Zaoral @zaoral@fedibird.com

Megatron-LMとGKEで作るMixtral 8x7Bを語彙拡張継続事前学習
アベジャ
https://tech-blog.abeja.asia/entry/abeja-nedo-project-part1-202404

https://twitter.com/shion_honda/status/1782137494584152126

https://huggingface.co/spaces/shionhonda/reviewer2-bot

> Reviewer #2 Bot
論文系

https://twitter.com/iwiwi/status/1782210345056268586
>進化的モデルマージを拡散モデルに適用した事例「EvoSDXL-JP」を公開

https://twitter.com/SakanaAILabs/status/1782207884170080407

https://huggingface.co/SakanaAI/EvoSDXL-JP-v1

LLMモデル "Llama3" を 4bit 量子化
https://qiita.com/akasakat/items/0855b5f05467cc8cbbf4?fbclid=IwZXh0bgNhZW0CMTEAAR0iwxt3JVJJN5xaEMQvQKIF4VkdwkoSgfauKciGmFoHf26vUfmsC1JCO5o_aem_AZ_mt31NjH3EHjXCjeoTnDCmYo--i6OUDk2br2wvB21On13inHxX-XpSu93ciYk1Rb4lSzAu3veBcGxiRNvDWqvU

https://twitter.com/noguchis/status/1781890513853116818
WSL2 llama3

https://snn.hirlab.net/
SNN

https://huggingface.co/blog/mlabonne/orpo-llama-3

https://huggingface.co/LoneStriker/Meta-Llama-3-70B-Instruct-GGUF

https://gihyo.jp/article/2024/04/meta-llama3

https://twitter.com/hpp_ricecake/status/1781660882835313084
Swallow-MX

https://huggingface.co/Crystalcareai/llama-3-4x8b

https://twitter.com/npaka123/status/1781050851672625232
TF4.40

https://twitter.com/jaguring1/status/1781253845986976186

https://twitter.com/_stakaya/status/1781293875992342714
rag--

https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct

https://ai.meta.com/blog/meta-llama-3/

https://twitter.com/hillbig/status/1781090215995830734

Llama 3
https://note.com/npaka/n/n73b0786f48e9?sub_rt=share_h

https://twitter.com/shion_honda/status/1781064322048798787

https://twitter.com/yuiseki_/status/1780794904312246691
>MistralのTokenizer

https://twitter.com/tdualdir/status/1780403302423023708
https://zenn.dev/knowledgesense/articles/508187f1c616e3
Rag CoA

https://twitter.com/Maxwell_110/status/1780354228831318146

StockGPT: A GenAI Model for Stock Prediction and Trading
https://arxiv.org/abs/2404.05101

https://twitter.com/kotoba_tech/status/1780080789637730749