diff --git "a/quantization_config.json" "b/quantization_config.json" new file mode 100644--- /dev/null +++ "b/quantization_config.json" @@ -0,0 +1,5967 @@ +{ + "quant_method": "exl3", + "version": "0.0.1", + "bits": 4.0, + "calibration": { + "rows": 100, + "cols": 2048 + }, + "tensor_storage": { + "model.embed_tokens": { + "stored_tensors": { + "model.embed_tokens.weight": { + "shape": [ + 151936, + 896 + ], + "n_bytes": 272269312, + "dtype": "torch.float16" + } + } + }, + "model.layers.0.input_layernorm": { + "stored_tensors": { + "model.layers.0.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.0.self_attn.q_proj": { + "stored_tensors": { + "model.layers.0.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.0.self_attn.k_proj": { + "stored_tensors": { + "model.layers.0.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.0.self_attn.v_proj": { + "stored_tensors": { + "model.layers.0.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.0.self_attn.o_proj": { + "stored_tensors": { + "model.layers.0.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.0.post_attention_layernorm": { + "stored_tensors": { + "model.layers.0.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.0.mlp.up_proj": { + "stored_tensors": { + "model.layers.0.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.0.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.0.mlp.gate_proj": { + "stored_tensors": { + "model.layers.0.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.0.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.0.mlp.down_proj": { + "stored_tensors": { + "model.layers.0.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.0.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.0.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.input_layernorm": { + "stored_tensors": { + "model.layers.1.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.1.self_attn.q_proj": { + "stored_tensors": { + "model.layers.1.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.self_attn.k_proj": { + "stored_tensors": { + "model.layers.1.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.self_attn.v_proj": { + "stored_tensors": { + "model.layers.1.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.self_attn.o_proj": { + "stored_tensors": { + "model.layers.1.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.post_attention_layernorm": { + "stored_tensors": { + "model.layers.1.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.1.mlp.up_proj": { + "stored_tensors": { + "model.layers.1.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.1.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.mlp.gate_proj": { + "stored_tensors": { + "model.layers.1.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.1.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.1.mlp.down_proj": { + "stored_tensors": { + "model.layers.1.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.1.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.1.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.input_layernorm": { + "stored_tensors": { + "model.layers.2.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.2.self_attn.q_proj": { + "stored_tensors": { + "model.layers.2.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.self_attn.k_proj": { + "stored_tensors": { + "model.layers.2.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.self_attn.v_proj": { + "stored_tensors": { + "model.layers.2.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.self_attn.o_proj": { + "stored_tensors": { + "model.layers.2.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.post_attention_layernorm": { + "stored_tensors": { + "model.layers.2.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.2.mlp.up_proj": { + "stored_tensors": { + "model.layers.2.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.2.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.mlp.gate_proj": { + "stored_tensors": { + "model.layers.2.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.2.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.2.mlp.down_proj": { + "stored_tensors": { + "model.layers.2.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.2.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.2.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.input_layernorm": { + "stored_tensors": { + "model.layers.3.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.3.self_attn.q_proj": { + "stored_tensors": { + "model.layers.3.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.self_attn.k_proj": { + "stored_tensors": { + "model.layers.3.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.self_attn.v_proj": { + "stored_tensors": { + "model.layers.3.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.self_attn.o_proj": { + "stored_tensors": { + "model.layers.3.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.post_attention_layernorm": { + "stored_tensors": { + "model.layers.3.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.3.mlp.up_proj": { + "stored_tensors": { + "model.layers.3.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.3.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.mlp.gate_proj": { + "stored_tensors": { + "model.layers.3.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.3.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.3.mlp.down_proj": { + "stored_tensors": { + "model.layers.3.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.3.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.3.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.input_layernorm": { + "stored_tensors": { + "model.layers.4.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.4.self_attn.q_proj": { + "stored_tensors": { + "model.layers.4.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.self_attn.k_proj": { + "stored_tensors": { + "model.layers.4.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.self_attn.v_proj": { + "stored_tensors": { + "model.layers.4.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.self_attn.o_proj": { + "stored_tensors": { + "model.layers.4.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.post_attention_layernorm": { + "stored_tensors": { + "model.layers.4.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.4.mlp.up_proj": { + "stored_tensors": { + "model.layers.4.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.4.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.mlp.gate_proj": { + "stored_tensors": { + "model.layers.4.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.4.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.4.mlp.down_proj": { + "stored_tensors": { + "model.layers.4.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.4.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.4.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.input_layernorm": { + "stored_tensors": { + "model.layers.5.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.5.self_attn.q_proj": { + "stored_tensors": { + "model.layers.5.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.self_attn.k_proj": { + "stored_tensors": { + "model.layers.5.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.self_attn.v_proj": { + "stored_tensors": { + "model.layers.5.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.self_attn.o_proj": { + "stored_tensors": { + "model.layers.5.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.post_attention_layernorm": { + "stored_tensors": { + "model.layers.5.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.5.mlp.up_proj": { + "stored_tensors": { + "model.layers.5.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.5.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.mlp.gate_proj": { + "stored_tensors": { + "model.layers.5.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.5.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.5.mlp.down_proj": { + "stored_tensors": { + "model.layers.5.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.5.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.5.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.input_layernorm": { + "stored_tensors": { + "model.layers.6.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.6.self_attn.q_proj": { + "stored_tensors": { + "model.layers.6.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.self_attn.k_proj": { + "stored_tensors": { + "model.layers.6.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.self_attn.v_proj": { + "stored_tensors": { + "model.layers.6.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.self_attn.o_proj": { + "stored_tensors": { + "model.layers.6.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.post_attention_layernorm": { + "stored_tensors": { + "model.layers.6.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.6.mlp.up_proj": { + "stored_tensors": { + "model.layers.6.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.6.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.mlp.gate_proj": { + "stored_tensors": { + "model.layers.6.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.6.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.6.mlp.down_proj": { + "stored_tensors": { + "model.layers.6.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.6.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.6.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.input_layernorm": { + "stored_tensors": { + "model.layers.7.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.7.self_attn.q_proj": { + "stored_tensors": { + "model.layers.7.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.self_attn.k_proj": { + "stored_tensors": { + "model.layers.7.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.self_attn.v_proj": { + "stored_tensors": { + "model.layers.7.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.self_attn.o_proj": { + "stored_tensors": { + "model.layers.7.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.post_attention_layernorm": { + "stored_tensors": { + "model.layers.7.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.7.mlp.up_proj": { + "stored_tensors": { + "model.layers.7.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.7.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.mlp.gate_proj": { + "stored_tensors": { + "model.layers.7.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.7.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.7.mlp.down_proj": { + "stored_tensors": { + "model.layers.7.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.7.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.7.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.input_layernorm": { + "stored_tensors": { + "model.layers.8.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.8.self_attn.q_proj": { + "stored_tensors": { + "model.layers.8.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.self_attn.k_proj": { + "stored_tensors": { + "model.layers.8.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.self_attn.v_proj": { + "stored_tensors": { + "model.layers.8.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.self_attn.o_proj": { + "stored_tensors": { + "model.layers.8.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.post_attention_layernorm": { + "stored_tensors": { + "model.layers.8.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.8.mlp.up_proj": { + "stored_tensors": { + "model.layers.8.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.8.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.mlp.gate_proj": { + "stored_tensors": { + "model.layers.8.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.8.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.8.mlp.down_proj": { + "stored_tensors": { + "model.layers.8.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.8.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.8.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.input_layernorm": { + "stored_tensors": { + "model.layers.9.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.9.self_attn.q_proj": { + "stored_tensors": { + "model.layers.9.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.self_attn.k_proj": { + "stored_tensors": { + "model.layers.9.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.self_attn.v_proj": { + "stored_tensors": { + "model.layers.9.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.self_attn.o_proj": { + "stored_tensors": { + "model.layers.9.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.post_attention_layernorm": { + "stored_tensors": { + "model.layers.9.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.9.mlp.up_proj": { + "stored_tensors": { + "model.layers.9.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.9.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.mlp.gate_proj": { + "stored_tensors": { + "model.layers.9.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.9.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.9.mlp.down_proj": { + "stored_tensors": { + "model.layers.9.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.9.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.9.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.input_layernorm": { + "stored_tensors": { + "model.layers.10.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.10.self_attn.q_proj": { + "stored_tensors": { + "model.layers.10.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.self_attn.k_proj": { + "stored_tensors": { + "model.layers.10.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.self_attn.v_proj": { + "stored_tensors": { + "model.layers.10.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.self_attn.o_proj": { + "stored_tensors": { + "model.layers.10.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.post_attention_layernorm": { + "stored_tensors": { + "model.layers.10.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.10.mlp.up_proj": { + "stored_tensors": { + "model.layers.10.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.10.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.mlp.gate_proj": { + "stored_tensors": { + "model.layers.10.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.10.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.10.mlp.down_proj": { + "stored_tensors": { + "model.layers.10.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.10.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.10.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.input_layernorm": { + "stored_tensors": { + "model.layers.11.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.11.self_attn.q_proj": { + "stored_tensors": { + "model.layers.11.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.self_attn.k_proj": { + "stored_tensors": { + "model.layers.11.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.self_attn.v_proj": { + "stored_tensors": { + "model.layers.11.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.self_attn.o_proj": { + "stored_tensors": { + "model.layers.11.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.post_attention_layernorm": { + "stored_tensors": { + "model.layers.11.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.11.mlp.up_proj": { + "stored_tensors": { + "model.layers.11.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.11.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.mlp.gate_proj": { + "stored_tensors": { + "model.layers.11.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.11.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.11.mlp.down_proj": { + "stored_tensors": { + "model.layers.11.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.11.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.11.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.input_layernorm": { + "stored_tensors": { + "model.layers.12.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.12.self_attn.q_proj": { + "stored_tensors": { + "model.layers.12.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.self_attn.k_proj": { + "stored_tensors": { + "model.layers.12.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.self_attn.v_proj": { + "stored_tensors": { + "model.layers.12.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.self_attn.o_proj": { + "stored_tensors": { + "model.layers.12.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.post_attention_layernorm": { + "stored_tensors": { + "model.layers.12.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.12.mlp.up_proj": { + "stored_tensors": { + "model.layers.12.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.12.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.mlp.gate_proj": { + "stored_tensors": { + "model.layers.12.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.12.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.12.mlp.down_proj": { + "stored_tensors": { + "model.layers.12.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.12.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.12.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.input_layernorm": { + "stored_tensors": { + "model.layers.13.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.13.self_attn.q_proj": { + "stored_tensors": { + "model.layers.13.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.self_attn.k_proj": { + "stored_tensors": { + "model.layers.13.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.self_attn.v_proj": { + "stored_tensors": { + "model.layers.13.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.self_attn.o_proj": { + "stored_tensors": { + "model.layers.13.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.post_attention_layernorm": { + "stored_tensors": { + "model.layers.13.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.13.mlp.up_proj": { + "stored_tensors": { + "model.layers.13.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.13.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.mlp.gate_proj": { + "stored_tensors": { + "model.layers.13.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.13.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.13.mlp.down_proj": { + "stored_tensors": { + "model.layers.13.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.13.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.13.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.input_layernorm": { + "stored_tensors": { + "model.layers.14.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.14.self_attn.q_proj": { + "stored_tensors": { + "model.layers.14.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.self_attn.k_proj": { + "stored_tensors": { + "model.layers.14.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.self_attn.v_proj": { + "stored_tensors": { + "model.layers.14.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.self_attn.o_proj": { + "stored_tensors": { + "model.layers.14.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.post_attention_layernorm": { + "stored_tensors": { + "model.layers.14.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.14.mlp.up_proj": { + "stored_tensors": { + "model.layers.14.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.14.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.mlp.gate_proj": { + "stored_tensors": { + "model.layers.14.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.14.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.14.mlp.down_proj": { + "stored_tensors": { + "model.layers.14.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.14.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.14.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.input_layernorm": { + "stored_tensors": { + "model.layers.15.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.15.self_attn.q_proj": { + "stored_tensors": { + "model.layers.15.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.self_attn.k_proj": { + "stored_tensors": { + "model.layers.15.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.self_attn.v_proj": { + "stored_tensors": { + "model.layers.15.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.self_attn.o_proj": { + "stored_tensors": { + "model.layers.15.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.post_attention_layernorm": { + "stored_tensors": { + "model.layers.15.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.15.mlp.up_proj": { + "stored_tensors": { + "model.layers.15.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.15.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.mlp.gate_proj": { + "stored_tensors": { + "model.layers.15.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.15.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.15.mlp.down_proj": { + "stored_tensors": { + "model.layers.15.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.15.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.15.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.input_layernorm": { + "stored_tensors": { + "model.layers.16.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.16.self_attn.q_proj": { + "stored_tensors": { + "model.layers.16.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.self_attn.k_proj": { + "stored_tensors": { + "model.layers.16.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.self_attn.v_proj": { + "stored_tensors": { + "model.layers.16.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.self_attn.o_proj": { + "stored_tensors": { + "model.layers.16.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.post_attention_layernorm": { + "stored_tensors": { + "model.layers.16.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.16.mlp.up_proj": { + "stored_tensors": { + "model.layers.16.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.16.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.mlp.gate_proj": { + "stored_tensors": { + "model.layers.16.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.16.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.16.mlp.down_proj": { + "stored_tensors": { + "model.layers.16.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.16.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.16.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.input_layernorm": { + "stored_tensors": { + "model.layers.17.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.17.self_attn.q_proj": { + "stored_tensors": { + "model.layers.17.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.self_attn.k_proj": { + "stored_tensors": { + "model.layers.17.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.self_attn.v_proj": { + "stored_tensors": { + "model.layers.17.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.self_attn.o_proj": { + "stored_tensors": { + "model.layers.17.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.post_attention_layernorm": { + "stored_tensors": { + "model.layers.17.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.17.mlp.up_proj": { + "stored_tensors": { + "model.layers.17.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.17.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.mlp.gate_proj": { + "stored_tensors": { + "model.layers.17.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.17.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.17.mlp.down_proj": { + "stored_tensors": { + "model.layers.17.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.17.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.17.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.input_layernorm": { + "stored_tensors": { + "model.layers.18.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.18.self_attn.q_proj": { + "stored_tensors": { + "model.layers.18.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.self_attn.k_proj": { + "stored_tensors": { + "model.layers.18.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.self_attn.v_proj": { + "stored_tensors": { + "model.layers.18.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.self_attn.o_proj": { + "stored_tensors": { + "model.layers.18.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.post_attention_layernorm": { + "stored_tensors": { + "model.layers.18.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.18.mlp.up_proj": { + "stored_tensors": { + "model.layers.18.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.18.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.mlp.gate_proj": { + "stored_tensors": { + "model.layers.18.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.18.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.18.mlp.down_proj": { + "stored_tensors": { + "model.layers.18.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.18.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.18.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.input_layernorm": { + "stored_tensors": { + "model.layers.19.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.19.self_attn.q_proj": { + "stored_tensors": { + "model.layers.19.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.self_attn.k_proj": { + "stored_tensors": { + "model.layers.19.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.self_attn.v_proj": { + "stored_tensors": { + "model.layers.19.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.self_attn.o_proj": { + "stored_tensors": { + "model.layers.19.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.post_attention_layernorm": { + "stored_tensors": { + "model.layers.19.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.19.mlp.up_proj": { + "stored_tensors": { + "model.layers.19.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.19.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.mlp.gate_proj": { + "stored_tensors": { + "model.layers.19.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.19.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.19.mlp.down_proj": { + "stored_tensors": { + "model.layers.19.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.19.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.19.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.input_layernorm": { + "stored_tensors": { + "model.layers.20.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.20.self_attn.q_proj": { + "stored_tensors": { + "model.layers.20.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.self_attn.k_proj": { + "stored_tensors": { + "model.layers.20.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.self_attn.v_proj": { + "stored_tensors": { + "model.layers.20.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.self_attn.o_proj": { + "stored_tensors": { + "model.layers.20.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.post_attention_layernorm": { + "stored_tensors": { + "model.layers.20.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.20.mlp.up_proj": { + "stored_tensors": { + "model.layers.20.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.20.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.mlp.gate_proj": { + "stored_tensors": { + "model.layers.20.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.20.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.20.mlp.down_proj": { + "stored_tensors": { + "model.layers.20.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.20.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.20.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.input_layernorm": { + "stored_tensors": { + "model.layers.21.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.21.self_attn.q_proj": { + "stored_tensors": { + "model.layers.21.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.self_attn.k_proj": { + "stored_tensors": { + "model.layers.21.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.self_attn.v_proj": { + "stored_tensors": { + "model.layers.21.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.self_attn.o_proj": { + "stored_tensors": { + "model.layers.21.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.post_attention_layernorm": { + "stored_tensors": { + "model.layers.21.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.21.mlp.up_proj": { + "stored_tensors": { + "model.layers.21.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.21.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.mlp.gate_proj": { + "stored_tensors": { + "model.layers.21.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.21.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.21.mlp.down_proj": { + "stored_tensors": { + "model.layers.21.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.21.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.21.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.input_layernorm": { + "stored_tensors": { + "model.layers.22.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.22.self_attn.q_proj": { + "stored_tensors": { + "model.layers.22.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.self_attn.k_proj": { + "stored_tensors": { + "model.layers.22.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.self_attn.v_proj": { + "stored_tensors": { + "model.layers.22.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.self_attn.o_proj": { + "stored_tensors": { + "model.layers.22.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.post_attention_layernorm": { + "stored_tensors": { + "model.layers.22.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.22.mlp.up_proj": { + "stored_tensors": { + "model.layers.22.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.22.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.mlp.gate_proj": { + "stored_tensors": { + "model.layers.22.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.22.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.22.mlp.down_proj": { + "stored_tensors": { + "model.layers.22.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.22.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.22.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.input_layernorm": { + "stored_tensors": { + "model.layers.23.input_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.23.self_attn.q_proj": { + "stored_tensors": { + "model.layers.23.self_attn.q_proj.bias": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.q_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.q_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.q_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.self_attn.k_proj": { + "stored_tensors": { + "model.layers.23.self_attn.k_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.k_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.k_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.k_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.self_attn.v_proj": { + "stored_tensors": { + "model.layers.23.self_attn.v_proj.bias": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.v_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.v_proj.svh": { + "shape": [ + 128 + ], + "n_bytes": 256, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.v_proj.trellis": { + "shape": [ + 56, + 8, + 64 + ], + "n_bytes": 57344, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.self_attn.o_proj": { + "stored_tensors": { + "model.layers.23.self_attn.o_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.o_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.self_attn.o_proj.trellis": { + "shape": [ + 56, + 56, + 64 + ], + "n_bytes": 401408, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.post_attention_layernorm": { + "stored_tensors": { + "model.layers.23.post_attention_layernorm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "model.layers.23.mlp.up_proj": { + "stored_tensors": { + "model.layers.23.mlp.up_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.mlp.up_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.23.mlp.up_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.mlp.gate_proj": { + "stored_tensors": { + "model.layers.23.mlp.gate_proj.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.mlp.gate_proj.svh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.23.mlp.gate_proj.trellis": { + "shape": [ + 56, + 304, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.layers.23.mlp.down_proj": { + "stored_tensors": { + "model.layers.23.mlp.down_proj.suh": { + "shape": [ + 4864 + ], + "n_bytes": 9728, + "dtype": "torch.float16" + }, + "model.layers.23.mlp.down_proj.svh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "model.layers.23.mlp.down_proj.trellis": { + "shape": [ + 304, + 56, + 64 + ], + "n_bytes": 2179072, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 4 + }, + "model.norm": { + "stored_tensors": { + "model.norm.weight": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + } + } + }, + "lm_head": { + "stored_tensors": { + "lm_head.suh": { + "shape": [ + 896 + ], + "n_bytes": 1792, + "dtype": "torch.float16" + }, + "lm_head.svh": { + "shape": [ + 151936 + ], + "n_bytes": 303872, + "dtype": "torch.float16" + }, + "lm_head.trellis": { + "shape": [ + 56, + 9496, + 96 + ], + "n_bytes": 102100992, + "dtype": "torch.int16" + } + }, + "quant_format": "exl3", + "bits_per_weight": 6 + } + } +} \ No newline at end of file