.coqui.ai/en/latest/models/tortoise.html https://elevenlabs.io/ are commercial

Livewhisper https://github.com/Nikorasu/LiveWhisper is live and more real-time, 2 years old https://github.com/facebookresearch/llama is just llama official inference https://github.com/yuyuyzl/EasyVtuber seems like talking head 3 fork Singing voice conversion https://github.com/svc-develop-team/so-vits-svc See readme, has e.g. Realtime version https://github.com/w-okada/voice-changer

https://github.com/Koischizo is other github account of same person The channel https://www.youtube.com/c/Koischizo is the result

STT

• https://github.com/openai/whisper
• https://github.com/Nikorasu/LiveWhisper

TTS

• https://github.com/coqui-ai/TTS

Live2D

• https://github.com/Live2D
• https://www.live2d.com/en/

databáze person https://www.reddit.com/r/KoboldAI/comments/117vomu/where_to_download_custom_characters/ https://www.dreamtavern.ai/ - No machine export, need to copypaste to sillytavern manually

- Guide  https://RENTRY.ORG/PYGMABOORU
- Contains pngs where all data is stored as metadata of the png

https://botprompts.net/ - Contains just jsons https://beta.character.ai/ https://chub.ai/ https://janitorai.com/s - Has character definitions enabled - Can be downloaded using https://www.reddit.com/r/SillyTavernAI/comments/168ya0v/i_made_a_page_where_you_can_easily_download_card/ just replace janitorai.com with janitorai.me in the url on page of certain character - https://janitorai.me/ works too

stable diffusion materiály https://stable-diffusion-art.com/

ostatní mé zdroje: https://gist.github.com/racinmat/efd25a488cac89f7c8fb65b5e9327452 https://gist.github.com/racinmat/b50f740fe2ce91325b3bf85497b7c2ea https://gist.github.com/racinmat/6f663fa82c44c8da0d2fcdb241054954 https://gist.github.com/racinmat/dd04c4df4289bdc546b6d690c93ee6ca