Organization Card

Multilingual open models with data provenance and source attribution.

Models in this space are reserved for leading efforts in responsible AI development. Few projects meet entry criteria; weak humane standards are still stringent for ML. This field... we humans... are capable of better.

Each of us make the world. This is a call to improve the situation. Learn, study, test, grow, and step by step create a better future. What does it look like? What can we achieve working together towards this dream?

We build our answers.

https://github.com/darkshapes


Scraped/unscrupulous fine-tuning data      ❌
Unreleased/unspecified datasets            ❌
High quality results only in English       ❌
Proprietary code                           ❌
Uncredited contributions                   ❌
Restrictive licencing                      ❌
Undocumented funding                       ❌



Consenting/fairly compensated data sources 🆗
Synthetic fine-tuning data                 🆗
English as 2nd language / Multilingual     🆗
Available source code                      🆗
CO^2 metrics / Training hardware specifics 🆗
Permissive licencing                       🆗
Disclosed revenue sources/training costs   🆗

models 16

datasets

None public yet

darkshapes

AI & ML interests

Recent Activity

Multilingual open models with data provenance and source attribution.

models 16

darkshapes/ruadaptqwen2.5-7B-ext-u48-instruct-gguf

darkshapes/ruadaptqwen2.5-32b-instruct-gguf

darkshapes/ruadaptqwen2.5-3b-ins-gguf

darkshapes/suzume-llama-3-8B-multilingual-orpo-borda-top25-gguf

darkshapes/mistral-7b-v0.3-chinese-chat-4bit

darkshapes/aya-23-8b-gguf

darkshapes/codeninja-1.0-openchat-7b-gguf

darkshapes/mistral-7b-openorca-gguf

darkshapes/mistral-7b-openorca-oasst_top1_230825v2-gguf

darkshapes/mt0-base

datasets

AI & ML interests

Recent Activity

Team members 2

Multilingual open models with data provenance and source attribution.

models 16 Sort: Recently updated

datasets

models 16