| { |
| "schema_version": 1, |
| "architecture": "causal_conv1d_fsq_action_decoder", |
| "representation": "human_action_138_v1", |
| "fps": 20, |
| "feature_size": 138, |
| "projected_feature_size": 139, |
| "codebook_size": 65536, |
| "levels": [ |
| 16, |
| 16, |
| 16, |
| 16 |
| ], |
| "decoder": { |
| "feature_size": 139, |
| "hidden_dim": 768, |
| "temporal_kernel_size": 3, |
| "num_stages": 20, |
| "channel_blocks_per_stage": 1, |
| "dilations": [ |
| 1, |
| 2, |
| 4, |
| 8, |
| 1, |
| 2, |
| 4, |
| 8, |
| 1, |
| 2, |
| 4, |
| 8, |
| 1, |
| 2, |
| 4, |
| 8, |
| 1, |
| 2, |
| 4, |
| 8 |
| ], |
| "expansion_ratio": 4, |
| "num_norm_groups": 32, |
| "norm_type": "weight_norm", |
| "causal": true, |
| "use_sandwich_block": false, |
| "proj_out_kernel_size": 5, |
| "z_channels": 4 |
| } |
| } |
|
|
