Spaces:

ASTERIZER
/

luna-300m-program

Running

App Files Files Community

luna-300m-program / README.md

REZIRETSA

Revert space gradient to gray

108e4aa verified 5 days ago

preview code

Raw

History Blame Contribute Delete

1.05 kB

metadata

title: LUNA-300M Program
emoji: 🌙
colorFrom: gray
colorTo: gray
sdk: static
pinned: true
license: other
short_description: LUNA-300M model, training data and program links

LUNA-300M Program

Program hub for LUNA-300M — a ~303M-parameter English causal language model pretrained from scratch by ASTERIZER on the 4.5B-token LUNA_PreTrain corpus.

Assets

Model: ASTERIZER/LUNA-300M (pretrained checkpoints, step 31,052)
Sibling: ASTERIZER/LUNA-100M (trained, RAG+MCP SFT)
Corpus: ASTERIZER/LUNA_PreTrain (4.5B tokens, 270 litdata chunks)
Workspace: ASTERIZER/LUNA (training code, train_300m.py)
Collection: LUNA-300M Program