- state.param_states.decoder.decoder.depth_decoder.depth_layers_0.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.depth_decoder.depth_layers_1.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.depth_decoder.depth_layers_1.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.depth_decoder.depth_layers_2.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.depth_decoder.depth_layers_3.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_0.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_0.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_11.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_12.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_13.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_15.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_16.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_19.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_19.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_2.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_3.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_3.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_4.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_5.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_5.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_5.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_6.pre_cross_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_6.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_6.pre_self_attention_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.layers_7.pre_mlp_layer_norm.scale.v
- state.param_states.decoder.decoder.temporal_decoder.relpos_bias.rel_embedding.v
- state.param_states.encoder.layers_1.pre_mlp_layer_norm.scale.v
- state.param_states.encoder.layers_11.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_12.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_15.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_15.pre_mlp_layer_norm.scale.v
- state.param_states.encoder.layers_16.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_19.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_2.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_21.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_22.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_23.pre_mlp_layer_norm.scale.v
- state.param_states.encoder.layers_4.pre_mlp_layer_norm.scale.v
- state.param_states.encoder.layers_5.pre_mlp_layer_norm.scale.v
- state.param_states.encoder.layers_6.pre_attention_layer_norm.scale.v
- state.param_states.encoder.layers_6.pre_mlp_layer_norm.scale.v
- state.param_states.encoder.layers_9.pre_attention_layer_norm.scale.v
- target.decoder.decoder.depth_decoder.depth_layers_0.mlp.wi_0.kernel
- target.decoder.decoder.depth_decoder.depth_layers_0.self_attention.key.kernel
- target.decoder.decoder.depth_decoder.depth_layers_0.self_attention.value.kernel
- target.decoder.decoder.depth_decoder.depth_layers_1.self_attention.out.kernel
- target.decoder.decoder.depth_decoder.depth_layers_2.pre_self_attention_layer_norm.scale
- target.decoder.decoder.depth_decoder.depth_layers_2.self_attention.key.kernel
- target.decoder.decoder.depth_decoder.depth_layers_2.self_attention.out.kernel
- target.decoder.decoder.depth_decoder.depth_layers_3.mlp.wi_1.kernel