| { |
| "format": "tensorstore", |
| "version": "0.0.98", |
| "prefixes": { |
| "model": [ |
| { |
| "path": "model/lm_head/kernel", |
| "shape": [ |
| 5120, |
| 248320 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/embed_tokens/embedding", |
| "shape": [ |
| 248320, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/0/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/1/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/2/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/3/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/4/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/5/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/6/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/7/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/8/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/9/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/10/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/11/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/12/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/13/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/14/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/15/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/16/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/17/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/18/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/19/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/20/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/21/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/22/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/23/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/24/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/25/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/26/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/27/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/28/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/29/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/30/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/31/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/32/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/33/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/34/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/35/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/36/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/37/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/38/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/39/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/40/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/41/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/42/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/43/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/44/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/45/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/46/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/47/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/48/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/49/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/50/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/51/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/52/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/53/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/54/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/55/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/56/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/57/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/58/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/59/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/60/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/61/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/A_log", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/conv1d/kernel", |
| "shape": [ |
| 4, |
| 1, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/dt_bias", |
| "shape": [ |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/in_proj_a/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/in_proj_b/kernel", |
| "shape": [ |
| 5120, |
| 48 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/in_proj_qkv/kernel", |
| "shape": [ |
| 5120, |
| 10240 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/in_proj_z/kernel", |
| "shape": [ |
| 5120, |
| 6144 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/norm/kernel", |
| "shape": [ |
| 128 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/linear_attn/out_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/62/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/input_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/mlp/down_proj/kernel", |
| "shape": [ |
| 17408, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/mlp/gate_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/mlp/up_proj/kernel", |
| "shape": [ |
| 5120, |
| 17408 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/post_attention_layernorm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/self_attn/k_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/self_attn/k_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/self_attn/o_proj/kernel", |
| "shape": [ |
| 6144, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/self_attn/q_norm/kernel", |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/self_attn/q_proj/kernel", |
| "shape": [ |
| 5120, |
| 12288 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/layers/63/self_attn/v_proj/kernel", |
| "shape": [ |
| 5120, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/language_model/norm/kernel", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/0/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/1/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/2/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/3/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/4/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/5/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/6/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/7/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/8/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/9/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/10/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/11/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/12/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/13/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/14/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/15/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/16/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/17/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/18/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/19/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/20/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/21/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/22/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/23/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/24/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/25/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/attn/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/attn/proj/kernel", |
| "shape": [ |
| 1152, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/attn/qkv/bias", |
| "shape": [ |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/attn/qkv/kernel", |
| "shape": [ |
| 1152, |
| 3456 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/mlp/linear_fc1/bias", |
| "shape": [ |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/mlp/linear_fc1/kernel", |
| "shape": [ |
| 1152, |
| 4304 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/mlp/linear_fc2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/mlp/linear_fc2/kernel", |
| "shape": [ |
| 4304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/norm1/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/norm1/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/norm2/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/blocks/26/norm2/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/merger/linear_fc1/bias", |
| "shape": [ |
| 4608 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/merger/linear_fc1/kernel", |
| "shape": [ |
| 4608, |
| 4608 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/merger/linear_fc2/bias", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/merger/linear_fc2/kernel", |
| "shape": [ |
| 4608, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/merger/norm/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/merger/norm/scale", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/patch_embed/proj/bias", |
| "shape": [ |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/patch_embed/proj/kernel", |
| "shape": [ |
| 2, |
| 16, |
| 16, |
| 3, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| }, |
| { |
| "path": "model/model/visual/pos_embed/embedding", |
| "shape": [ |
| 2304, |
| 1152 |
| ], |
| "dtype": "bfloat16" |
| } |
| ] |
| } |
| } |
