Hugging Face's logo

JRamirez-UAB
/
hybridagent-8b-exl3-vllm-experimental

Text Generation
Transformers
Safetensors
English
qwen3
hybrid-reasoning
latent-reasoning
agent
tool-use
exl3
vllm
text-generation-inference
quantization
experimental
rtx3090
ampere
conversational
5-bit
Model card Files
xet
 Community
hybridagent-8b-exl3-vllm-experimental / quantization_config.json
JRamirez-UAB's picture
JRamirez-UAB
Upload quantization_config.json with huggingface_hub
0f896bf verified
Raw History Blame Contribute Delete
350 kB
{ "quant_method": "exl3", "version": "1.5.1", "bits": 5.0, "head_bits": 5, "calibration": { "rows": 196, "cols": 2048 }, "out_scales": "always", "codebook": "mul1", "tensor_storage": { "model.embed_tokens": { "stored_tensors": { "model.embed_tokens.weight": { "shape": [ 151671, 4096 ], "n_bytes": 1242488832, "dtype": "torch.bfloat16" } } }, "model.layers.0.input_layernorm": { "stored_tensors": { "model.layers.0.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.0.self_attn.q_proj": { "stored_tensors": { "model.layers.0.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.0.self_attn.k_proj": { "stored_tensors": { "model.layers.0.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.0.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.0.self_attn.v_proj": { "stored_tensors": { "model.layers.0.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.0.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.0.self_attn.o_proj": { "stored_tensors": { "model.layers.0.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.0.self_attn.q_norm": { "stored_tensors": { "model.layers.0.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.0.self_attn.k_norm": { "stored_tensors": { "model.layers.0.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.0.post_attention_layernorm": { "stored_tensors": { "model.layers.0.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.0.mlp.up_proj": { "stored_tensors": { "model.layers.0.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.0.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.mlp.up_proj.trellis": { "shape": [ 256, 768, 48 ], "n_bytes": 18874368, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.0.mlp.gate_proj": { "stored_tensors": { "model.layers.0.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.0.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.mlp.gate_proj.trellis": { "shape": [ 256, 768, 48 ], "n_bytes": 18874368, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.0.mlp.down_proj": { "stored_tensors": { "model.layers.0.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.0.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.0.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.0.mlp.down_proj.trellis": { "shape": [ 768, 256, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.1.input_layernorm": { "stored_tensors": { "model.layers.1.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.1.self_attn.q_proj": { "stored_tensors": { "model.layers.1.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.self_attn.q_proj.trellis": { "shape": [ 256, 256, 48 ], "n_bytes": 6291456, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.1.self_attn.k_proj": { "stored_tensors": { "model.layers.1.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.1.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.self_attn.k_proj.trellis": { "shape": [ 256, 64, 48 ], "n_bytes": 1572864, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.1.self_attn.v_proj": { "stored_tensors": { "model.layers.1.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.1.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.self_attn.v_proj.trellis": { "shape": [ 256, 64, 48 ], "n_bytes": 1572864, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.1.self_attn.o_proj": { "stored_tensors": { "model.layers.1.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.self_attn.o_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.1.self_attn.q_norm": { "stored_tensors": { "model.layers.1.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.1.self_attn.k_norm": { "stored_tensors": { "model.layers.1.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.1.post_attention_layernorm": { "stored_tensors": { "model.layers.1.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.1.mlp.up_proj": { "stored_tensors": { "model.layers.1.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.1.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.mlp.up_proj.trellis": { "shape": [ 256, 768, 48 ], "n_bytes": 18874368, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.1.mlp.gate_proj": { "stored_tensors": { "model.layers.1.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.1.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.mlp.gate_proj.trellis": { "shape": [ 256, 768, 48 ], "n_bytes": 18874368, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.1.mlp.down_proj": { "stored_tensors": { "model.layers.1.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.1.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.1.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.1.mlp.down_proj.trellis": { "shape": [ 768, 256, 48 ], "n_bytes": 18874368, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.2.input_layernorm": { "stored_tensors": { "model.layers.2.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.2.self_attn.q_proj": { "stored_tensors": { "model.layers.2.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.self_attn.q_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.2.self_attn.k_proj": { "stored_tensors": { "model.layers.2.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.2.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.self_attn.k_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.2.self_attn.v_proj": { "stored_tensors": { "model.layers.2.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.2.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.self_attn.v_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.2.self_attn.o_proj": { "stored_tensors": { "model.layers.2.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.self_attn.o_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.2.self_attn.q_norm": { "stored_tensors": { "model.layers.2.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.2.self_attn.k_norm": { "stored_tensors": { "model.layers.2.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.2.post_attention_layernorm": { "stored_tensors": { "model.layers.2.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.2.mlp.up_proj": { "stored_tensors": { "model.layers.2.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.2.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.mlp.up_proj.trellis": { "shape": [ 256, 768, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.2.mlp.gate_proj": { "stored_tensors": { "model.layers.2.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.2.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.mlp.gate_proj.trellis": { "shape": [ 256, 768, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.2.mlp.down_proj": { "stored_tensors": { "model.layers.2.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.2.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.2.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.2.mlp.down_proj.trellis": { "shape": [ 768, 256, 48 ], "n_bytes": 18874368, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 3, "mul1_multiplier": 2212286765 }, "model.layers.3.input_layernorm": { "stored_tensors": { "model.layers.3.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.3.self_attn.q_proj": { "stored_tensors": { "model.layers.3.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.q_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.3.self_attn.k_proj": { "stored_tensors": { "model.layers.3.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.3.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.k_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.3.self_attn.v_proj": { "stored_tensors": { "model.layers.3.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.3.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.v_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.3.self_attn.o_proj": { "stored_tensors": { "model.layers.3.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.self_attn.o_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.3.self_attn.q_norm": { "stored_tensors": { "model.layers.3.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.3.self_attn.k_norm": { "stored_tensors": { "model.layers.3.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.3.post_attention_layernorm": { "stored_tensors": { "model.layers.3.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.3.mlp.up_proj": { "stored_tensors": { "model.layers.3.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.3.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.mlp.up_proj.trellis": { "shape": [ 256, 768, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.3.mlp.gate_proj": { "stored_tensors": { "model.layers.3.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.3.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.mlp.gate_proj.trellis": { "shape": [ 256, 768, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.3.mlp.down_proj": { "stored_tensors": { "model.layers.3.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.3.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.3.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.3.mlp.down_proj.trellis": { "shape": [ 768, 256, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.input_layernorm": { "stored_tensors": { "model.layers.4.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.4.self_attn.q_proj": { "stored_tensors": { "model.layers.4.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.self_attn.q_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.self_attn.k_proj": { "stored_tensors": { "model.layers.4.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.4.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.self_attn.k_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.self_attn.v_proj": { "stored_tensors": { "model.layers.4.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.4.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.self_attn.v_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.self_attn.o_proj": { "stored_tensors": { "model.layers.4.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.self_attn.o_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.self_attn.q_norm": { "stored_tensors": { "model.layers.4.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.4.self_attn.k_norm": { "stored_tensors": { "model.layers.4.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.4.post_attention_layernorm": { "stored_tensors": { "model.layers.4.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.4.mlp.up_proj": { "stored_tensors": { "model.layers.4.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.4.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.mlp.up_proj.trellis": { "shape": [ 256, 768, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.mlp.gate_proj": { "stored_tensors": { "model.layers.4.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.4.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.mlp.gate_proj.trellis": { "shape": [ 256, 768, 64 ], "n_bytes": 25165824, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.4.mlp.down_proj": { "stored_tensors": { "model.layers.4.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.4.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.4.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.4.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.5.input_layernorm": { "stored_tensors": { "model.layers.5.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.5.self_attn.q_proj": { "stored_tensors": { "model.layers.5.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.self_attn.q_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.5.self_attn.k_proj": { "stored_tensors": { "model.layers.5.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.5.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.self_attn.k_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.5.self_attn.v_proj": { "stored_tensors": { "model.layers.5.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.5.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.self_attn.v_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.5.self_attn.o_proj": { "stored_tensors": { "model.layers.5.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.self_attn.o_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.5.self_attn.q_norm": { "stored_tensors": { "model.layers.5.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.5.self_attn.k_norm": { "stored_tensors": { "model.layers.5.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.5.post_attention_layernorm": { "stored_tensors": { "model.layers.5.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.5.mlp.up_proj": { "stored_tensors": { "model.layers.5.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.5.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.5.mlp.gate_proj": { "stored_tensors": { "model.layers.5.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.5.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.5.mlp.down_proj": { "stored_tensors": { "model.layers.5.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.5.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.5.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.5.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.6.input_layernorm": { "stored_tensors": { "model.layers.6.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.6.self_attn.q_proj": { "stored_tensors": { "model.layers.6.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.self_attn.q_proj.trellis": { "shape": [ 256, 256, 64 ], "n_bytes": 8388608, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.6.self_attn.k_proj": { "stored_tensors": { "model.layers.6.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.6.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.self_attn.k_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.6.self_attn.v_proj": { "stored_tensors": { "model.layers.6.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.6.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.self_attn.v_proj.trellis": { "shape": [ 256, 64, 64 ], "n_bytes": 2097152, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 4, "mul1_multiplier": 2212286765 }, "model.layers.6.self_attn.o_proj": { "stored_tensors": { "model.layers.6.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.6.self_attn.q_norm": { "stored_tensors": { "model.layers.6.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.6.self_attn.k_norm": { "stored_tensors": { "model.layers.6.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.6.post_attention_layernorm": { "stored_tensors": { "model.layers.6.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.6.mlp.up_proj": { "stored_tensors": { "model.layers.6.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.6.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.6.mlp.gate_proj": { "stored_tensors": { "model.layers.6.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.6.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.6.mlp.down_proj": { "stored_tensors": { "model.layers.6.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.6.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.6.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.6.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.7.input_layernorm": { "stored_tensors": { "model.layers.7.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.7.self_attn.q_proj": { "stored_tensors": { "model.layers.7.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.7.self_attn.k_proj": { "stored_tensors": { "model.layers.7.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.7.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.7.self_attn.v_proj": { "stored_tensors": { "model.layers.7.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.7.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.7.self_attn.o_proj": { "stored_tensors": { "model.layers.7.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.7.self_attn.q_norm": { "stored_tensors": { "model.layers.7.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.7.self_attn.k_norm": { "stored_tensors": { "model.layers.7.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.7.post_attention_layernorm": { "stored_tensors": { "model.layers.7.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.7.mlp.up_proj": { "stored_tensors": { "model.layers.7.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.7.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.7.mlp.gate_proj": { "stored_tensors": { "model.layers.7.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.7.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.7.mlp.down_proj": { "stored_tensors": { "model.layers.7.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.7.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.7.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.7.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.input_layernorm": { "stored_tensors": { "model.layers.8.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.8.self_attn.q_proj": { "stored_tensors": { "model.layers.8.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.self_attn.k_proj": { "stored_tensors": { "model.layers.8.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.8.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.self_attn.v_proj": { "stored_tensors": { "model.layers.8.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.8.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.self_attn.o_proj": { "stored_tensors": { "model.layers.8.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.self_attn.q_norm": { "stored_tensors": { "model.layers.8.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.8.self_attn.k_norm": { "stored_tensors": { "model.layers.8.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.8.post_attention_layernorm": { "stored_tensors": { "model.layers.8.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.8.mlp.up_proj": { "stored_tensors": { "model.layers.8.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.8.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.mlp.gate_proj": { "stored_tensors": { "model.layers.8.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.8.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.8.mlp.down_proj": { "stored_tensors": { "model.layers.8.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.8.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.8.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.8.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.input_layernorm": { "stored_tensors": { "model.layers.9.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.9.self_attn.q_proj": { "stored_tensors": { "model.layers.9.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.self_attn.k_proj": { "stored_tensors": { "model.layers.9.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.9.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.self_attn.v_proj": { "stored_tensors": { "model.layers.9.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.9.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.self_attn.o_proj": { "stored_tensors": { "model.layers.9.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.self_attn.q_norm": { "stored_tensors": { "model.layers.9.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.9.self_attn.k_norm": { "stored_tensors": { "model.layers.9.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.9.post_attention_layernorm": { "stored_tensors": { "model.layers.9.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.9.mlp.up_proj": { "stored_tensors": { "model.layers.9.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.9.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.mlp.gate_proj": { "stored_tensors": { "model.layers.9.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.9.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.9.mlp.down_proj": { "stored_tensors": { "model.layers.9.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.9.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.9.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.9.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.10.input_layernorm": { "stored_tensors": { "model.layers.10.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.10.self_attn.q_proj": { "stored_tensors": { "model.layers.10.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.10.self_attn.k_proj": { "stored_tensors": { "model.layers.10.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.10.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.10.self_attn.v_proj": { "stored_tensors": { "model.layers.10.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.10.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.10.self_attn.o_proj": { "stored_tensors": { "model.layers.10.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.10.self_attn.q_norm": { "stored_tensors": { "model.layers.10.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.10.self_attn.k_norm": { "stored_tensors": { "model.layers.10.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.10.post_attention_layernorm": { "stored_tensors": { "model.layers.10.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.10.mlp.up_proj": { "stored_tensors": { "model.layers.10.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.10.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.10.mlp.gate_proj": { "stored_tensors": { "model.layers.10.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.10.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.10.mlp.down_proj": { "stored_tensors": { "model.layers.10.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.10.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.10.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.10.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.11.input_layernorm": { "stored_tensors": { "model.layers.11.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.11.self_attn.q_proj": { "stored_tensors": { "model.layers.11.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.11.self_attn.k_proj": { "stored_tensors": { "model.layers.11.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.11.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.11.self_attn.v_proj": { "stored_tensors": { "model.layers.11.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.11.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.11.self_attn.o_proj": { "stored_tensors": { "model.layers.11.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.11.self_attn.q_norm": { "stored_tensors": { "model.layers.11.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.11.self_attn.k_norm": { "stored_tensors": { "model.layers.11.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.11.post_attention_layernorm": { "stored_tensors": { "model.layers.11.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.11.mlp.up_proj": { "stored_tensors": { "model.layers.11.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.11.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.11.mlp.gate_proj": { "stored_tensors": { "model.layers.11.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.11.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.11.mlp.down_proj": { "stored_tensors": { "model.layers.11.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.11.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.11.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.11.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.12.input_layernorm": { "stored_tensors": { "model.layers.12.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.12.self_attn.q_proj": { "stored_tensors": { "model.layers.12.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.12.self_attn.k_proj": { "stored_tensors": { "model.layers.12.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.12.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.12.self_attn.v_proj": { "stored_tensors": { "model.layers.12.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.12.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.12.self_attn.o_proj": { "stored_tensors": { "model.layers.12.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.12.self_attn.q_norm": { "stored_tensors": { "model.layers.12.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.12.self_attn.k_norm": { "stored_tensors": { "model.layers.12.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.12.post_attention_layernorm": { "stored_tensors": { "model.layers.12.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.12.mlp.up_proj": { "stored_tensors": { "model.layers.12.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.12.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.12.mlp.gate_proj": { "stored_tensors": { "model.layers.12.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.12.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.12.mlp.down_proj": { "stored_tensors": { "model.layers.12.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.12.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.12.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.12.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.13.input_layernorm": { "stored_tensors": { "model.layers.13.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.13.self_attn.q_proj": { "stored_tensors": { "model.layers.13.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.13.self_attn.k_proj": { "stored_tensors": { "model.layers.13.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.13.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.13.self_attn.v_proj": { "stored_tensors": { "model.layers.13.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.13.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.13.self_attn.o_proj": { "stored_tensors": { "model.layers.13.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.13.self_attn.q_norm": { "stored_tensors": { "model.layers.13.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.13.self_attn.k_norm": { "stored_tensors": { "model.layers.13.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.13.post_attention_layernorm": { "stored_tensors": { "model.layers.13.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.13.mlp.up_proj": { "stored_tensors": { "model.layers.13.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.13.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.13.mlp.gate_proj": { "stored_tensors": { "model.layers.13.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.13.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.13.mlp.down_proj": { "stored_tensors": { "model.layers.13.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.13.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.13.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.13.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.14.input_layernorm": { "stored_tensors": { "model.layers.14.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.14.self_attn.q_proj": { "stored_tensors": { "model.layers.14.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.14.self_attn.k_proj": { "stored_tensors": { "model.layers.14.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.14.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.14.self_attn.v_proj": { "stored_tensors": { "model.layers.14.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.14.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.14.self_attn.o_proj": { "stored_tensors": { "model.layers.14.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.14.self_attn.q_norm": { "stored_tensors": { "model.layers.14.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.14.self_attn.k_norm": { "stored_tensors": { "model.layers.14.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.14.post_attention_layernorm": { "stored_tensors": { "model.layers.14.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.14.mlp.up_proj": { "stored_tensors": { "model.layers.14.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.14.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.14.mlp.gate_proj": { "stored_tensors": { "model.layers.14.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.14.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.14.mlp.down_proj": { "stored_tensors": { "model.layers.14.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.14.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.14.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.14.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.15.input_layernorm": { "stored_tensors": { "model.layers.15.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.15.self_attn.q_proj": { "stored_tensors": { "model.layers.15.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.15.self_attn.k_proj": { "stored_tensors": { "model.layers.15.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.15.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.15.self_attn.v_proj": { "stored_tensors": { "model.layers.15.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.15.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.15.self_attn.o_proj": { "stored_tensors": { "model.layers.15.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.15.self_attn.q_norm": { "stored_tensors": { "model.layers.15.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.15.self_attn.k_norm": { "stored_tensors": { "model.layers.15.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.15.post_attention_layernorm": { "stored_tensors": { "model.layers.15.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.15.mlp.up_proj": { "stored_tensors": { "model.layers.15.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.15.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.15.mlp.gate_proj": { "stored_tensors": { "model.layers.15.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.15.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.15.mlp.down_proj": { "stored_tensors": { "model.layers.15.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.15.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.15.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.15.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.16.input_layernorm": { "stored_tensors": { "model.layers.16.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.16.self_attn.q_proj": { "stored_tensors": { "model.layers.16.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.16.self_attn.k_proj": { "stored_tensors": { "model.layers.16.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.16.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.16.self_attn.v_proj": { "stored_tensors": { "model.layers.16.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.16.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.16.self_attn.o_proj": { "stored_tensors": { "model.layers.16.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.16.self_attn.q_norm": { "stored_tensors": { "model.layers.16.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.16.self_attn.k_norm": { "stored_tensors": { "model.layers.16.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.16.post_attention_layernorm": { "stored_tensors": { "model.layers.16.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.16.mlp.up_proj": { "stored_tensors": { "model.layers.16.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.16.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.16.mlp.gate_proj": { "stored_tensors": { "model.layers.16.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.16.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.16.mlp.down_proj": { "stored_tensors": { "model.layers.16.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.16.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.16.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.16.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.17.input_layernorm": { "stored_tensors": { "model.layers.17.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.17.self_attn.q_proj": { "stored_tensors": { "model.layers.17.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.17.self_attn.k_proj": { "stored_tensors": { "model.layers.17.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.17.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.17.self_attn.v_proj": { "stored_tensors": { "model.layers.17.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.17.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.17.self_attn.o_proj": { "stored_tensors": { "model.layers.17.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.17.self_attn.q_norm": { "stored_tensors": { "model.layers.17.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.17.self_attn.k_norm": { "stored_tensors": { "model.layers.17.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.17.post_attention_layernorm": { "stored_tensors": { "model.layers.17.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.17.mlp.up_proj": { "stored_tensors": { "model.layers.17.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.17.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.17.mlp.gate_proj": { "stored_tensors": { "model.layers.17.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.17.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.17.mlp.down_proj": { "stored_tensors": { "model.layers.17.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.17.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.17.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.17.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.18.input_layernorm": { "stored_tensors": { "model.layers.18.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.18.self_attn.q_proj": { "stored_tensors": { "model.layers.18.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.18.self_attn.k_proj": { "stored_tensors": { "model.layers.18.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.18.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.18.self_attn.v_proj": { "stored_tensors": { "model.layers.18.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.18.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.18.self_attn.o_proj": { "stored_tensors": { "model.layers.18.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.18.self_attn.q_norm": { "stored_tensors": { "model.layers.18.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.18.self_attn.k_norm": { "stored_tensors": { "model.layers.18.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.18.post_attention_layernorm": { "stored_tensors": { "model.layers.18.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.18.mlp.up_proj": { "stored_tensors": { "model.layers.18.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.18.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.18.mlp.gate_proj": { "stored_tensors": { "model.layers.18.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.18.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.18.mlp.down_proj": { "stored_tensors": { "model.layers.18.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.18.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.18.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.18.mlp.down_proj.trellis": { "shape": [ 768, 256, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.19.input_layernorm": { "stored_tensors": { "model.layers.19.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.19.self_attn.q_proj": { "stored_tensors": { "model.layers.19.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.19.self_attn.k_proj": { "stored_tensors": { "model.layers.19.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.19.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.19.self_attn.v_proj": { "stored_tensors": { "model.layers.19.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.19.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.19.self_attn.o_proj": { "stored_tensors": { "model.layers.19.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.19.self_attn.q_norm": { "stored_tensors": { "model.layers.19.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.19.self_attn.k_norm": { "stored_tensors": { "model.layers.19.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.19.post_attention_layernorm": { "stored_tensors": { "model.layers.19.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.19.mlp.up_proj": { "stored_tensors": { "model.layers.19.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.19.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.19.mlp.gate_proj": { "stored_tensors": { "model.layers.19.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.19.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.19.mlp.down_proj": { "stored_tensors": { "model.layers.19.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.19.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.19.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.19.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.20.input_layernorm": { "stored_tensors": { "model.layers.20.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.20.self_attn.q_proj": { "stored_tensors": { "model.layers.20.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.20.self_attn.k_proj": { "stored_tensors": { "model.layers.20.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.20.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.20.self_attn.v_proj": { "stored_tensors": { "model.layers.20.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.20.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.20.self_attn.o_proj": { "stored_tensors": { "model.layers.20.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.20.self_attn.q_norm": { "stored_tensors": { "model.layers.20.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.20.self_attn.k_norm": { "stored_tensors": { "model.layers.20.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.20.post_attention_layernorm": { "stored_tensors": { "model.layers.20.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.20.mlp.up_proj": { "stored_tensors": { "model.layers.20.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.20.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.20.mlp.gate_proj": { "stored_tensors": { "model.layers.20.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.20.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.20.mlp.down_proj": { "stored_tensors": { "model.layers.20.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.20.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.20.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.20.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.21.input_layernorm": { "stored_tensors": { "model.layers.21.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.21.self_attn.q_proj": { "stored_tensors": { "model.layers.21.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.21.self_attn.k_proj": { "stored_tensors": { "model.layers.21.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.21.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.21.self_attn.v_proj": { "stored_tensors": { "model.layers.21.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.21.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.21.self_attn.o_proj": { "stored_tensors": { "model.layers.21.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.21.self_attn.q_norm": { "stored_tensors": { "model.layers.21.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.21.self_attn.k_norm": { "stored_tensors": { "model.layers.21.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.21.post_attention_layernorm": { "stored_tensors": { "model.layers.21.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.21.mlp.up_proj": { "stored_tensors": { "model.layers.21.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.21.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.21.mlp.gate_proj": { "stored_tensors": { "model.layers.21.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.21.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.21.mlp.down_proj": { "stored_tensors": { "model.layers.21.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.21.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.21.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.21.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.22.input_layernorm": { "stored_tensors": { "model.layers.22.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.22.self_attn.q_proj": { "stored_tensors": { "model.layers.22.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.22.self_attn.k_proj": { "stored_tensors": { "model.layers.22.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.22.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.22.self_attn.v_proj": { "stored_tensors": { "model.layers.22.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.22.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.22.self_attn.o_proj": { "stored_tensors": { "model.layers.22.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.22.self_attn.q_norm": { "stored_tensors": { "model.layers.22.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.22.self_attn.k_norm": { "stored_tensors": { "model.layers.22.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.22.post_attention_layernorm": { "stored_tensors": { "model.layers.22.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.22.mlp.up_proj": { "stored_tensors": { "model.layers.22.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.22.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.22.mlp.gate_proj": { "stored_tensors": { "model.layers.22.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.22.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.22.mlp.down_proj": { "stored_tensors": { "model.layers.22.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.22.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.22.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.22.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.23.input_layernorm": { "stored_tensors": { "model.layers.23.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.23.self_attn.q_proj": { "stored_tensors": { "model.layers.23.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.23.self_attn.k_proj": { "stored_tensors": { "model.layers.23.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.23.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.23.self_attn.v_proj": { "stored_tensors": { "model.layers.23.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.23.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.23.self_attn.o_proj": { "stored_tensors": { "model.layers.23.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.23.self_attn.q_norm": { "stored_tensors": { "model.layers.23.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.23.self_attn.k_norm": { "stored_tensors": { "model.layers.23.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.23.post_attention_layernorm": { "stored_tensors": { "model.layers.23.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.23.mlp.up_proj": { "stored_tensors": { "model.layers.23.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.23.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.23.mlp.gate_proj": { "stored_tensors": { "model.layers.23.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.23.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.23.mlp.down_proj": { "stored_tensors": { "model.layers.23.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.23.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.23.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.23.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.24.input_layernorm": { "stored_tensors": { "model.layers.24.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.24.self_attn.q_proj": { "stored_tensors": { "model.layers.24.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.24.self_attn.k_proj": { "stored_tensors": { "model.layers.24.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.24.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.24.self_attn.v_proj": { "stored_tensors": { "model.layers.24.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.24.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.24.self_attn.o_proj": { "stored_tensors": { "model.layers.24.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.self_attn.o_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.24.self_attn.q_norm": { "stored_tensors": { "model.layers.24.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.24.self_attn.k_norm": { "stored_tensors": { "model.layers.24.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.24.post_attention_layernorm": { "stored_tensors": { "model.layers.24.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.24.mlp.up_proj": { "stored_tensors": { "model.layers.24.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.24.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.24.mlp.gate_proj": { "stored_tensors": { "model.layers.24.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.24.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.24.mlp.down_proj": { "stored_tensors": { "model.layers.24.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.24.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.24.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.24.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.25.input_layernorm": { "stored_tensors": { "model.layers.25.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.25.self_attn.q_proj": { "stored_tensors": { "model.layers.25.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.25.self_attn.k_proj": { "stored_tensors": { "model.layers.25.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.25.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.25.self_attn.v_proj": { "stored_tensors": { "model.layers.25.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.25.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.25.self_attn.o_proj": { "stored_tensors": { "model.layers.25.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.25.self_attn.q_norm": { "stored_tensors": { "model.layers.25.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.25.self_attn.k_norm": { "stored_tensors": { "model.layers.25.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.25.post_attention_layernorm": { "stored_tensors": { "model.layers.25.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.25.mlp.up_proj": { "stored_tensors": { "model.layers.25.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.25.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.25.mlp.gate_proj": { "stored_tensors": { "model.layers.25.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.25.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.25.mlp.down_proj": { "stored_tensors": { "model.layers.25.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.25.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.25.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.25.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.26.input_layernorm": { "stored_tensors": { "model.layers.26.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.26.self_attn.q_proj": { "stored_tensors": { "model.layers.26.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.26.self_attn.k_proj": { "stored_tensors": { "model.layers.26.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.26.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.26.self_attn.v_proj": { "stored_tensors": { "model.layers.26.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.26.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.26.self_attn.o_proj": { "stored_tensors": { "model.layers.26.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.26.self_attn.q_norm": { "stored_tensors": { "model.layers.26.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.26.self_attn.k_norm": { "stored_tensors": { "model.layers.26.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.26.post_attention_layernorm": { "stored_tensors": { "model.layers.26.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.26.mlp.up_proj": { "stored_tensors": { "model.layers.26.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.26.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.26.mlp.gate_proj": { "stored_tensors": { "model.layers.26.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.26.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.26.mlp.down_proj": { "stored_tensors": { "model.layers.26.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.26.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.26.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.26.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.27.input_layernorm": { "stored_tensors": { "model.layers.27.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.27.self_attn.q_proj": { "stored_tensors": { "model.layers.27.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.27.self_attn.k_proj": { "stored_tensors": { "model.layers.27.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.27.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.27.self_attn.v_proj": { "stored_tensors": { "model.layers.27.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.27.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.27.self_attn.o_proj": { "stored_tensors": { "model.layers.27.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.27.self_attn.q_norm": { "stored_tensors": { "model.layers.27.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.27.self_attn.k_norm": { "stored_tensors": { "model.layers.27.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.27.post_attention_layernorm": { "stored_tensors": { "model.layers.27.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.27.mlp.up_proj": { "stored_tensors": { "model.layers.27.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.27.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.27.mlp.gate_proj": { "stored_tensors": { "model.layers.27.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.27.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.27.mlp.down_proj": { "stored_tensors": { "model.layers.27.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.27.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.27.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.27.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.28.input_layernorm": { "stored_tensors": { "model.layers.28.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.28.self_attn.q_proj": { "stored_tensors": { "model.layers.28.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.28.self_attn.k_proj": { "stored_tensors": { "model.layers.28.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.28.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.28.self_attn.v_proj": { "stored_tensors": { "model.layers.28.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.28.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.28.self_attn.o_proj": { "stored_tensors": { "model.layers.28.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.28.self_attn.q_norm": { "stored_tensors": { "model.layers.28.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.28.self_attn.k_norm": { "stored_tensors": { "model.layers.28.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.28.post_attention_layernorm": { "stored_tensors": { "model.layers.28.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.28.mlp.up_proj": { "stored_tensors": { "model.layers.28.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.28.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.28.mlp.gate_proj": { "stored_tensors": { "model.layers.28.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.28.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.28.mlp.down_proj": { "stored_tensors": { "model.layers.28.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.28.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.28.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.28.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.input_layernorm": { "stored_tensors": { "model.layers.29.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.29.self_attn.q_proj": { "stored_tensors": { "model.layers.29.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.self_attn.k_proj": { "stored_tensors": { "model.layers.29.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.29.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.self_attn.v_proj": { "stored_tensors": { "model.layers.29.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.29.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.self_attn.o_proj": { "stored_tensors": { "model.layers.29.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.self_attn.q_norm": { "stored_tensors": { "model.layers.29.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.29.self_attn.k_norm": { "stored_tensors": { "model.layers.29.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.29.post_attention_layernorm": { "stored_tensors": { "model.layers.29.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.29.mlp.up_proj": { "stored_tensors": { "model.layers.29.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.29.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.mlp.gate_proj": { "stored_tensors": { "model.layers.29.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.29.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.29.mlp.down_proj": { "stored_tensors": { "model.layers.29.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.29.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.29.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.29.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.30.input_layernorm": { "stored_tensors": { "model.layers.30.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.30.self_attn.q_proj": { "stored_tensors": { "model.layers.30.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.30.self_attn.k_proj": { "stored_tensors": { "model.layers.30.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.30.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.30.self_attn.v_proj": { "stored_tensors": { "model.layers.30.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.30.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.30.self_attn.o_proj": { "stored_tensors": { "model.layers.30.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.30.self_attn.q_norm": { "stored_tensors": { "model.layers.30.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.30.self_attn.k_norm": { "stored_tensors": { "model.layers.30.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.30.post_attention_layernorm": { "stored_tensors": { "model.layers.30.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.30.mlp.up_proj": { "stored_tensors": { "model.layers.30.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.30.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.30.mlp.gate_proj": { "stored_tensors": { "model.layers.30.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.30.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.30.mlp.down_proj": { "stored_tensors": { "model.layers.30.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.30.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.30.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.30.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.input_layernorm": { "stored_tensors": { "model.layers.31.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.31.self_attn.q_proj": { "stored_tensors": { "model.layers.31.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.self_attn.k_proj": { "stored_tensors": { "model.layers.31.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.31.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.self_attn.v_proj": { "stored_tensors": { "model.layers.31.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.31.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.self_attn.o_proj": { "stored_tensors": { "model.layers.31.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.self_attn.q_norm": { "stored_tensors": { "model.layers.31.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.31.self_attn.k_norm": { "stored_tensors": { "model.layers.31.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.31.post_attention_layernorm": { "stored_tensors": { "model.layers.31.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.31.mlp.up_proj": { "stored_tensors": { "model.layers.31.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.31.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.mlp.gate_proj": { "stored_tensors": { "model.layers.31.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.31.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.31.mlp.down_proj": { "stored_tensors": { "model.layers.31.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.31.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.31.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.31.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.input_layernorm": { "stored_tensors": { "model.layers.32.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.32.self_attn.q_proj": { "stored_tensors": { "model.layers.32.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.self_attn.k_proj": { "stored_tensors": { "model.layers.32.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.32.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.self_attn.v_proj": { "stored_tensors": { "model.layers.32.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.32.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.self_attn.o_proj": { "stored_tensors": { "model.layers.32.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.self_attn.q_norm": { "stored_tensors": { "model.layers.32.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.32.self_attn.k_norm": { "stored_tensors": { "model.layers.32.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.32.post_attention_layernorm": { "stored_tensors": { "model.layers.32.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.32.mlp.up_proj": { "stored_tensors": { "model.layers.32.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.32.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.mlp.gate_proj": { "stored_tensors": { "model.layers.32.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.32.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.32.mlp.down_proj": { "stored_tensors": { "model.layers.32.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.32.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.32.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.32.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.input_layernorm": { "stored_tensors": { "model.layers.33.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.33.self_attn.q_proj": { "stored_tensors": { "model.layers.33.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.self_attn.k_proj": { "stored_tensors": { "model.layers.33.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.33.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.self_attn.v_proj": { "stored_tensors": { "model.layers.33.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.33.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.self_attn.o_proj": { "stored_tensors": { "model.layers.33.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.self_attn.q_norm": { "stored_tensors": { "model.layers.33.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.33.self_attn.k_norm": { "stored_tensors": { "model.layers.33.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.33.post_attention_layernorm": { "stored_tensors": { "model.layers.33.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.33.mlp.up_proj": { "stored_tensors": { "model.layers.33.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.33.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.mlp.up_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.mlp.gate_proj": { "stored_tensors": { "model.layers.33.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.33.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.mlp.gate_proj.trellis": { "shape": [ 256, 768, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.33.mlp.down_proj": { "stored_tensors": { "model.layers.33.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.33.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.33.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.33.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.34.input_layernorm": { "stored_tensors": { "model.layers.34.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.34.self_attn.q_proj": { "stored_tensors": { "model.layers.34.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.self_attn.q_proj.trellis": { "shape": [ 256, 256, 96 ], "n_bytes": 12582912, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.34.self_attn.k_proj": { "stored_tensors": { "model.layers.34.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.34.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.self_attn.k_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.34.self_attn.v_proj": { "stored_tensors": { "model.layers.34.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.34.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.self_attn.v_proj.trellis": { "shape": [ 256, 64, 96 ], "n_bytes": 3145728, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.34.self_attn.o_proj": { "stored_tensors": { "model.layers.34.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.34.self_attn.q_norm": { "stored_tensors": { "model.layers.34.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.34.self_attn.k_norm": { "stored_tensors": { "model.layers.34.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.34.post_attention_layernorm": { "stored_tensors": { "model.layers.34.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.34.mlp.up_proj": { "stored_tensors": { "model.layers.34.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.34.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.mlp.up_proj.trellis": { "shape": [ 256, 768, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.34.mlp.gate_proj": { "stored_tensors": { "model.layers.34.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.34.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.mlp.gate_proj.trellis": { "shape": [ 256, 768, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.34.mlp.down_proj": { "stored_tensors": { "model.layers.34.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.34.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.34.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.34.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.35.input_layernorm": { "stored_tensors": { "model.layers.35.input_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.35.self_attn.q_proj": { "stored_tensors": { "model.layers.35.self_attn.q_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.self_attn.q_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.self_attn.q_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.self_attn.q_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.35.self_attn.k_proj": { "stored_tensors": { "model.layers.35.self_attn.k_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.self_attn.k_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.35.self_attn.k_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.self_attn.k_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.35.self_attn.v_proj": { "stored_tensors": { "model.layers.35.self_attn.v_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.self_attn.v_proj.svh": { "shape": [ 1024 ], "n_bytes": 2048, "dtype": "torch.float16" }, "model.layers.35.self_attn.v_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.self_attn.v_proj.trellis": { "shape": [ 256, 64, 80 ], "n_bytes": 2621440, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.35.self_attn.o_proj": { "stored_tensors": { "model.layers.35.self_attn.o_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.self_attn.o_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.self_attn.o_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.self_attn.o_proj.trellis": { "shape": [ 256, 256, 80 ], "n_bytes": 10485760, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.layers.35.self_attn.q_norm": { "stored_tensors": { "model.layers.35.self_attn.q_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.35.self_attn.k_norm": { "stored_tensors": { "model.layers.35.self_attn.k_norm.weight": { "shape": [ 128 ], "n_bytes": 256, "dtype": "torch.bfloat16" } } }, "model.layers.35.post_attention_layernorm": { "stored_tensors": { "model.layers.35.post_attention_layernorm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "model.layers.35.mlp.up_proj": { "stored_tensors": { "model.layers.35.mlp.up_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.mlp.up_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.35.mlp.up_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.mlp.up_proj.trellis": { "shape": [ 256, 768, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.35.mlp.gate_proj": { "stored_tensors": { "model.layers.35.mlp.gate_proj.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.mlp.gate_proj.svh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.35.mlp.gate_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.mlp.gate_proj.trellis": { "shape": [ 256, 768, 96 ], "n_bytes": 37748736, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 6, "mul1_multiplier": 2212286765 }, "model.layers.35.mlp.down_proj": { "stored_tensors": { "model.layers.35.mlp.down_proj.suh": { "shape": [ 12288 ], "n_bytes": 24576, "dtype": "torch.float16" }, "model.layers.35.mlp.down_proj.svh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "model.layers.35.mlp.down_proj.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "model.layers.35.mlp.down_proj.trellis": { "shape": [ 768, 256, 80 ], "n_bytes": 31457280, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 }, "model.norm": { "stored_tensors": { "model.norm.weight": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.bfloat16" } } }, "lm_head": { "stored_tensors": { "lm_head.suh": { "shape": [ 4096 ], "n_bytes": 8192, "dtype": "torch.float16" }, "lm_head.svh": { "shape": [ 151680 ], "n_bytes": 303360, "dtype": "torch.float16" }, "lm_head.mul1": { "shape": [], "n_bytes": 4, "dtype": "torch.int32" }, "lm_head.trellis": { "shape": [ 256, 9480, 80 ], "n_bytes": 388300800, "dtype": "torch.int16" } }, "quant_format": "exl3", "bits_per_weight": 5, "mul1_multiplier": 2212286765 } } }
{
"quant_method": "exl3",
"version": "1.5.1",
"bits": 5.0,
"head_bits": 5,
"calibration": {
"rows": 196,
"cols": 2048
},
"out_scales": "always",
"codebook": "mul1",
"tensor_storage": {
"model.embed_tokens": {
"stored_tensors": {
"model.embed_tokens.weight": {
"shape": [
151671,
4096
],
"n_bytes": 1242488832,
"dtype": "torch.bfloat16"
}
}
},
"model.layers.0.input_layernorm": {
"stored_tensors": {
"model.layers.0.input_layernorm.weight": {
"shape": [
4096
],
"n_bytes": 8192,
"dtype": "torch.bfloat16"
}
}
},
"model.layers.0.self_attn.q_proj": {
"stored_tensors": {
"model.layers.0.self_attn.q_proj.suh": {
"shape": [
4096
],
"n_bytes": 8192,
"dtype": "torch.float16"
},
"model.layers.0.self_attn.q_proj.svh": {
"shape": [
4096
],
"n_bytes": 8192,
"dtype": "torch.float16"
},
"model.layers.0.self_attn.q_proj.mul1": {
"shape": [],
"n_bytes": 4,
"dtype": "torch.int32"
},
"model.layers.0.self_attn.q_proj.trellis": {
"shape": [
256,
256,
80
],
"n_bytes": 10485760,
"dtype": "torch.int16"
}
},
"quant_format": "exl3",
"bits_per_weight": 5,
"mul1_multiplier": 2212286765
},
"model.layers.0.self_attn.k_proj": {
"stored_tensors": {
"model.layers.0.self_attn.k_proj.suh": {
"shape": [
4096
],
"n_bytes": 8192,
"dtype": "torch.float16"
},
"model.layers.0.self_attn.k_proj.svh": {
"shape": [
1024
],
"n_bytes": 2048,
"dtype": "torch.float16"
},
"model.layers.0.self_attn.k_proj.mul1": {
"shape": [],
"n_bytes": 4,
"dtype": "torch.int32"
},
"model.layers.0.self_attn.k_proj.trellis": {
"shape": [
256,
64,
80
],
"n_bytes": 2621440,
"dtype": "torch.int16"
}