Download checkpoint_keys.json from HorizonRobotics/CogWAM: direct link, hf CLI and curl.
- Browser
- Download file 220 kB
-
https://huggingface.co/HorizonRobotics/CogWAM/resolve/main/checkpoint_keys.json
- Command line
-
hf download hf://HorizonRobotics/CogWAM/checkpoint_keys.json
-
curl -L -o checkpoint_keys.json https://huggingface.co/HorizonRobotics/CogWAM/resolve/main/checkpoint_keys.json
220 kB
| { | |
| "action_model.action_context.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_context.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.action_context.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_context.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_input.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_input.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 14 | |
| ] | |
| ], | |
| "action_model.action_output.bias": [ | |
| "bfloat16", | |
| [ | |
| 14 | |
| ] | |
| ], | |
| "action_model.action_output.weight": [ | |
| "bfloat16", | |
| [ | |
| 14, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_time_embedding.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_time_embedding.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 256 | |
| ] | |
| ], | |
| "action_model.action_time_embedding.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_time_embedding.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.action_time_projection.1.bias": [ | |
| "bfloat16", | |
| [ | |
| 6144 | |
| ] | |
| ], | |
| "action_model.action_time_projection.1.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.0.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.0.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.0.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.0.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.0.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.1.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.1.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.1.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.1.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.1.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.10.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.10.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.10.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.10.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.10.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.11.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.11.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.11.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.11.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.11.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.12.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.12.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.12.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.12.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.12.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.13.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.13.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.13.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.13.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.13.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.14.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.14.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.14.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.14.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.14.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.15.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.15.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.15.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.15.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.15.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.16.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.16.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.16.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.16.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.16.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.17.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.17.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.17.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.17.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.17.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.18.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.18.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.18.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.18.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.18.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.19.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.19.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.19.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.19.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.19.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.2.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.2.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.2.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.2.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.2.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.20.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.20.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.20.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.20.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.20.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.21.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.21.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.21.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.21.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.21.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.22.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.22.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.22.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.22.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.22.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.23.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.23.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.23.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.23.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.23.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.24.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.24.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.24.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.24.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.24.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.25.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.25.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.25.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.25.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.25.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.26.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.26.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.26.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.26.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.26.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.27.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.27.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.27.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.27.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.27.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.28.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.28.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.28.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.28.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.28.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.29.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.29.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.29.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.29.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.29.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.3.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.3.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.3.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.3.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.3.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.4.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.4.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.4.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.4.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.4.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.5.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.5.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.5.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.5.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.5.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.6.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.6.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.6.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.6.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.6.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.7.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.7.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.7.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.7.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.7.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.8.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.8.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.8.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.8.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.8.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.9.action.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "action_model.layers.9.action.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.action.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.cross_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.ffn.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.9.world.ffn.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.ffn.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.ffn.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.layers.9.world.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 6, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.norm3.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.norm3.weight": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.k.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.norm_k.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.norm_q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.o.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.o.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.q.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.q.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.v.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.layers.9.world.self_attn.v.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_model.state_to_planner.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "action_model.state_to_planner.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 14 | |
| ] | |
| ], | |
| "action_model.world_context.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_context.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "action_model.world_context.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_context.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_frame_embedding": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 2, | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_input.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_input.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 768 | |
| ] | |
| ], | |
| "action_model.world_output.modulation": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 2, | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_output.output.bias": [ | |
| "bfloat16", | |
| [ | |
| 768 | |
| ] | |
| ], | |
| "action_model.world_output.output.weight": [ | |
| "bfloat16", | |
| [ | |
| 768, | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_time_embedding.0.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_time_embedding.0.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 256 | |
| ] | |
| ], | |
| "action_model.world_time_embedding.2.bias": [ | |
| "bfloat16", | |
| [ | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_time_embedding.2.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 512 | |
| ] | |
| ], | |
| "action_model.world_time_projection.1.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "action_model.world_time_projection.1.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 512 | |
| ] | |
| ], | |
| "action_plan_queries.embedding": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 25, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.lm_head.weight": [ | |
| "bfloat16", | |
| [ | |
| 248081, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.embed_tokens.weight": [ | |
| "bfloat16", | |
| [ | |
| 248081, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.0.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.1.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.10.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.self_attn.k_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.self_attn.k_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.self_attn.o_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.self_attn.q_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.self_attn.q_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.11.self_attn.v_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.12.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.13.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.14.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.self_attn.k_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.self_attn.k_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.self_attn.o_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.self_attn.q_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.self_attn.q_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.15.self_attn.v_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.16.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.17.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.18.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.self_attn.k_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.self_attn.k_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.self_attn.o_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.self_attn.q_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.self_attn.q_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.19.self_attn.v_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.2.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.20.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.21.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.22.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.self_attn.k_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.self_attn.k_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.self_attn.o_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.self_attn.q_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.self_attn.q_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.23.self_attn.v_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.self_attn.k_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.self_attn.k_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.self_attn.o_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.self_attn.q_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.self_attn.q_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.3.self_attn.v_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.4.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.5.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.6.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.self_attn.k_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.self_attn.k_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.self_attn.o_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.self_attn.q_norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 256 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.self_attn.q_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.7.self_attn.v_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 512, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.8.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.input_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.A_log": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.conv1d.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 1, | |
| 4 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.dt_bias": [ | |
| "bfloat16", | |
| [ | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.in_proj_a.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.in_proj_b.weight": [ | |
| "bfloat16", | |
| [ | |
| 16, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.in_proj_qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.in_proj_z.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 128 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.linear_attn.out_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.mlp.down_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 6144 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.mlp.gate_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.mlp.up_proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 6144, | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.layers.9.post_attention_layernorm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.language_model.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.0.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.1.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.10.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.11.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.12.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.13.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.14.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.15.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.16.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.17.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.18.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.19.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.2.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.20.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.21.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.22.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.23.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.3.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.4.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.5.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.6.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.7.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.8.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.attn.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.attn.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.attn.qkv.bias": [ | |
| "bfloat16", | |
| [ | |
| 3072 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.attn.qkv.weight": [ | |
| "bfloat16", | |
| [ | |
| 3072, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.mlp.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.mlp.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.mlp.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.mlp.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.norm1.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.norm1.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.norm2.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.blocks.9.norm2.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.merger.linear_fc1.bias": [ | |
| "bfloat16", | |
| [ | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.merger.linear_fc1.weight": [ | |
| "bfloat16", | |
| [ | |
| 4096, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.merger.linear_fc2.bias": [ | |
| "bfloat16", | |
| [ | |
| 2048 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.merger.linear_fc2.weight": [ | |
| "bfloat16", | |
| [ | |
| 2048, | |
| 4096 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.merger.norm.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.merger.norm.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.patch_embed.proj.bias": [ | |
| "bfloat16", | |
| [ | |
| 1024 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.patch_embed.proj.weight": [ | |
| "bfloat16", | |
| [ | |
| 1024, | |
| 3, | |
| 2, | |
| 16, | |
| 16 | |
| ] | |
| ], | |
| "qwen_vl_interface.model.model.visual.pos_embed.weight": [ | |
| "bfloat16", | |
| [ | |
| 2304, | |
| 1024 | |
| ] | |
| ], | |
| "world_plan_queries.embedding": [ | |
| "bfloat16", | |
| [ | |
| 1, | |
| 16, | |
| 2048 | |
| ] | |
| ] | |
| } | |