| # -*- coding: utf-8 -*- | |
| """ | |
| Subspace Decoder Package | |
| A Transformer decoder implementation with Multi-Head Latent Attention (MLA) | |
| and decomposed MLP layers for efficient parameter usage. | |
| """ | |
| # Import all the main classes from models | |
| from .models import ( | |
| SharedSpaceDecoderConfig, | |
| SharedSpaceDecoderPreTrainedModel, | |
| SharedSpaceDecoderModel, | |
| SharedSpaceDecoderForCausalLM, | |
| ) | |
| __version__ = "0.1.0" | |
| __all__ = [ | |
| "SharedSpaceDecoderConfig", | |
| "SharedSpaceDecoderPreTrainedModel", | |
| "SharedSpaceDecoderModel", | |
| "SharedSpaceDecoderForCausalLM", | |
| ] | |