/lmg/ - A general dedicated to the discussion and development of local language models.

Unanism edition

Previous Threads: >>98760271 & >>98764931

►News
>(01/31) miqu-1-70b is confirmed to be "a quantised (and watermarked) version of an early model we trained and distributed quite openly" by Mistral CEO https://huggingface.co/miqudev/miqu-1-70b/discussions/10
>(01/30) PR for another SoTA lcpp quant, this time is Q3 merged https://github.com/ggerganov/llama.cpp/pull/5196
>(01/30) PR for DynaTemp finally merged in exl2 https://github.com/turboderp/exllamav2/pull/263
>(01/30) New sampler PR by Kalomaze for ooba is open https://github.com/oobabooga/text-generation-webui/pull/5403
>(01/29) Meta released three new CodeLlama-70B https://huggingface.co/codellama

►FAQ: https://wikia.schneedc.com
►Glossary: https://archive.today/E013q
►Links: https://rentry.org/LocalModelsLinks
►Official /lmg/ card: https://files.catbox.moe/cbclyf.png

►Getting Started
https://rentry.org/llama-mini-guide
https://rentry.org/8-step-llm-guide
https://rentry.org/llama_v2_sillytavern
https://rentry.org/lmg-spoonfeed-guide

►Further Learning
https://rentry.org/machine-learning-roadmap
https://rentry.org/llm-training
https://rentry.org/LocalModelsPapers

►Benchmarks
General Purpose:
https://hf.co/spaces/HuggingFaceH4/open_llm_leaderboard
https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard
Programming: https://hf.co/spaces/bigcode/bigcode-models-leaderboard
Censorbench: https://codeberg.org/jts2323/censorbench

►ERP/RP Datasets
https://rentry.org/qib8f
https://rentry.org/ashh2

►Alpha Calculator
https://desmos.com/calculator/ffngla98yc

►Text Gen. UI, Inference Engines
https://github.com/oobabooga/text-generation-webui
https://github.com/LostRuins/koboldcpp
https://github.com/lmg-anon/mikupad
https://github.com/turboderp/exui
https://github.com/ggerganov/llama.cpp
https://github.com/turboderp/exllamav2