{ "quant_method": "exl3", "version": "0.0.25", "bits": 3.5, "head_bits": 6, "calibration": { "rows": 250, "cols": 2048 }, "out_scales": "always", "codebook": "mcg", "tensor_storage": { "model.embed_tokens": { "stored_tensors": { "model.embed_tokens.weight": { "shape": [ 100352, 3840 ], "n_bytes": 770703360, "dtype": "torch.float16" } } }, "model.layers.0.input_layernorm": { "stored_tensors": { "model.layers.0.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.0.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.0.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.0.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.0.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.0.linear_attn.g_proj": { "stored_tensors": { "model.layers.0.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.0.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.0.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.0.linear_attn.b_proj": { "stored_tensors": { "model.layers.0.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.0.linear_attn.a_proj": { "stored_tensors": { "model.layers.0.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.0.linear_attn.o_proj": { "stored_tensors": { "model.layers.0.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.0.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.0.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.0.linear_attn.o_norm": { "stored_tensors": { "model.layers.0.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.0.post_attention_layernorm": { "stored_tensors": { "model.layers.0.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.0.mlp.up_proj": { "stored_tensors": { "model.layers.0.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.0.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.0.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.0.mlp.gate_proj": { "stored_tensors": { "model.layers.0.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.0.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.0.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.0.mlp.down_proj": { "stored_tensors": { "model.layers.0.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.0.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.0.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.1.input_layernorm": { "stored_tensors": { "model.layers.1.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.1.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.1.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.1.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.1.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.1.linear_attn.g_proj": { "stored_tensors": { "model.layers.1.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.1.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.1.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.1.linear_attn.b_proj": { "stored_tensors": { "model.layers.1.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.1.linear_attn.a_proj": { "stored_tensors": { "model.layers.1.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.1.linear_attn.o_proj": { "stored_tensors": { "model.layers.1.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.1.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.1.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.1.linear_attn.o_norm": { "stored_tensors": { "model.layers.1.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.1.post_attention_layernorm": { "stored_tensors": { "model.layers.1.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.1.mlp.up_proj": { "stored_tensors": { "model.layers.1.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.1.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.1.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.1.mlp.gate_proj": { "stored_tensors": { "model.layers.1.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.1.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.1.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.1.mlp.down_proj": { "stored_tensors": { "model.layers.1.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.1.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.1.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.2.input_layernorm": { "stored_tensors": { "model.layers.2.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.2.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.2.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.2.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.2.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.2.linear_attn.g_proj": { "stored_tensors": { "model.layers.2.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.2.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.2.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.2.linear_attn.b_proj": { "stored_tensors": { "model.layers.2.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.2.linear_attn.a_proj": { "stored_tensors": { "model.layers.2.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.2.linear_attn.o_proj": { "stored_tensors": { "model.layers.2.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.2.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.2.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.2.linear_attn.o_norm": { "stored_tensors": { "model.layers.2.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.2.post_attention_layernorm": { "stored_tensors": { "model.layers.2.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.2.mlp.up_proj": { "stored_tensors": { "model.layers.2.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.2.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.2.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.2.mlp.gate_proj": { "stored_tensors": { "model.layers.2.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.2.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.2.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.2.mlp.down_proj": { "stored_tensors": { "model.layers.2.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.2.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.2.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.3.self_attn.q_proj": { "stored_tensors": { "model.layers.3.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.3.self_attn.k_proj": { "stored_tensors": { "model.layers.3.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.3.self_attn.v_proj": { "stored_tensors": { "model.layers.3.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.3.self_attn.o_proj": { "stored_tensors": { "model.layers.3.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.3.self_attn.q_norm": { "stored_tensors": { "model.layers.3.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.3.self_attn.k_norm": { "stored_tensors": { "model.layers.3.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.3.post_attention_layernorm": { "stored_tensors": { "model.layers.3.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.3.mlp.up_proj": { "stored_tensors": { "model.layers.3.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.3.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.3.mlp.gate_proj": { "stored_tensors": { "model.layers.3.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.3.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.3.mlp.down_proj": { "stored_tensors": { "model.layers.3.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.3.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.3.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.3.post_feedforward_layernorm": { "stored_tensors": { "model.layers.3.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.4.input_layernorm": { "stored_tensors": { "model.layers.4.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.4.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.4.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.4.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.4.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.4.linear_attn.g_proj": { "stored_tensors": { "model.layers.4.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.4.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.4.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.4.linear_attn.b_proj": { "stored_tensors": { "model.layers.4.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.4.linear_attn.a_proj": { "stored_tensors": { "model.layers.4.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.4.linear_attn.o_proj": { "stored_tensors": { "model.layers.4.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.4.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.4.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.4.linear_attn.o_norm": { "stored_tensors": { "model.layers.4.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.4.post_attention_layernorm": { "stored_tensors": { "model.layers.4.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.4.mlp.up_proj": { "stored_tensors": { "model.layers.4.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.4.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.4.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.4.mlp.gate_proj": { "stored_tensors": { "model.layers.4.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.4.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.4.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.4.mlp.down_proj": { "stored_tensors": { "model.layers.4.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.4.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.4.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.5.input_layernorm": { "stored_tensors": { "model.layers.5.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.5.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.5.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.5.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.5.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.5.linear_attn.g_proj": { "stored_tensors": { "model.layers.5.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.5.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.5.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.5.linear_attn.b_proj": { "stored_tensors": { "model.layers.5.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.5.linear_attn.a_proj": { "stored_tensors": { "model.layers.5.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.5.linear_attn.o_proj": { "stored_tensors": { "model.layers.5.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.5.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.5.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.5.linear_attn.o_norm": { "stored_tensors": { "model.layers.5.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.5.post_attention_layernorm": { "stored_tensors": { "model.layers.5.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.5.mlp.up_proj": { "stored_tensors": { "model.layers.5.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.5.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.5.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.5.mlp.gate_proj": { "stored_tensors": { "model.layers.5.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.5.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.5.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.5.mlp.down_proj": { "stored_tensors": { "model.layers.5.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.5.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.5.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.6.input_layernorm": { "stored_tensors": { "model.layers.6.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.6.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.6.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.6.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.6.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.6.linear_attn.g_proj": { "stored_tensors": { "model.layers.6.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.6.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.6.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.6.linear_attn.b_proj": { "stored_tensors": { "model.layers.6.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.6.linear_attn.a_proj": { "stored_tensors": { "model.layers.6.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.6.linear_attn.o_proj": { "stored_tensors": { "model.layers.6.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.6.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.6.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.6.linear_attn.o_norm": { "stored_tensors": { "model.layers.6.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.6.post_attention_layernorm": { "stored_tensors": { "model.layers.6.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.6.mlp.up_proj": { "stored_tensors": { "model.layers.6.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.6.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.6.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.6.mlp.gate_proj": { "stored_tensors": { "model.layers.6.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.6.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.6.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.6.mlp.down_proj": { "stored_tensors": { "model.layers.6.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.6.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.6.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.7.self_attn.q_proj": { "stored_tensors": { "model.layers.7.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.7.self_attn.k_proj": { "stored_tensors": { "model.layers.7.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.7.self_attn.v_proj": { "stored_tensors": { "model.layers.7.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.7.self_attn.o_proj": { "stored_tensors": { "model.layers.7.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.7.self_attn.q_norm": { "stored_tensors": { "model.layers.7.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.7.self_attn.k_norm": { "stored_tensors": { "model.layers.7.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.7.post_attention_layernorm": { "stored_tensors": { "model.layers.7.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.7.mlp.up_proj": { "stored_tensors": { "model.layers.7.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.7.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.7.mlp.gate_proj": { "stored_tensors": { "model.layers.7.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.7.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.7.mlp.down_proj": { "stored_tensors": { "model.layers.7.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.7.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.7.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.7.post_feedforward_layernorm": { "stored_tensors": { "model.layers.7.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.8.input_layernorm": { "stored_tensors": { "model.layers.8.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.8.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.8.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.8.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.8.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.8.linear_attn.g_proj": { "stored_tensors": { "model.layers.8.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.8.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.8.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.8.linear_attn.b_proj": { "stored_tensors": { "model.layers.8.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.8.linear_attn.a_proj": { "stored_tensors": { "model.layers.8.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.8.linear_attn.o_proj": { "stored_tensors": { "model.layers.8.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.8.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.8.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.8.linear_attn.o_norm": { "stored_tensors": { "model.layers.8.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.8.post_attention_layernorm": { "stored_tensors": { "model.layers.8.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.8.mlp.up_proj": { "stored_tensors": { "model.layers.8.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.8.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.8.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.8.mlp.gate_proj": { "stored_tensors": { "model.layers.8.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.8.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.8.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.8.mlp.down_proj": { "stored_tensors": { "model.layers.8.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.8.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.8.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.9.input_layernorm": { "stored_tensors": { "model.layers.9.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.9.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.9.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.9.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.9.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.9.linear_attn.g_proj": { "stored_tensors": { "model.layers.9.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.9.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.9.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.9.linear_attn.b_proj": { "stored_tensors": { "model.layers.9.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.9.linear_attn.a_proj": { "stored_tensors": { "model.layers.9.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.9.linear_attn.o_proj": { "stored_tensors": { "model.layers.9.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.9.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.9.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.9.linear_attn.o_norm": { "stored_tensors": { "model.layers.9.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.9.post_attention_layernorm": { "stored_tensors": { "model.layers.9.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.9.mlp.up_proj": { "stored_tensors": { "model.layers.9.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.9.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.9.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.9.mlp.gate_proj": { "stored_tensors": { "model.layers.9.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.9.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.9.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.9.mlp.down_proj": { "stored_tensors": { "model.layers.9.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.9.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.9.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.10.input_layernorm": { "stored_tensors": { "model.layers.10.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.10.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.10.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.10.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.10.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.10.linear_attn.g_proj": { "stored_tensors": { "model.layers.10.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.10.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.10.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.10.linear_attn.b_proj": { "stored_tensors": { "model.layers.10.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.10.linear_attn.a_proj": { "stored_tensors": { "model.layers.10.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.10.linear_attn.o_proj": { "stored_tensors": { "model.layers.10.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.10.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.10.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.10.linear_attn.o_norm": { "stored_tensors": { "model.layers.10.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.10.post_attention_layernorm": { "stored_tensors": { "model.layers.10.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.10.mlp.up_proj": { "stored_tensors": { "model.layers.10.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.10.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.10.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.10.mlp.gate_proj": { "stored_tensors": { "model.layers.10.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.10.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.10.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.10.mlp.down_proj": { "stored_tensors": { "model.layers.10.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.10.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.10.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.11.self_attn.q_proj": { "stored_tensors": { "model.layers.11.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.11.self_attn.k_proj": { "stored_tensors": { "model.layers.11.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.11.self_attn.v_proj": { "stored_tensors": { "model.layers.11.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.11.self_attn.o_proj": { "stored_tensors": { "model.layers.11.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.11.self_attn.q_norm": { "stored_tensors": { "model.layers.11.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.11.self_attn.k_norm": { "stored_tensors": { "model.layers.11.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.11.post_attention_layernorm": { "stored_tensors": { "model.layers.11.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.11.mlp.up_proj": { "stored_tensors": { "model.layers.11.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.11.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.11.mlp.gate_proj": { "stored_tensors": { "model.layers.11.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.11.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.11.mlp.down_proj": { "stored_tensors": { "model.layers.11.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.11.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.11.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.11.post_feedforward_layernorm": { "stored_tensors": { "model.layers.11.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.12.input_layernorm": { "stored_tensors": { "model.layers.12.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.12.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.12.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.12.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.12.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.12.linear_attn.g_proj": { "stored_tensors": { "model.layers.12.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.12.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.12.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.12.linear_attn.b_proj": { "stored_tensors": { "model.layers.12.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.12.linear_attn.a_proj": { "stored_tensors": { "model.layers.12.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.12.linear_attn.o_proj": { "stored_tensors": { "model.layers.12.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.12.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.12.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.12.linear_attn.o_norm": { "stored_tensors": { "model.layers.12.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.12.post_attention_layernorm": { "stored_tensors": { "model.layers.12.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.12.mlp.up_proj": { "stored_tensors": { "model.layers.12.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.12.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.12.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.12.mlp.gate_proj": { "stored_tensors": { "model.layers.12.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.12.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.12.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.12.mlp.down_proj": { "stored_tensors": { "model.layers.12.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.12.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.12.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.13.input_layernorm": { "stored_tensors": { "model.layers.13.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.13.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.13.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.13.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.13.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.13.linear_attn.g_proj": { "stored_tensors": { "model.layers.13.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.13.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.13.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.13.linear_attn.b_proj": { "stored_tensors": { "model.layers.13.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.13.linear_attn.a_proj": { "stored_tensors": { "model.layers.13.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.13.linear_attn.o_proj": { "stored_tensors": { "model.layers.13.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.13.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.13.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.13.linear_attn.o_norm": { "stored_tensors": { "model.layers.13.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.13.post_attention_layernorm": { "stored_tensors": { "model.layers.13.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.13.mlp.up_proj": { "stored_tensors": { "model.layers.13.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.13.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.13.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.13.mlp.gate_proj": { "stored_tensors": { "model.layers.13.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.13.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.13.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.13.mlp.down_proj": { "stored_tensors": { "model.layers.13.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.13.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.13.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.14.input_layernorm": { "stored_tensors": { "model.layers.14.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.14.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.14.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.14.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.14.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.14.linear_attn.g_proj": { "stored_tensors": { "model.layers.14.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.14.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.14.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.14.linear_attn.b_proj": { "stored_tensors": { "model.layers.14.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.14.linear_attn.a_proj": { "stored_tensors": { "model.layers.14.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.14.linear_attn.o_proj": { "stored_tensors": { "model.layers.14.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.14.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.14.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.14.linear_attn.o_norm": { "stored_tensors": { "model.layers.14.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.14.post_attention_layernorm": { "stored_tensors": { "model.layers.14.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.14.mlp.up_proj": { "stored_tensors": { "model.layers.14.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.14.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.14.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.14.mlp.gate_proj": { "stored_tensors": { "model.layers.14.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.14.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.14.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.14.mlp.down_proj": { "stored_tensors": { "model.layers.14.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.14.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.14.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.15.self_attn.q_proj": { "stored_tensors": { "model.layers.15.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.15.self_attn.k_proj": { "stored_tensors": { "model.layers.15.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.15.self_attn.v_proj": { "stored_tensors": { "model.layers.15.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.15.self_attn.o_proj": { "stored_tensors": { "model.layers.15.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.15.self_attn.q_norm": { "stored_tensors": { "model.layers.15.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.15.self_attn.k_norm": { "stored_tensors": { "model.layers.15.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.15.post_attention_layernorm": { "stored_tensors": { "model.layers.15.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.15.mlp.up_proj": { "stored_tensors": { "model.layers.15.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.15.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.15.mlp.gate_proj": { "stored_tensors": { "model.layers.15.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.15.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.15.mlp.down_proj": { "stored_tensors": { "model.layers.15.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.15.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.15.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.15.post_feedforward_layernorm": { "stored_tensors": { "model.layers.15.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.16.input_layernorm": { "stored_tensors": { "model.layers.16.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.16.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.16.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.16.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.16.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.16.linear_attn.g_proj": { "stored_tensors": { "model.layers.16.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.16.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.16.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.16.linear_attn.b_proj": { "stored_tensors": { "model.layers.16.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.16.linear_attn.a_proj": { "stored_tensors": { "model.layers.16.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.16.linear_attn.o_proj": { "stored_tensors": { "model.layers.16.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.16.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.16.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.16.linear_attn.o_norm": { "stored_tensors": { "model.layers.16.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.16.post_attention_layernorm": { "stored_tensors": { "model.layers.16.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.16.mlp.up_proj": { "stored_tensors": { "model.layers.16.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.16.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.16.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.16.mlp.gate_proj": { "stored_tensors": { "model.layers.16.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.16.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.16.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.16.mlp.down_proj": { "stored_tensors": { "model.layers.16.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.16.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.16.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.17.input_layernorm": { "stored_tensors": { "model.layers.17.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.17.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.17.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.17.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.17.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.17.linear_attn.g_proj": { "stored_tensors": { "model.layers.17.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.17.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.17.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.17.linear_attn.b_proj": { "stored_tensors": { "model.layers.17.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.17.linear_attn.a_proj": { "stored_tensors": { "model.layers.17.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.17.linear_attn.o_proj": { "stored_tensors": { "model.layers.17.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.17.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.17.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.17.linear_attn.o_norm": { "stored_tensors": { "model.layers.17.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.17.post_attention_layernorm": { "stored_tensors": { "model.layers.17.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.17.mlp.up_proj": { "stored_tensors": { "model.layers.17.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.17.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.17.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.17.mlp.gate_proj": { "stored_tensors": { "model.layers.17.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.17.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.17.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.17.mlp.down_proj": { "stored_tensors": { "model.layers.17.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.17.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.17.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.18.input_layernorm": { "stored_tensors": { "model.layers.18.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.18.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.18.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.18.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.18.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.18.linear_attn.g_proj": { "stored_tensors": { "model.layers.18.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.18.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.18.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.18.linear_attn.b_proj": { "stored_tensors": { "model.layers.18.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.18.linear_attn.a_proj": { "stored_tensors": { "model.layers.18.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.18.linear_attn.o_proj": { "stored_tensors": { "model.layers.18.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.18.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.18.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.18.linear_attn.o_norm": { "stored_tensors": { "model.layers.18.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.18.post_attention_layernorm": { "stored_tensors": { "model.layers.18.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.18.mlp.up_proj": { "stored_tensors": { "model.layers.18.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.18.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.18.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.18.mlp.gate_proj": { "stored_tensors": { "model.layers.18.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.18.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.18.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.18.mlp.down_proj": { "stored_tensors": { "model.layers.18.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.18.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.18.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.19.self_attn.q_proj": { "stored_tensors": { "model.layers.19.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.19.self_attn.k_proj": { "stored_tensors": { "model.layers.19.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.19.self_attn.v_proj": { "stored_tensors": { "model.layers.19.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.19.self_attn.o_proj": { "stored_tensors": { "model.layers.19.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.19.self_attn.q_norm": { "stored_tensors": { "model.layers.19.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.19.self_attn.k_norm": { "stored_tensors": { "model.layers.19.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.19.post_attention_layernorm": { "stored_tensors": { "model.layers.19.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.19.mlp.up_proj": { "stored_tensors": { "model.layers.19.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.19.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.19.mlp.gate_proj": { "stored_tensors": { "model.layers.19.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.19.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.19.mlp.down_proj": { "stored_tensors": { "model.layers.19.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.19.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.19.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.19.post_feedforward_layernorm": { "stored_tensors": { "model.layers.19.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.20.input_layernorm": { "stored_tensors": { "model.layers.20.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.20.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.20.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.20.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.20.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.20.linear_attn.g_proj": { "stored_tensors": { "model.layers.20.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.20.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.20.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.20.linear_attn.b_proj": { "stored_tensors": { "model.layers.20.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.20.linear_attn.a_proj": { "stored_tensors": { "model.layers.20.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.20.linear_attn.o_proj": { "stored_tensors": { "model.layers.20.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.20.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.20.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.20.linear_attn.o_norm": { "stored_tensors": { "model.layers.20.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.20.post_attention_layernorm": { "stored_tensors": { "model.layers.20.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.20.mlp.up_proj": { "stored_tensors": { "model.layers.20.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.20.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.20.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.20.mlp.gate_proj": { "stored_tensors": { "model.layers.20.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.20.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.20.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.20.mlp.down_proj": { "stored_tensors": { "model.layers.20.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.20.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.20.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.21.input_layernorm": { "stored_tensors": { "model.layers.21.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.21.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.21.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.21.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.21.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.21.linear_attn.g_proj": { "stored_tensors": { "model.layers.21.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.21.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.21.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.21.linear_attn.b_proj": { "stored_tensors": { "model.layers.21.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.21.linear_attn.a_proj": { "stored_tensors": { "model.layers.21.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.21.linear_attn.o_proj": { "stored_tensors": { "model.layers.21.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.21.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.21.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.21.linear_attn.o_norm": { "stored_tensors": { "model.layers.21.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.21.post_attention_layernorm": { "stored_tensors": { "model.layers.21.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.21.mlp.up_proj": { "stored_tensors": { "model.layers.21.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.21.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.21.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.21.mlp.gate_proj": { "stored_tensors": { "model.layers.21.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.21.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.21.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.21.mlp.down_proj": { "stored_tensors": { "model.layers.21.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.21.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.21.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.22.input_layernorm": { "stored_tensors": { "model.layers.22.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.22.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.22.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.22.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.22.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.22.linear_attn.g_proj": { "stored_tensors": { "model.layers.22.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.22.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.22.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.22.linear_attn.b_proj": { "stored_tensors": { "model.layers.22.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.22.linear_attn.a_proj": { "stored_tensors": { "model.layers.22.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.22.linear_attn.o_proj": { "stored_tensors": { "model.layers.22.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.22.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.22.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.22.linear_attn.o_norm": { "stored_tensors": { "model.layers.22.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.22.post_attention_layernorm": { "stored_tensors": { "model.layers.22.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.22.mlp.up_proj": { "stored_tensors": { "model.layers.22.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.22.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.22.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.22.mlp.gate_proj": { "stored_tensors": { "model.layers.22.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.22.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.22.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.22.mlp.down_proj": { "stored_tensors": { "model.layers.22.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.22.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.22.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.23.self_attn.q_proj": { "stored_tensors": { "model.layers.23.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.23.self_attn.k_proj": { "stored_tensors": { "model.layers.23.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.23.self_attn.v_proj": { "stored_tensors": { "model.layers.23.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.23.self_attn.o_proj": { "stored_tensors": { "model.layers.23.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.23.self_attn.q_norm": { "stored_tensors": { "model.layers.23.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.23.self_attn.k_norm": { "stored_tensors": { "model.layers.23.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.23.post_attention_layernorm": { "stored_tensors": { "model.layers.23.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.23.mlp.up_proj": { "stored_tensors": { "model.layers.23.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.23.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.23.mlp.gate_proj": { "stored_tensors": { "model.layers.23.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.23.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.23.mlp.down_proj": { "stored_tensors": { "model.layers.23.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.23.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.23.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.23.post_feedforward_layernorm": { "stored_tensors": { "model.layers.23.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.24.input_layernorm": { "stored_tensors": { "model.layers.24.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.24.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.24.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.24.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.24.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.24.linear_attn.g_proj": { "stored_tensors": { "model.layers.24.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.24.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.24.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.24.linear_attn.b_proj": { "stored_tensors": { "model.layers.24.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.24.linear_attn.a_proj": { "stored_tensors": { "model.layers.24.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.24.linear_attn.o_proj": { "stored_tensors": { "model.layers.24.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.24.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.24.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.24.linear_attn.o_norm": { "stored_tensors": { "model.layers.24.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.24.post_attention_layernorm": { "stored_tensors": { "model.layers.24.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.24.mlp.up_proj": { "stored_tensors": { "model.layers.24.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.24.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.24.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.24.mlp.gate_proj": { "stored_tensors": { "model.layers.24.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.24.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.24.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.24.mlp.down_proj": { "stored_tensors": { "model.layers.24.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.24.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.24.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.25.input_layernorm": { "stored_tensors": { "model.layers.25.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.25.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.25.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.25.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.25.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.25.linear_attn.g_proj": { "stored_tensors": { "model.layers.25.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.25.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.25.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.25.linear_attn.b_proj": { "stored_tensors": { "model.layers.25.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.25.linear_attn.a_proj": { "stored_tensors": { "model.layers.25.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.25.linear_attn.o_proj": { "stored_tensors": { "model.layers.25.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.25.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.25.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.25.linear_attn.o_norm": { "stored_tensors": { "model.layers.25.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.25.post_attention_layernorm": { "stored_tensors": { "model.layers.25.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.25.mlp.up_proj": { "stored_tensors": { "model.layers.25.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.25.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.25.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.25.mlp.gate_proj": { "stored_tensors": { "model.layers.25.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.25.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.25.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.25.mlp.down_proj": { "stored_tensors": { "model.layers.25.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.25.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.25.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.26.input_layernorm": { "stored_tensors": { "model.layers.26.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.26.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.26.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.26.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.26.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.26.linear_attn.g_proj": { "stored_tensors": { "model.layers.26.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.26.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.26.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.26.linear_attn.b_proj": { "stored_tensors": { "model.layers.26.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.26.linear_attn.a_proj": { "stored_tensors": { "model.layers.26.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.26.linear_attn.o_proj": { "stored_tensors": { "model.layers.26.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.26.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.26.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.26.linear_attn.o_norm": { "stored_tensors": { "model.layers.26.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.26.post_attention_layernorm": { "stored_tensors": { "model.layers.26.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.26.mlp.up_proj": { "stored_tensors": { "model.layers.26.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.26.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.26.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.26.mlp.gate_proj": { "stored_tensors": { "model.layers.26.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.26.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.26.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.26.mlp.down_proj": { "stored_tensors": { "model.layers.26.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.26.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.26.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.27.self_attn.q_proj": { "stored_tensors": { "model.layers.27.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.27.self_attn.k_proj": { "stored_tensors": { "model.layers.27.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.27.self_attn.v_proj": { "stored_tensors": { "model.layers.27.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.27.self_attn.o_proj": { "stored_tensors": { "model.layers.27.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.27.self_attn.q_norm": { "stored_tensors": { "model.layers.27.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.27.self_attn.k_norm": { "stored_tensors": { "model.layers.27.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.27.post_attention_layernorm": { "stored_tensors": { "model.layers.27.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.27.mlp.up_proj": { "stored_tensors": { "model.layers.27.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.27.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.27.mlp.gate_proj": { "stored_tensors": { "model.layers.27.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.27.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.27.mlp.down_proj": { "stored_tensors": { "model.layers.27.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.27.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.27.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.27.post_feedforward_layernorm": { "stored_tensors": { "model.layers.27.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.28.input_layernorm": { "stored_tensors": { "model.layers.28.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.28.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.28.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.28.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.28.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.28.linear_attn.g_proj": { "stored_tensors": { "model.layers.28.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.28.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.28.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.28.linear_attn.b_proj": { "stored_tensors": { "model.layers.28.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.28.linear_attn.a_proj": { "stored_tensors": { "model.layers.28.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.28.linear_attn.o_proj": { "stored_tensors": { "model.layers.28.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.28.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.28.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.28.linear_attn.o_norm": { "stored_tensors": { "model.layers.28.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.28.post_attention_layernorm": { "stored_tensors": { "model.layers.28.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.28.mlp.up_proj": { "stored_tensors": { "model.layers.28.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.28.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.28.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.28.mlp.gate_proj": { "stored_tensors": { "model.layers.28.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.28.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.28.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.28.mlp.down_proj": { "stored_tensors": { "model.layers.28.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.28.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.28.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.29.input_layernorm": { "stored_tensors": { "model.layers.29.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.29.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.29.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.29.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.29.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.29.linear_attn.g_proj": { "stored_tensors": { "model.layers.29.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.29.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.29.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 80 ], "n_bytes": 13824000, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mcg_multiplier": 3417055213 }, "model.layers.29.linear_attn.b_proj": { "stored_tensors": { "model.layers.29.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.29.linear_attn.a_proj": { "stored_tensors": { "model.layers.29.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.29.linear_attn.o_proj": { "stored_tensors": { "model.layers.29.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.29.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.29.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 48 ], "n_bytes": 8294400, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.29.linear_attn.o_norm": { "stored_tensors": { "model.layers.29.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.29.post_attention_layernorm": { "stored_tensors": { "model.layers.29.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.29.mlp.up_proj": { "stored_tensors": { "model.layers.29.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.29.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.29.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.29.mlp.gate_proj": { "stored_tensors": { "model.layers.29.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.29.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.29.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.29.mlp.down_proj": { "stored_tensors": { "model.layers.29.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.29.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.29.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.30.input_layernorm": { "stored_tensors": { "model.layers.30.input_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.30.linear_attn.in_proj_qkv": { "stored_tensors": { "model.layers.30.linear_attn.in_proj_qkv.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.30.linear_attn.in_proj_qkv.svh": { "shape": [ 11520 ], "n_bytes": 23040, "dtype": "torch.float16" }, "model.layers.30.linear_attn.in_proj_qkv.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.linear_attn.in_proj_qkv.trellis": { "shape": [ 240, 720, 48 ], "n_bytes": 16588800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.30.linear_attn.g_proj": { "stored_tensors": { "model.layers.30.linear_attn.g_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.30.linear_attn.g_proj.svh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.30.linear_attn.g_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.linear_attn.g_proj.trellis": { "shape": [ 240, 360, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.30.linear_attn.b_proj": { "stored_tensors": { "model.layers.30.linear_attn.b_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.30.linear_attn.a_proj": { "stored_tensors": { "model.layers.30.linear_attn.a_proj.weight": { "shape": [ 30, 3840 ], "n_bytes": 230400, "dtype": "torch.float16" } } }, "model.layers.30.linear_attn.o_proj": { "stored_tensors": { "model.layers.30.linear_attn.o_proj.suh": { "shape": [ 5760 ], "n_bytes": 11520, "dtype": "torch.float16" }, "model.layers.30.linear_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.30.linear_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.linear_attn.o_proj.trellis": { "shape": [ 360, 240, 64 ], "n_bytes": 11059200, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.30.linear_attn.o_norm": { "stored_tensors": { "model.layers.30.linear_attn.o_norm.weight": { "shape": [ 192 ], "n_bytes": 384, "dtype": "torch.bfloat16" } } }, "model.layers.30.post_attention_layernorm": { "stored_tensors": { "model.layers.30.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.30.mlp.up_proj": { "stored_tensors": { "model.layers.30.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.30.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.30.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.mlp.up_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.30.mlp.gate_proj": { "stored_tensors": { "model.layers.30.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.30.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.30.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.30.mlp.down_proj": { "stored_tensors": { "model.layers.30.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.30.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.30.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.31.self_attn.q_proj": { "stored_tensors": { "model.layers.31.self_attn.q_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.q_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.q_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.q_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.31.self_attn.k_proj": { "stored_tensors": { "model.layers.31.self_attn.k_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.k_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.k_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.k_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.31.self_attn.v_proj": { "stored_tensors": { "model.layers.31.self_attn.v_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.v_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.v_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.v_proj.trellis": { "shape": [ 240, 240, 64 ], "n_bytes": 7372800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.31.self_attn.o_proj": { "stored_tensors": { "model.layers.31.self_attn.o_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.o_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.self_attn.o_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.o_proj.trellis": { "shape": [ 240, 240, 48 ], "n_bytes": 5529600, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.31.self_attn.q_norm": { "stored_tensors": { "model.layers.31.self_attn.q_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.31.self_attn.k_norm": { "stored_tensors": { "model.layers.31.self_attn.k_norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.31.post_attention_layernorm": { "stored_tensors": { "model.layers.31.post_attention_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.layers.31.mlp.up_proj": { "stored_tensors": { "model.layers.31.mlp.up_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.mlp.up_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.31.mlp.up_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.mlp.up_proj.trellis": { "shape": [ 240, 688, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.31.mlp.gate_proj": { "stored_tensors": { "model.layers.31.mlp.gate_proj.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.mlp.gate_proj.svh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.31.mlp.gate_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.mlp.gate_proj.trellis": { "shape": [ 240, 688, 48 ], "n_bytes": 15851520, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mcg_multiplier": 3417055213 }, "model.layers.31.mlp.down_proj": { "stored_tensors": { "model.layers.31.mlp.down_proj.suh": { "shape": [ 11008 ], "n_bytes": 22016, "dtype": "torch.float16" }, "model.layers.31.mlp.down_proj.svh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "model.layers.31.mlp.down_proj.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.mlp.down_proj.trellis": { "shape": [ 688, 240, 64 ], "n_bytes": 21135360, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mcg_multiplier": 3417055213 }, "model.layers.31.post_feedforward_layernorm": { "stored_tensors": { "model.layers.31.post_feedforward_layernorm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "model.norm": { "stored_tensors": { "model.norm.weight": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.bfloat16" } } }, "lm_head": { "stored_tensors": { "lm_head.suh": { "shape": [ 3840 ], "n_bytes": 7680, "dtype": "torch.float16" }, "lm_head.svh": { "shape": [ 100352 ], "n_bytes": 200704, "dtype": "torch.float16" }, "lm_head.mcg": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "lm_head.trellis": { "shape": [ 240, 6272, 96 ], "n_bytes": 289013760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mcg_multiplier": 3417055213 } } }