my collection of pretrained weights for SRLM model.
The pretraining is incomplete. Also, most of the weights here are old and irrelevant. Newer weights coming for 'attn' -branch soon.
SRLM is a language model built from S5-layers. It is a diffusion model as in "Diffusion Modeling by Estimating the Ratios of the Data Distribution", And it's equipped with a hierarchical reasoning model as a middle-layer.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support