You can deploy this model and start earning money today!

tiagoCuervo/
gslm-scaling-85m-43p5b

Name: gslm-scaling-85m-43p5b
Availability: InStock

slm-research

safetensors

slm

speech-language-model

audio

text-generation

apache-2.0

Views

No views yet

gslm-scaling-85m-43p5b

Model-only FP32 weights for a final-budget scaling-curve endpoint from Scaling Properties of Speech Language Models (EMNLP 2024). This checkpoint has 85,339,392 parameters and saw exactly 43,545,280,000 tokens at 262,400 tokens per update.

python

import torch
from slm import TransformerLM

model = TransformerLM.from_pretrained("tiagoCuervo/gslm-scaling-85m-43p5b", device="cuda")
prompt = torch.tensor([[12, 91, 204]], device="cuda")
units = model.generate(prompt, max_new_tokens=200, temperature=0.8, top_k=50)

Evaluation scores are 58.90% sBLIMP phenomenon macro accuracy, 56.04% sBLIMP voice-pair accuracy, 54.57% sStoryCloze, and 72.37% tStoryCloze.

The model predicts consecutive-run-collapsed 25 Hz layer-11 mHuBERT K-means units (K=500; EOS=500; PAD=501). A vocoder is not bundled.

Speech-tokenizer and vocoder artifacts are credited in the SLM third-party notices. See provenance.json for exact hashes and evaluation metadata.