You need to agree to share your Hugging Face information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

AuriStream1B_40Pred_BabyView_200k

AuriStream 1B (48 layers, d=1280, 16 heads, 40 future-prediction heads) trained for 200k steps on BabyView audio, tokenized with the 5 ms / 8192-code WavCoch tokenizer TuKoResearch/WavCochCausalV8192-5ms-babyview. Inputs must be codes from that tokenizer.

Source run: AuriStream1B_40Pred_BabyView5ms8k_64pod_wd020_do005 (weight decay 0.20, dropout 0.05), checkpoint model_00200000.pt.

from transformers import AutoModel
model = AutoModel.from_pretrained("TuKoResearch/AuriStream1B_40Pred_BabyView_200k", trust_remote_code=True).eval()

Downloads last month: 5

Safetensors

Model size

1B params

Tensor type

F32