/lmg/ - a general dedicated to the discussion and development of local language models.Previous threads: >>100180197 & >>100173514►News>(04/24) Snowflake Arctic Instruct 128x3B MoE released: https://hf.co/Snowflake/snowflake-arctic-instruct>(04/23) Phi-3 Mini model released: https://hf.co/microsoft/Phi-3-mini-128k-instruct-onnx>(04/21) Llama3 70B pruned to 42B parameters: https://hf.co/chargoddard/llama3-42b-v0>(04/18) Llama3 8B, 70B pretrained and instruction-tuned models released: https://llama.meta.com/llama3/>(04/17) Mixtral-8x22B-Instruct-v0.1 released: https://mistral.ai/news/mixtral-8x22b/►News Archive: https://rentry.org/lmg-news-archive►FAQ: https://wikia.schneedc.com►Glossary: https://rentry.org/lmg-glossary►Links: https://rentry.org/LocalModelsLinks►Official /lmg/ card: https://files.catbox.moe/cbclyf.png►Getting Startedhttps://rentry.org/llama-mini-guidehttps://rentry.org/8-step-llm-guidehttps://rentry.org/llama_v2_sillytavernhttps://rentry.org/lmg-spoonfeed-guidehttps://rentry.org/rocm-llamacpp►Further Learninghttps://rentry.org/machine-learning-roadmaphttps://rentry.org/llm-traininghttps://rentry.org/LocalModelsPapers►BenchmarksChatbot Arena: https://chat.lmsys.org/?leaderboardProgramming: https://hf.co/spaces/bigcode/bigcode-models-leaderboardCensorship: https://hf.co/spaces/DontPlanToEnd/UGI-LeaderboardCensorbench: https://codeberg.org/jts2323/censorbench►ToolsAlpha Calculator: https://desmos.com/calculator/ffngla98ycGGUF VRAM Calculator: https://hf.co/spaces/NyxKrage/LLM-Model-VRAM-CalculatorSampler visualizer: https://artefact2.github.io/llm-sampling/index.xhtml►Text Gen. UI, Inference Engineshttps://github.com/oobabooga/text-generation-webuihttps://github.com/LostRuins/koboldcpphttps://github.com/lmg-anon/mikupadhttps://github.com/turboderp/exuihttps://github.com/ggerganov/llama.cpp
Why on Earth are you doing a page 6 bake?
> transgender mikukill yourself
https://www.youtube.com/watch?v=X0Le56V6feg
>>100184962Is there a llama3 model for ooba already? I heard there's a new one which is particularly fast, but I usually just get them through thebloke and I couldn't find it.
>>100187293There were quants of llama3 like 10 minutes after release. TheBloke retired a while ago. Just search for the quants on huggingface or download the model and quant it yourself.
>>100187341Oh well, that's a shame. I wish the guy a happy retirement, though. He was pretty cool. Also thanks
Any good l3 8b finetunes yet?
>>100184962Kill trannies. Behead trannies. Roundhouse kick a tranny into the concrete. Slam dunk a tranny into the trashcan. Crucify filthy trannies. Defecate in a trannies food. Launch trannies into the sun. Stir fry trannies in a wok. Toss trannies into active volcanoes. Urinate into a trannies' gas tank. Judo throw trannies into a wood chipper. Twist tranny heads off. Report trannies to the IRS. Karate chop trannies in half. Curb stomp trannies. Trap trannies in quicksand. Crush trannies in the trash compactor. Liquefy trannies in a vat of acid. Dissect trannies. Exterminate trannies in the gas chamber. Stomp tranny skulls with steel toed boots. Cremate trannies in the oven. Lobotomize tranny. Drown trannies in fried chicken grease. Vaporize trannies with a ray gun. Kick trannies down the stairs. Feed trannies to alligators. Slice trannies with a suzuki katana.
>nobody talking about Moistral despite it literally being a Euryale-tier 11B with better formatting and very creative vocabulary
>>100184962good llama3 params?
>>100184962
looking to ween myself off of GPT4 codegen sota. I basically use it to instruct what i want, i know what i want, give it context, only use simple c prototypes. how realistic is it to rely on l370b instruct?
How tf to even use phi 3 mini when it's "unsupported" by llama.cpp?
>>100184962What an absolutely foul looking miku
>>100184962not sure which thread is legit so I'll ask here: >>100194467also, have a bump so you can reuse this thread later
>>100184962why is tranny "art" so vile goddamn