anlp-a2-part1-moe_top2
Decoder-only transformer trained from scratch to translate Vietnamese and Japanese to English
(belumind/en-vi-ja-curated-500k-triplets). FFN variant: moe_top2.
- total params: 17,020,800 | active params per token: 13,481,856
- trained on 39,003,133 tokens
- W&B run: https://wandb.ai/dunkronit-iiit-hyderabad/anlp-a2-part1/runs/moe_top2-8210c110
Input format: <vi> source <en> or <ja> source <en>; the model continues with English and <eos>.
Load with src.part1.model.Transformer.from_pretrained("DunkRonit/anlp-a2-part1-moe_top2") from the assignment repo.
- Downloads last month
- 27