| { | |
| "metadata": { | |
| "total_size": 132834094880 | |
| }, | |
| "weight_map": { | |
| "lm_head.weight": "model-00031-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.v1_input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.v1_weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.v1_weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.w1_input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.w1_weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.w1_weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.w2_input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.w2_weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.experts.mlp.w2_weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.ffn.router.layer.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.Wqkv.input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.Wqkv.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.Wqkv.weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.kv_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.out_proj.input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.out_proj.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.attn.out_proj.weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.norm_1.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.0.norm_attn_norm.norm_2.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.v1_input_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.v1_weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.v1_weight_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.w1_input_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.w1_weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.w1_weight_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.w2_input_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.w2_weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.experts.mlp.w2_weight_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.1.ffn.router.layer.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.Wqkv.input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.Wqkv.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.Wqkv.weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.kv_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.out_proj.input_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.out_proj.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.attn.out_proj.weight_scale": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.norm_1.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.1.norm_attn_norm.norm_2.weight": "model-00001-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.v1_input_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.v1_weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.v1_weight_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.w1_input_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.w1_weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.w1_weight_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.w2_input_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.w2_weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.experts.mlp.w2_weight_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.10.ffn.router.layer.weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.Wqkv.input_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.Wqkv.weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.Wqkv.weight_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.kv_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.out_proj.input_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.out_proj.weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.attn.out_proj.weight_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.norm_1.weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.10.norm_attn_norm.norm_2.weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.v1_input_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.v1_weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.v1_weight_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.w1_input_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.w1_weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.w1_weight_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.w2_input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.w2_weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.experts.mlp.w2_weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.11.ffn.router.layer.weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.Wqkv.input_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.Wqkv.weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.Wqkv.weight_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.kv_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.out_proj.input_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.out_proj.weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.attn.out_proj.weight_scale": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.norm_1.weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.11.norm_attn_norm.norm_2.weight": "model-00009-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.v1_input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.v1_weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.v1_weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.w1_input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.w1_weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.w1_weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.w2_input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.w2_weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.experts.mlp.w2_weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.ffn.router.layer.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.Wqkv.input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.Wqkv.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.Wqkv.weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.kv_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.out_proj.input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.out_proj.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.attn.out_proj.weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.norm_1.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.12.norm_attn_norm.norm_2.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.v1_input_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.v1_weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.v1_weight_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.w1_input_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.w1_weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.w1_weight_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.w2_input_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.w2_weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.experts.mlp.w2_weight_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.13.ffn.router.layer.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.Wqkv.input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.Wqkv.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.Wqkv.weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.kv_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.out_proj.input_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.out_proj.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.attn.out_proj.weight_scale": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.norm_1.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.13.norm_attn_norm.norm_2.weight": "model-00010-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.v1_input_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.v1_weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.v1_weight_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.w1_input_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.w1_weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.w1_weight_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.w2_input_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.w2_weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.experts.mlp.w2_weight_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.14.ffn.router.layer.weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.Wqkv.input_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.Wqkv.weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.Wqkv.weight_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.kv_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.out_proj.input_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.out_proj.weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.attn.out_proj.weight_scale": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.norm_1.weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.14.norm_attn_norm.norm_2.weight": "model-00011-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.v1_input_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.v1_weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.v1_weight_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.w1_input_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.w1_weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.w1_weight_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.w2_input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.w2_weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.experts.mlp.w2_weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.15.ffn.router.layer.weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.Wqkv.input_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.Wqkv.weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.Wqkv.weight_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.kv_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.out_proj.input_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.out_proj.weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.attn.out_proj.weight_scale": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.norm_1.weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.15.norm_attn_norm.norm_2.weight": "model-00012-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.v1_input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.v1_weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.v1_weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.w1_input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.w1_weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.w1_weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.w2_input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.w2_weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.experts.mlp.w2_weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.ffn.router.layer.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.Wqkv.input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.Wqkv.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.Wqkv.weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.kv_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.out_proj.input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.out_proj.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.attn.out_proj.weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.norm_1.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.16.norm_attn_norm.norm_2.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.v1_input_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.v1_weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.v1_weight_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.w1_input_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.w1_weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.w1_weight_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.w2_input_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.w2_weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.experts.mlp.w2_weight_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.17.ffn.router.layer.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.Wqkv.input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.Wqkv.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.Wqkv.weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.kv_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.out_proj.input_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.out_proj.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.attn.out_proj.weight_scale": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.norm_1.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.17.norm_attn_norm.norm_2.weight": "model-00013-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.v1_input_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.v1_weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.v1_weight_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.w1_input_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.w1_weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.w1_weight_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.w2_input_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.w2_weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.experts.mlp.w2_weight_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.18.ffn.router.layer.weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.Wqkv.input_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.Wqkv.weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.Wqkv.weight_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.kv_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.out_proj.input_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.out_proj.weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.attn.out_proj.weight_scale": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.norm_1.weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.18.norm_attn_norm.norm_2.weight": "model-00014-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.v1_input_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.v1_weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.v1_weight_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.w1_input_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.w1_weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.w1_weight_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.w2_input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.w2_weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.experts.mlp.w2_weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.19.ffn.router.layer.weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.Wqkv.input_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.Wqkv.weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.Wqkv.weight_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.kv_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.out_proj.input_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.out_proj.weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.attn.out_proj.weight_scale": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.norm_1.weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.19.norm_attn_norm.norm_2.weight": "model-00015-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.v1_input_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.v1_weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.v1_weight_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.w1_input_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.w1_weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.w1_weight_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.w2_input_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.w2_weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.experts.mlp.w2_weight_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.2.ffn.router.layer.weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.Wqkv.input_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.Wqkv.weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.Wqkv.weight_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.kv_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.out_proj.input_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.out_proj.weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.attn.out_proj.weight_scale": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.norm_1.weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.2.norm_attn_norm.norm_2.weight": "model-00002-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.v1_input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.v1_weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.v1_weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.w1_input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.w1_weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.w1_weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.w2_input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.w2_weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.experts.mlp.w2_weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.ffn.router.layer.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.Wqkv.input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.Wqkv.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.Wqkv.weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.kv_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.out_proj.input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.out_proj.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.attn.out_proj.weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.norm_1.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.20.norm_attn_norm.norm_2.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.v1_input_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.v1_weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.v1_weight_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.w1_input_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.w1_weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.w1_weight_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.w2_input_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.w2_weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.experts.mlp.w2_weight_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.21.ffn.router.layer.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.Wqkv.input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.Wqkv.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.Wqkv.weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.kv_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.out_proj.input_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.out_proj.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.attn.out_proj.weight_scale": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.norm_1.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.21.norm_attn_norm.norm_2.weight": "model-00016-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.v1_input_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.v1_weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.v1_weight_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.w1_input_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.w1_weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.w1_weight_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.w2_input_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.w2_weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.experts.mlp.w2_weight_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.22.ffn.router.layer.weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.Wqkv.input_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.Wqkv.weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.Wqkv.weight_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.kv_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.out_proj.input_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.out_proj.weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.attn.out_proj.weight_scale": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.norm_1.weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.22.norm_attn_norm.norm_2.weight": "model-00017-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.v1_input_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.v1_weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.v1_weight_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.w1_input_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.w1_weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.w1_weight_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.w2_input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.w2_weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.experts.mlp.w2_weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.23.ffn.router.layer.weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.Wqkv.input_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.Wqkv.weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.Wqkv.weight_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.kv_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.out_proj.input_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.out_proj.weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.attn.out_proj.weight_scale": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.norm_1.weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.23.norm_attn_norm.norm_2.weight": "model-00018-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.v1_input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.v1_weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.v1_weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.w1_input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.w1_weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.w1_weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.w2_input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.w2_weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.experts.mlp.w2_weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.ffn.router.layer.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.Wqkv.input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.Wqkv.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.Wqkv.weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.kv_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.out_proj.input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.out_proj.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.attn.out_proj.weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.norm_1.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.24.norm_attn_norm.norm_2.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.v1_input_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.v1_weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.v1_weight_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.w1_input_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.w1_weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.w1_weight_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.w2_input_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.w2_weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.experts.mlp.w2_weight_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.25.ffn.router.layer.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.Wqkv.input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.Wqkv.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.Wqkv.weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.kv_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.out_proj.input_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.out_proj.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.attn.out_proj.weight_scale": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.norm_1.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.25.norm_attn_norm.norm_2.weight": "model-00019-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.v1_input_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.v1_weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.v1_weight_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.w1_input_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.w1_weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.w1_weight_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.w2_input_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.w2_weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.experts.mlp.w2_weight_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.26.ffn.router.layer.weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.Wqkv.input_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.Wqkv.weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.Wqkv.weight_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.kv_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.out_proj.input_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.out_proj.weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.attn.out_proj.weight_scale": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.norm_1.weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.26.norm_attn_norm.norm_2.weight": "model-00020-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.v1_input_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.v1_weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.v1_weight_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.w1_input_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.w1_weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.w1_weight_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.w2_input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.w2_weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.experts.mlp.w2_weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.27.ffn.router.layer.weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.Wqkv.input_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.Wqkv.weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.Wqkv.weight_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.kv_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.out_proj.input_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.out_proj.weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.attn.out_proj.weight_scale": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.norm_1.weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.27.norm_attn_norm.norm_2.weight": "model-00021-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.v1_input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.v1_weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.v1_weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.w1_input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.w1_weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.w1_weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.w2_input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.w2_weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.experts.mlp.w2_weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.ffn.router.layer.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.Wqkv.input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.Wqkv.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.Wqkv.weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.kv_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.out_proj.input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.out_proj.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.attn.out_proj.weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.norm_1.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.28.norm_attn_norm.norm_2.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.v1_input_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.v1_weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.v1_weight_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.w1_input_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.w1_weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.w1_weight_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.w2_input_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.w2_weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.experts.mlp.w2_weight_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.29.ffn.router.layer.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.Wqkv.input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.Wqkv.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.Wqkv.weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.kv_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.out_proj.input_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.out_proj.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.attn.out_proj.weight_scale": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.norm_1.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.29.norm_attn_norm.norm_2.weight": "model-00022-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.v1_input_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.v1_weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.v1_weight_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.w1_input_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.w1_weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.w1_weight_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.w2_input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.w2_weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.experts.mlp.w2_weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.3.ffn.router.layer.weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.Wqkv.input_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.Wqkv.weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.Wqkv.weight_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.kv_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.out_proj.input_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.out_proj.weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.attn.out_proj.weight_scale": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.norm_1.weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.3.norm_attn_norm.norm_2.weight": "model-00003-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.v1_input_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.v1_weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.v1_weight_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.w1_input_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.w1_weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.w1_weight_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.w2_input_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.w2_weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.experts.mlp.w2_weight_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.30.ffn.router.layer.weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.Wqkv.input_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.Wqkv.weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.Wqkv.weight_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.kv_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.out_proj.input_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.out_proj.weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.attn.out_proj.weight_scale": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.norm_1.weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.30.norm_attn_norm.norm_2.weight": "model-00023-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.v1_input_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.v1_weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.v1_weight_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.w1_input_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.w1_weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.w1_weight_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.w2_input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.w2_weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.experts.mlp.w2_weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.31.ffn.router.layer.weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.Wqkv.input_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.Wqkv.weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.Wqkv.weight_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.kv_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.out_proj.input_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.out_proj.weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.attn.out_proj.weight_scale": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.norm_1.weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.31.norm_attn_norm.norm_2.weight": "model-00024-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.v1_input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.v1_weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.v1_weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.w1_input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.w1_weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.w1_weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.w2_input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.w2_weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.experts.mlp.w2_weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.ffn.router.layer.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.Wqkv.input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.Wqkv.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.Wqkv.weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.kv_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.out_proj.input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.out_proj.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.attn.out_proj.weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.norm_1.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.32.norm_attn_norm.norm_2.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.v1_input_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.v1_weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.v1_weight_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.w1_input_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.w1_weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.w1_weight_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.w2_input_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.w2_weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.experts.mlp.w2_weight_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.33.ffn.router.layer.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.Wqkv.input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.Wqkv.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.Wqkv.weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.kv_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.out_proj.input_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.out_proj.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.attn.out_proj.weight_scale": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.norm_1.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.33.norm_attn_norm.norm_2.weight": "model-00025-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.v1_input_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.v1_weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.v1_weight_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.w1_input_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.w1_weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.w1_weight_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.w2_input_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.w2_weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.experts.mlp.w2_weight_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.34.ffn.router.layer.weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.Wqkv.input_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.Wqkv.weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.Wqkv.weight_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.kv_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.out_proj.input_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.out_proj.weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.attn.out_proj.weight_scale": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.norm_1.weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.34.norm_attn_norm.norm_2.weight": "model-00026-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.v1_input_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.v1_weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.v1_weight_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.w1_input_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.w1_weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.w1_weight_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.w2_input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.w2_weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.experts.mlp.w2_weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.35.ffn.router.layer.weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.Wqkv.input_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.Wqkv.weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.Wqkv.weight_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.kv_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.out_proj.input_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.out_proj.weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.attn.out_proj.weight_scale": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.norm_1.weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.35.norm_attn_norm.norm_2.weight": "model-00027-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.v1_input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.v1_weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.v1_weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.w1_input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.w1_weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.w1_weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.w2_input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.w2_weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.experts.mlp.w2_weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.ffn.router.layer.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.Wqkv.input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.Wqkv.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.Wqkv.weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.kv_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.out_proj.input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.out_proj.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.attn.out_proj.weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.norm_1.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.36.norm_attn_norm.norm_2.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.v1_input_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.v1_weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.v1_weight_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.w1_input_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.w1_weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.w1_weight_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.w2_input_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.w2_weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.experts.mlp.w2_weight_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.37.ffn.router.layer.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.Wqkv.input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.Wqkv.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.Wqkv.weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.kv_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.out_proj.input_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.out_proj.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.attn.out_proj.weight_scale": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.norm_1.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.37.norm_attn_norm.norm_2.weight": "model-00028-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.v1_input_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.v1_weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.v1_weight_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.w1_input_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.w1_weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.w1_weight_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.w2_input_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.w2_weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.experts.mlp.w2_weight_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.38.ffn.router.layer.weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.Wqkv.input_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.Wqkv.weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.Wqkv.weight_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.kv_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.out_proj.input_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.out_proj.weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.attn.out_proj.weight_scale": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.norm_1.weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.38.norm_attn_norm.norm_2.weight": "model-00029-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.v1_input_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.v1_weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.v1_weight_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.w1_input_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.w1_weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.w1_weight_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.w2_input_scale": "model-00031-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.w2_weight": "model-00031-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.experts.mlp.w2_weight_scale": "model-00031-of-00031.safetensors", | |
| "transformer.blocks.39.ffn.router.layer.weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.Wqkv.input_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.Wqkv.weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.Wqkv.weight_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.kv_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.out_proj.input_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.out_proj.weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.attn.out_proj.weight_scale": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.norm_1.weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.39.norm_attn_norm.norm_2.weight": "model-00030-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.v1_input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.v1_weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.v1_weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.w1_input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.w1_weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.w1_weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.w2_input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.w2_weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.experts.mlp.w2_weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.ffn.router.layer.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.Wqkv.input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.Wqkv.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.Wqkv.weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.kv_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.out_proj.input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.out_proj.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.attn.out_proj.weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.norm_1.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.4.norm_attn_norm.norm_2.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.v1_input_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.v1_weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.v1_weight_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.w1_input_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.w1_weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.w1_weight_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.w2_input_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.w2_weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.experts.mlp.w2_weight_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.5.ffn.router.layer.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.Wqkv.input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.Wqkv.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.Wqkv.weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.kv_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.out_proj.input_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.out_proj.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.attn.out_proj.weight_scale": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.norm_1.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.5.norm_attn_norm.norm_2.weight": "model-00004-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.v1_input_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.v1_weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.v1_weight_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.w1_input_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.w1_weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.w1_weight_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.w2_input_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.w2_weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.experts.mlp.w2_weight_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.6.ffn.router.layer.weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.Wqkv.input_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.Wqkv.weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.Wqkv.weight_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.kv_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.out_proj.input_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.out_proj.weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.attn.out_proj.weight_scale": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.norm_1.weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.6.norm_attn_norm.norm_2.weight": "model-00005-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.v1_input_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.v1_weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.v1_weight_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.w1_input_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.w1_weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.w1_weight_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.w2_input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.w2_weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.experts.mlp.w2_weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.7.ffn.router.layer.weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.Wqkv.input_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.Wqkv.weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.Wqkv.weight_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.kv_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.out_proj.input_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.out_proj.weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.attn.out_proj.weight_scale": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.norm_1.weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.7.norm_attn_norm.norm_2.weight": "model-00006-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.v1_input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.v1_weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.v1_weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.w1_input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.w1_weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.w1_weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.w2_input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.w2_weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.experts.mlp.w2_weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.ffn.router.layer.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.Wqkv.input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.Wqkv.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.Wqkv.weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.kv_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.out_proj.input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.out_proj.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.attn.out_proj.weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.norm_1.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.8.norm_attn_norm.norm_2.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.v1_input_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.v1_weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.v1_weight_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.w1_input_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.w1_weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.w1_weight_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.w2_input_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.w2_weight": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.experts.mlp.w2_weight_scale": "model-00008-of-00031.safetensors", | |
| "transformer.blocks.9.ffn.router.layer.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.Wqkv.input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.Wqkv.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.Wqkv.weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.kv_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.out_proj.input_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.out_proj.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.attn.out_proj.weight_scale": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.norm_1.weight": "model-00007-of-00031.safetensors", | |
| "transformer.blocks.9.norm_attn_norm.norm_2.weight": "model-00007-of-00031.safetensors", | |
| "transformer.norm_f.weight": "model-00031-of-00031.safetensors", | |
| "transformer.wte.weight": "model-00001-of-00031.safetensors" | |
| } | |
| } | |