| """Structural diagnostics for attention distributions, MoE expert balance, and gradient flow.""" | |
| from .attention_analysis import compute_attention_entropy, compute_attention_sparsity | |
| from .moe_analysis import compute_moe_load_balance | |
| from .gradient_analysis import compute_gradient_norm_stats | |
| __all__ = [ | |
| "compute_attention_entropy", | |
| "compute_attention_sparsity", | |
| "compute_moe_load_balance", | |
| "compute_gradient_norm_stats", | |
| ] | |
Xet Storage Details
- Size:
- 455 Bytes
- Xet hash:
- 2a358f3ccc0f08f55a2fb1326b1ce4ea0f3fb5f38ed2ec963c64abff61c7dae3
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.