The pretrained uTCN-300 checkpoint (causal, four blocks) for the micro-TCN Space, an emulation of the LA-2A compressor, mirrored from the GitHub repository of its author. The file is unchanged, a TorchScript trace whose weights the Space loads into its own copy of the network.
traced_3-uTCN-300__causal__4-10-13__fraction-1.0-bs32.pt: from https://github.com/csteinmetz1/neural-2a/blob/d9098285685231f20ccdfa13a3c65f3478585851/models/traced_3-uTCN-300__causal__4-10-13__fraction-1.0-bs32.pt, SHA-256d4f48bcab59ab8163448c7a2c39e7c2e993e32ba2821b61f8869169631e7c5f5.
By Christian J. Steinmetz. The micro-tcn code is Apache-2.0, and the neural-2a repository states that anything from its author in it is free to use for any purpose.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support