layer_10.bin.json

{"model.layers.10.self_attn.q_proj.weight.bits": {"offset": 0, "shape": [16384, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_proj.weight.su": {"offset": 1572864, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_proj.weight.sv": {"offset": 1576960, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_proj.weight.isc": {"offset": 1581056, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_proj.weight.osc": {"offset": 1585152, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_proj.weight.scale": {"offset": 1589248, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_proj.weight": {"quant": true, "n_out": 2048, "k_in": 2048, "K": 3, "block": 128, "tiles_out": 128, "tiles_in": 128, "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight.bits": {"offset": 1589252, "shape": [8192, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight.su": {"offset": 2375684, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight.sv": {"offset": 2379780, "shape": [1024], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight.isc": {"offset": 2381828, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight.osc": {"offset": 2385924, "shape": [1024], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight.scale": {"offset": 2387972, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_proj.weight": {"quant": true, "n_out": 1024, "k_in": 2048, "K": 3, "block": 128, "tiles_out": 64, "tiles_in": 128, "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight.bits": {"offset": 2387976, "shape": [8192, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight.su": {"offset": 3174408, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight.sv": {"offset": 3178504, "shape": [1024], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight.isc": {"offset": 3180552, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight.osc": {"offset": 3184648, "shape": [1024], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight.scale": {"offset": 3186696, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.v_proj.weight": {"quant": true, "n_out": 1024, "k_in": 2048, "K": 3, "block": 128, "tiles_out": 64, "tiles_in": 128, "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight.bits": {"offset": 3186700, "shape": [16384, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight.su": {"offset": 4759564, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight.sv": {"offset": 4763660, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight.isc": {"offset": 4767756, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight.osc": {"offset": 4771852, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight.scale": {"offset": 4775948, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.self_attn.o_proj.weight": {"quant": true, "n_out": 2048, "k_in": 2048, "K": 3, "block": 128, "tiles_out": 128, "tiles_in": 128, "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight.bits": {"offset": 4775952, "shape": [49152, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight.su": {"offset": 9494544, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight.sv": {"offset": 9498640, "shape": [6144], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight.isc": {"offset": 9510928, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight.osc": {"offset": 9515024, "shape": [6144], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight.scale": {"offset": 9527312, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.mlp.gate_proj.weight": {"quant": true, "n_out": 6144, "k_in": 2048, "K": 3, "block": 128, "tiles_out": 384, "tiles_in": 128, "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight.bits": {"offset": 9527316, "shape": [49152, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight.su": {"offset": 14245908, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight.sv": {"offset": 14250004, "shape": [6144], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight.isc": {"offset": 14262292, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight.osc": {"offset": 14266388, "shape": [6144], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight.scale": {"offset": 14278676, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.mlp.up_proj.weight": {"quant": true, "n_out": 6144, "k_in": 2048, "K": 3, "block": 128, "tiles_out": 384, "tiles_in": 128, "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight.bits": {"offset": 14278680, "shape": [49152, 24], "dtype": "u32", "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight.su": {"offset": 18997272, "shape": [6144], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight.sv": {"offset": 19009560, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight.isc": {"offset": 19013656, "shape": [6144], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight.osc": {"offset": 19025944, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight.scale": {"offset": 19030040, "shape": [1], "dtype": "f32", "shard": "layer_10.bin"}, "model.layers.10.mlp.down_proj.weight": {"quant": true, "n_out": 2048, "k_in": 6144, "K": 3, "block": 128, "tiles_out": 128, "tiles_in": 384, "shard": "layer_10.bin"}, "model.layers.10.input_layernorm.weight": {"offset": 19030044, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.post_attention_layernorm.weight": {"offset": 19034140, "shape": [2048], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.q_norm.weight": {"offset": 19038236, "shape": [128], "dtype": "f16", "shard": "layer_10.bin"}, "model.layers.10.self_attn.k_norm.weight": {"offset": 19038492, "shape": [128], "dtype": "f16", "shard": "layer_10.bin"}}