Hugging Face's logo

Ezekiel999
/
AksaraLLM-20B-Instruct

Transformers
Safetensors
Indonesian
aksarallm
indonesian
from-scratch
smoke-test
Model card Files
xet
 Community
AksaraLLM-20B-Instruct / README.md
Ezekiel999's picture
Ezekiel999
Upload tiny dry-run checkpoint (smoke test, not production)
4a62ed1 verified
preview code
|
raw
history blame contribute delete
614 Bytes
metadata 
license: apache-2.0
language: id
library_name: transformers
tags:
  - aksarallm
  - indonesian
  - from-scratch
  - smoke-test

Ezekiel999/AksaraLLM-20B-Instruct (smoke-test checkpoint)

This is NOT the production 20B model. It is a randomly-initialized tiny preset (2 layers, 64-dim, vocab 256) pushed from a Devin scaffolding session to validate the aksaraLLMModel.save_pretrained → HF → aksaraLLMModel.from_pretrained round-trip.

The real 20B model (42 layers, 6144-dim, vocab 131 072) must be trained from random initialisation on a TPU v5p pod using aksara-train/scripts/train_20b_pretrain.py.