aryan5v's picture
Add BF16-sourced Trim INT6 and native NVFP4 encoder for Track C
dc6ab6c verified
Raw History Blame Contribute Delete
50.5 kB
{
"format_version": 1,
"language_layers": 50,
"matrices": {
"model.language_model.layers.0.mlp.down_proj.weight": {
"global_scale": 2626.44287109375
},
"model.language_model.layers.0.mlp.gate_proj.weight": {
"global_scale": 7864.31982421875
},
"model.language_model.layers.0.mlp.up_proj.weight": {
"global_scale": 8290.6982421875
},
"model.language_model.layers.0.self_attn.k_proj.weight": {
"global_scale": 8765.9619140625
},
"model.language_model.layers.0.self_attn.o_proj.weight": {
"global_scale": 3780.923095703125
},
"model.language_model.layers.0.self_attn.q_proj.weight": {
"global_scale": 7864.31982421875
},
"model.language_model.layers.0.self_attn.v_proj.weight": {
"global_scale": 19802.244140625
},
"model.language_model.layers.1.mlp.down_proj.weight": {
"global_scale": 2422.98583984375
},
"model.language_model.layers.1.mlp.gate_proj.weight": {
"global_scale": 7955.23681640625
},
"model.language_model.layers.1.mlp.up_proj.weight": {
"global_scale": 5461.33349609375
},
"model.language_model.layers.1.self_attn.k_proj.weight": {
"global_scale": 7094.10302734375
},
"model.language_model.layers.1.self_attn.o_proj.weight": {
"global_scale": 6284.27392578125
},
"model.language_model.layers.1.self_attn.q_proj.weight": {
"global_scale": 10836.6611328125
},
"model.language_model.layers.1.self_attn.v_proj.weight": {
"global_scale": 14411.0576171875
},
"model.language_model.layers.10.mlp.down_proj.weight": {
"global_scale": 1911.4666748046875
},
"model.language_model.layers.10.mlp.gate_proj.weight": {
"global_scale": 5213.0908203125
},
"model.language_model.layers.10.mlp.up_proj.weight": {
"global_scale": 6616.61572265625
},
"model.language_model.layers.10.self_attn.k_proj.weight": {
"global_scale": 3373.176513671875
},
"model.language_model.layers.10.self_attn.o_proj.weight": {
"global_scale": 2688.0
},
"model.language_model.layers.10.self_attn.q_proj.weight": {
"global_scale": 11421.2119140625
},
"model.language_model.layers.10.self_attn.v_proj.weight": {
"global_scale": 9972.8701171875
},
"model.language_model.layers.11.mlp.down_proj.weight": {
"global_scale": 1890.4615478515625
},
"model.language_model.layers.11.mlp.gate_proj.weight": {
"global_scale": 5806.9873046875
},
"model.language_model.layers.11.mlp.up_proj.weight": {
"global_scale": 7561.84619140625
},
"model.language_model.layers.11.self_attn.k_proj.weight": {
"global_scale": 3292.478271484375
},
"model.language_model.layers.11.self_attn.o_proj.weight": {
"global_scale": 5831.59326171875
},
"model.language_model.layers.11.self_attn.q_proj.weight": {
"global_scale": 10668.6513671875
},
"model.language_model.layers.11.self_attn.v_proj.weight": {
"global_scale": 21011.54296875
},
"model.language_model.layers.12.mlp.down_proj.weight": {
"global_scale": 2060.263671875
},
"model.language_model.layers.12.mlp.gate_proj.weight": {
"global_scale": 5571.88671875
},
"model.language_model.layers.12.mlp.up_proj.weight": {
"global_scale": 7021.71435546875
},
"model.language_model.layers.12.self_attn.k_proj.weight": {
"global_scale": 5252.8857421875
},
"model.language_model.layers.12.self_attn.o_proj.weight": {
"global_scale": 3171.0966796875
},
"model.language_model.layers.12.self_attn.q_proj.weight": {
"global_scale": 10347.7900390625
},
"model.language_model.layers.12.self_attn.v_proj.weight": {
"global_scale": 12684.38671875
},
"model.language_model.layers.13.mlp.down_proj.weight": {
"global_scale": 2219.767822265625
},
"model.language_model.layers.13.mlp.gate_proj.weight": {
"global_scale": 4047.811767578125
},
"model.language_model.layers.13.mlp.up_proj.weight": {
"global_scale": 8241.0546875
},
"model.language_model.layers.13.self_attn.k_proj.weight": {
"global_scale": 4247.70361328125
},
"model.language_model.layers.13.self_attn.o_proj.weight": {
"global_scale": 3423.522216796875
},
"model.language_model.layers.13.self_attn.q_proj.weight": {
"global_scale": 9299.02734375
},
"model.language_model.layers.13.self_attn.v_proj.weight": {
"global_scale": 15728.6396484375
},
"model.language_model.layers.14.mlp.down_proj.weight": {
"global_scale": 2234.181884765625
},
"model.language_model.layers.14.mlp.gate_proj.weight": {
"global_scale": 4120.52734375
},
"model.language_model.layers.14.mlp.up_proj.weight": {
"global_scale": 6881.27978515625
},
"model.language_model.layers.14.self_attn.k_proj.weight": {
"global_scale": 5213.0908203125
},
"model.language_model.layers.14.self_attn.o_proj.weight": {
"global_scale": 2797.26806640625
},
"model.language_model.layers.14.self_attn.q_proj.weight": {
"global_scale": 12125.603515625
},
"model.language_model.layers.14.self_attn.v_proj.weight": {
"global_scale": 13045.0810546875
},
"model.language_model.layers.15.mlp.down_proj.weight": {
"global_scale": 2309.154296875
},
"model.language_model.layers.15.mlp.gate_proj.weight": {
"global_scale": 4497.56884765625
},
"model.language_model.layers.15.mlp.up_proj.weight": {
"global_scale": 7399.2255859375
},
"model.language_model.layers.15.self_attn.k_proj.weight": {
"global_scale": 4300.80029296875
},
"model.language_model.layers.15.self_attn.o_proj.weight": {
"global_scale": 5252.8857421875
},
"model.language_model.layers.15.self_attn.q_proj.weight": {
"global_scale": 8655.6982421875
},
"model.language_model.layers.15.self_attn.v_proj.weight": {
"global_scale": 16581.396484375
},
"model.language_model.layers.16.mlp.down_proj.weight": {
"global_scale": 1988.8092041015625
},
"model.language_model.layers.16.mlp.gate_proj.weight": {
"global_scale": 3423.522216796875
},
"model.language_model.layers.16.mlp.up_proj.weight": {
"global_scale": 7603.62451171875
},
"model.language_model.layers.16.self_attn.k_proj.weight": {
"global_scale": 4382.98095703125
},
"model.language_model.layers.16.self_attn.o_proj.weight": {
"global_scale": 4300.80029296875
},
"model.language_model.layers.16.self_attn.q_proj.weight": {
"global_scale": 8765.9619140625
},
"model.language_model.layers.16.self_attn.v_proj.weight": {
"global_scale": 13694.0888671875
},
"model.language_model.layers.17.mlp.down_proj.weight": {
"global_scale": 2191.490478515625
},
"model.language_model.layers.17.mlp.gate_proj.weight": {
"global_scale": 3185.77783203125
},
"model.language_model.layers.17.mlp.up_proj.weight": {
"global_scale": 9830.400390625
},
"model.language_model.layers.17.self_attn.k_proj.weight": {
"global_scale": 4355.24072265625
},
"model.language_model.layers.17.self_attn.o_proj.weight": {
"global_scale": 6171.5517578125
},
"model.language_model.layers.17.self_attn.q_proj.weight": {
"global_scale": 11054.2646484375
},
"model.language_model.layers.17.self_attn.v_proj.weight": {
"global_scale": 19660.80078125
},
"model.language_model.layers.18.mlp.down_proj.weight": {
"global_scale": 2511.416015625
},
"model.language_model.layers.18.mlp.gate_proj.weight": {
"global_scale": 3142.136962890625
},
"model.language_model.layers.18.mlp.up_proj.weight": {
"global_scale": 5549.4189453125
},
"model.language_model.layers.18.self_attn.k_proj.weight": {
"global_scale": 6089.62841796875
},
"model.language_model.layers.18.self_attn.o_proj.weight": {
"global_scale": 6199.3515625
},
"model.language_model.layers.18.self_attn.q_proj.weight": {
"global_scale": 11967.443359375
},
"model.language_model.layers.18.self_attn.v_proj.weight": {
"global_scale": 19521.361328125
},
"model.language_model.layers.19.mlp.down_proj.weight": {
"global_scale": 1764.4307861328125
},
"model.language_model.layers.19.mlp.gate_proj.weight": {
"global_scale": 3200.59521484375
},
"model.language_model.layers.19.mlp.up_proj.weight": {
"global_scale": 3865.8876953125
},
"model.language_model.layers.19.self_attn.k_proj.weight": {
"global_scale": 4024.140380859375
},
"model.language_model.layers.19.self_attn.o_proj.weight": {
"global_scale": 8936.7275390625
},
"model.language_model.layers.19.self_attn.q_proj.weight": {
"global_scale": 10668.6513671875
},
"model.language_model.layers.19.self_attn.v_proj.weight": {
"global_scale": 20695.580078125
},
"model.language_model.layers.2.mlp.down_proj.weight": {
"global_scale": 2340.5712890625
},
"model.language_model.layers.2.mlp.gate_proj.weight": {
"global_scale": 7819.63671875
},
"model.language_model.layers.2.mlp.up_proj.weight": {
"global_scale": 6813.1484375
},
"model.language_model.layers.2.self_attn.k_proj.weight": {
"global_scale": 6144.00048828125
},
"model.language_model.layers.2.self_attn.o_proj.weight": {
"global_scale": 6461.2958984375
},
"model.language_model.layers.2.self_attn.q_proj.weight": {
"global_scale": 7819.63671875
},
"model.language_model.layers.2.self_attn.v_proj.weight": {
"global_scale": 16886.576171875
},
"model.language_model.layers.20.mlp.down_proj.weight": {
"global_scale": 2205.53857421875
},
"model.language_model.layers.20.mlp.gate_proj.weight": {
"global_scale": 3292.478271484375
},
"model.language_model.layers.20.mlp.up_proj.weight": {
"global_scale": 7281.77734375
},
"model.language_model.layers.20.self_attn.k_proj.weight": {
"global_scale": 4327.84912109375
},
"model.language_model.layers.20.self_attn.o_proj.weight": {
"global_scale": 6746.35302734375
},
"model.language_model.layers.20.self_attn.q_proj.weight": {
"global_scale": 10794.1650390625
},
"model.language_model.layers.20.self_attn.v_proj.weight": {
"global_scale": 18350.080078125
},
"model.language_model.layers.21.mlp.down_proj.weight": {
"global_scale": 2475.280517578125
},
"model.language_model.layers.21.mlp.gate_proj.weight": {
"global_scale": 3276.800048828125
},
"model.language_model.layers.21.mlp.up_proj.weight": {
"global_scale": 4986.43505859375
},
"model.language_model.layers.21.self_attn.k_proj.weight": {
"global_scale": 5376.0
},
"model.language_model.layers.21.self_attn.o_proj.weight": {
"global_scale": 4681.142578125
},
"model.language_model.layers.21.self_attn.q_proj.weight": {
"global_scale": 11098.837890625
},
"model.language_model.layers.21.self_attn.v_proj.weight": {
"global_scale": 17531.923828125
},
"model.language_model.layers.22.mlp.down_proj.weight": {
"global_scale": 2060.263671875
},
"model.language_model.layers.22.mlp.gate_proj.weight": {
"global_scale": 3493.035400390625
},
"model.language_model.layers.22.mlp.up_proj.weight": {
"global_scale": 6144.00048828125
},
"model.language_model.layers.22.self_attn.k_proj.weight": {
"global_scale": 6144.00048828125
},
"model.language_model.layers.22.self_attn.o_proj.weight": {
"global_scale": 5527.13232421875
},
"model.language_model.layers.22.self_attn.q_proj.weight": {
"global_scale": 9760.6806640625
},
"model.language_model.layers.22.self_attn.v_proj.weight": {
"global_scale": 15463.55078125
},
"model.language_model.layers.23.mlp.down_proj.weight": {
"global_scale": 2060.263671875
},
"model.language_model.layers.23.mlp.gate_proj.weight": {
"global_scale": 3822.933349609375
},
"model.language_model.layers.23.mlp.up_proj.weight": {
"global_scale": 8143.52685546875
},
"model.language_model.layers.23.self_attn.k_proj.weight": {
"global_scale": 5097.244140625
},
"model.language_model.layers.23.self_attn.o_proj.weight": {
"global_scale": 6584.95654296875
},
"model.language_model.layers.23.self_attn.q_proj.weight": {
"global_scale": 12233.38671875
},
"model.language_model.layers.23.self_attn.v_proj.weight": {
"global_scale": 13901.576171875
},
"model.language_model.layers.24.mlp.down_proj.weight": {
"global_scale": 3602.764404296875
},
"model.language_model.layers.24.mlp.gate_proj.weight": {
"global_scale": 4812.083984375
},
"model.language_model.layers.24.mlp.up_proj.weight": {
"global_scale": 8601.6005859375
},
"model.language_model.layers.24.self_attn.k_proj.weight": {
"global_scale": 5687.0078125
},
"model.language_model.layers.24.self_attn.o_proj.weight": {
"global_scale": 8340.9453125
},
"model.language_model.layers.24.self_attn.q_proj.weight": {
"global_scale": 6713.44384765625
},
"model.language_model.layers.24.self_attn.v_proj.weight": {
"global_scale": 15463.55078125
},
"model.language_model.layers.25.mlp.down_proj.weight": {
"global_scale": 1922.1451416015625
},
"model.language_model.layers.25.mlp.gate_proj.weight": {
"global_scale": 4950.56103515625
},
"model.language_model.layers.25.mlp.up_proj.weight": {
"global_scale": 6431.1025390625
},
"model.language_model.layers.25.self_attn.k_proj.weight": {
"global_scale": 4915.2001953125
},
"model.language_model.layers.25.self_attn.o_proj.weight": {
"global_scale": 5527.13232421875
},
"model.language_model.layers.25.self_attn.q_proj.weight": {
"global_scale": 12179.2568359375
},
"model.language_model.layers.25.self_attn.v_proj.weight": {
"global_scale": 17758.142578125
},
"model.language_model.layers.26.mlp.down_proj.weight": {
"global_scale": 2177.620361328125
},
"model.language_model.layers.26.mlp.gate_proj.weight": {
"global_scale": 4024.140380859375
},
"model.language_model.layers.26.mlp.up_proj.weight": {
"global_scale": 3440.639892578125
},
"model.language_model.layers.26.self_attn.k_proj.weight": {
"global_scale": 5549.4189453125
},
"model.language_model.layers.26.self_attn.o_proj.weight": {
"global_scale": 7320.51025390625
},
"model.language_model.layers.26.self_attn.q_proj.weight": {
"global_scale": 10752.0
},
"model.language_model.layers.26.self_attn.v_proj.weight": {
"global_scale": 18350.080078125
},
"model.language_model.layers.27.mlp.down_proj.weight": {
"global_scale": 1849.806396484375
},
"model.language_model.layers.27.mlp.gate_proj.weight": {
"global_scale": 4300.80029296875
},
"model.language_model.layers.27.mlp.up_proj.weight": {
"global_scale": 4024.140380859375
},
"model.language_model.layers.27.self_attn.k_proj.weight": {
"global_scale": 4497.56884765625
},
"model.language_model.layers.27.self_attn.o_proj.weight": {
"global_scale": 2567.641845703125
},
"model.language_model.layers.27.self_attn.q_proj.weight": {
"global_scale": 8290.6982421875
},
"model.language_model.layers.27.self_attn.v_proj.weight": {
"global_scale": 11613.974609375
},
"model.language_model.layers.28.mlp.down_proj.weight": {
"global_scale": 2493.217529296875
},
"model.language_model.layers.28.mlp.gate_proj.weight": {
"global_scale": 5376.0
},
"model.language_model.layers.28.mlp.up_proj.weight": {
"global_scale": 7094.10302734375
},
"model.language_model.layers.28.self_attn.k_proj.weight": {
"global_scale": 4986.43505859375
},
"model.language_model.layers.28.self_attn.o_proj.weight": {
"global_scale": 9362.28515625
},
"model.language_model.layers.28.self_attn.q_proj.weight": {
"global_scale": 11915.63671875
},
"model.language_model.layers.28.self_attn.v_proj.weight": {
"global_scale": 13626.296875
},
"model.language_model.layers.29.mlp.down_proj.weight": {
"global_scale": 1869.9130859375
},
"model.language_model.layers.29.mlp.gate_proj.weight": {
"global_scale": 4439.53564453125
},
"model.language_model.layers.29.mlp.up_proj.weight": {
"global_scale": 6847.04443359375
},
"model.language_model.layers.29.self_attn.k_proj.weight": {
"global_scale": 5334.32568359375
},
"model.language_model.layers.29.self_attn.o_proj.weight": {
"global_scale": 4327.84912109375
},
"model.language_model.layers.29.self_attn.q_proj.weight": {
"global_scale": 12072.4208984375
},
"model.language_model.layers.29.self_attn.v_proj.weight": {
"global_scale": 19114.666015625
},
"model.language_model.layers.3.mlp.down_proj.weight": {
"global_scale": 2163.924560546875
},
"model.language_model.layers.3.mlp.gate_proj.weight": {
"global_scale": 7775.45751953125
},
"model.language_model.layers.3.mlp.up_proj.weight": {
"global_scale": 5059.7646484375
},
"model.language_model.layers.3.self_attn.k_proj.weight": {
"global_scale": 6847.04443359375
},
"model.language_model.layers.3.self_attn.o_proj.weight": {
"global_scale": 6062.8017578125
},
"model.language_model.layers.3.self_attn.q_proj.weight": {
"global_scale": 10045.6640625
},
"model.language_model.layers.3.self_attn.v_proj.weight": {
"global_scale": 14798.451171875
},
"model.language_model.layers.30.mlp.down_proj.weight": {
"global_scale": 3584.0
},
"model.language_model.layers.30.mlp.gate_proj.weight": {
"global_scale": 5710.60595703125
},
"model.language_model.layers.30.mlp.up_proj.weight": {
"global_scale": 8601.6005859375
},
"model.language_model.layers.30.self_attn.k_proj.weight": {
"global_scale": 5213.0908203125
},
"model.language_model.layers.30.self_attn.o_proj.weight": {
"global_scale": 3602.764404296875
},
"model.language_model.layers.30.self_attn.q_proj.weight": {
"global_scale": 10752.0
},
"model.language_model.layers.30.self_attn.v_proj.weight": {
"global_scale": 11054.2646484375
},
"model.language_model.layers.31.mlp.down_proj.weight": {
"global_scale": 1977.3792724609375
},
"model.language_model.layers.31.mlp.gate_proj.weight": {
"global_scale": 6616.61572265625
},
"model.language_model.layers.31.mlp.up_proj.weight": {
"global_scale": 7205.52880859375
},
"model.language_model.layers.31.self_attn.k_proj.weight": {
"global_scale": 4024.140380859375
},
"model.language_model.layers.31.self_attn.o_proj.weight": {
"global_scale": 3760.26220703125
},
"model.language_model.layers.31.self_attn.q_proj.weight": {
"global_scale": 8095.62353515625
},
"model.language_model.layers.31.self_attn.v_proj.weight": {
"global_scale": 16287.0537109375
},
"model.language_model.layers.32.mlp.down_proj.weight": {
"global_scale": 1988.8092041015625
},
"model.language_model.layers.32.mlp.gate_proj.weight": {
"global_scale": 4880.34033203125
},
"model.language_model.layers.32.mlp.up_proj.weight": {
"global_scale": 6847.04443359375
},
"model.language_model.layers.32.self_attn.k_proj.weight": {
"global_scale": 5376.0
},
"model.language_model.layers.32.self_attn.o_proj.weight": {
"global_scale": 5334.32568359375
},
"model.language_model.layers.32.self_attn.q_proj.weight": {
"global_scale": 5594.5361328125
},
"model.language_model.layers.32.self_attn.v_proj.weight": {
"global_scale": 12862.205078125
},
"model.language_model.layers.33.mlp.down_proj.weight": {
"global_scale": 1746.5177001953125
},
"model.language_model.layers.33.mlp.gate_proj.weight": {
"global_scale": 4527.15771484375
},
"model.language_model.layers.33.mlp.up_proj.weight": {
"global_scale": 5806.9873046875
},
"model.language_model.layers.33.self_attn.k_proj.weight": {
"global_scale": 6062.8017578125
},
"model.language_model.layers.33.self_attn.o_proj.weight": {
"global_scale": 4587.52001953125
},
"model.language_model.layers.33.self_attn.q_proj.weight": {
"global_scale": 7688.58056640625
},
"model.language_model.layers.33.self_attn.v_proj.weight": {
"global_scale": 12568.5478515625
},
"model.language_model.layers.34.mlp.down_proj.weight": {
"global_scale": 2035.8817138671875
},
"model.language_model.layers.34.mlp.gate_proj.weight": {
"global_scale": 5173.89501953125
},
"model.language_model.layers.34.mlp.up_proj.weight": {
"global_scale": 5059.7646484375
},
"model.language_model.layers.34.self_attn.k_proj.weight": {
"global_scale": 4145.34912109375
},
"model.language_model.layers.34.self_attn.o_proj.weight": {
"global_scale": 2293.760009765625
},
"model.language_model.layers.34.self_attn.q_proj.weight": {
"global_scale": 9426.4111328125
},
"model.language_model.layers.34.self_attn.v_proj.weight": {
"global_scale": 10794.1650390625
},
"model.language_model.layers.35.mlp.down_proj.weight": {
"global_scale": 2023.9058837890625
},
"model.language_model.layers.35.mlp.gate_proj.weight": {
"global_scale": 4649.513671875
},
"model.language_model.layers.35.mlp.up_proj.weight": {
"global_scale": 6062.8017578125
},
"model.language_model.layers.35.self_attn.k_proj.weight": {
"global_scale": 4745.71044921875
},
"model.language_model.layers.35.self_attn.o_proj.weight": {
"global_scale": 2234.181884765625
},
"model.language_model.layers.35.self_attn.q_proj.weight": {
"global_scale": 7130.85986328125
},
"model.language_model.layers.35.self_attn.v_proj.weight": {
"global_scale": 14959.3046875
},
"model.language_model.layers.36.mlp.down_proj.weight": {
"global_scale": 1782.7149658203125
},
"model.language_model.layers.36.mlp.gate_proj.weight": {
"global_scale": 5640.39306640625
},
"model.language_model.layers.36.mlp.up_proj.weight": {
"global_scale": 6648.57958984375
},
"model.language_model.layers.36.self_attn.k_proj.weight": {
"global_scale": 5334.32568359375
},
"model.language_model.layers.36.self_attn.o_proj.weight": {
"global_scale": 2309.154296875
},
"model.language_model.layers.36.self_attn.q_proj.weight": {
"global_scale": 8241.0546875
},
"model.language_model.layers.36.self_attn.v_proj.weight": {
"global_scale": 17420.962890625
},
"model.language_model.layers.37.mlp.down_proj.weight": {
"global_scale": 1670.213623046875
},
"model.language_model.layers.37.mlp.gate_proj.weight": {
"global_scale": 4170.47265625
},
"model.language_model.layers.37.mlp.up_proj.weight": {
"global_scale": 2843.50390625
},
"model.language_model.layers.37.self_attn.k_proj.weight": {
"global_scale": 6813.1484375
},
"model.language_model.layers.37.self_attn.o_proj.weight": {
"global_scale": 4382.98095703125
},
"model.language_model.layers.37.self_attn.q_proj.weight": {
"global_scale": 5881.43603515625
},
"model.language_model.layers.37.self_attn.v_proj.weight": {
"global_scale": 14188.2060546875
},
"model.language_model.layers.38.mlp.down_proj.weight": {
"global_scale": 1801.3822021484375
},
"model.language_model.layers.38.mlp.gate_proj.weight": {
"global_scale": 4439.53564453125
},
"model.language_model.layers.38.mlp.up_proj.weight": {
"global_scale": 4618.30859375
},
"model.language_model.layers.38.self_attn.k_proj.weight": {
"global_scale": 6461.2958984375
},
"model.language_model.layers.38.self_attn.o_proj.weight": {
"global_scale": 1515.700439453125
},
"model.language_model.layers.38.self_attn.q_proj.weight": {
"global_scale": 9901.1220703125
},
"model.language_model.layers.38.self_attn.v_proj.weight": {
"global_scale": 15728.6396484375
},
"model.language_model.layers.39.mlp.down_proj.weight": {
"global_scale": 1810.86328125
},
"model.language_model.layers.39.mlp.gate_proj.weight": {
"global_scale": 3640.888671875
},
"model.language_model.layers.39.mlp.up_proj.weight": {
"global_scale": 5293.29248046875
},
"model.language_model.layers.39.self_attn.k_proj.weight": {
"global_scale": 6915.859375
},
"model.language_model.layers.39.self_attn.o_proj.weight": {
"global_scale": 2667.162841796875
},
"model.language_model.layers.39.self_attn.q_proj.weight": {
"global_scale": 7205.52880859375
},
"model.language_model.layers.39.self_attn.v_proj.weight": {
"global_scale": 13626.296875
},
"model.language_model.layers.4.mlp.down_proj.weight": {
"global_scale": 2493.217529296875
},
"model.language_model.layers.4.mlp.gate_proj.weight": {
"global_scale": 7057.72314453125
},
"model.language_model.layers.4.mlp.up_proj.weight": {
"global_scale": 5376.0
},
"model.language_model.layers.4.self_attn.k_proj.weight": {
"global_scale": 7094.10302734375
},
"model.language_model.layers.4.self_attn.o_proj.weight": {
"global_scale": 5831.59326171875
},
"model.language_model.layers.4.self_attn.q_proj.weight": {
"global_scale": 10922.6669921875
},
"model.language_model.layers.4.self_attn.v_proj.weight": {
"global_scale": 16681.890625
},
"model.language_model.layers.40.mlp.down_proj.weight": {
"global_scale": 2110.822021484375
},
"model.language_model.layers.40.mlp.gate_proj.weight": {
"global_scale": 4915.2001953125
},
"model.language_model.layers.40.mlp.up_proj.weight": {
"global_scale": 6144.00048828125
},
"model.language_model.layers.40.self_attn.k_proj.weight": {
"global_scale": 6847.04443359375
},
"model.language_model.layers.40.self_attn.o_proj.weight": {
"global_scale": 2150.400146484375
},
"model.language_model.layers.40.self_attn.q_proj.weight": {
"global_scale": 7731.775390625
},
"model.language_model.layers.40.self_attn.v_proj.weight": {
"global_scale": 15639.2734375
},
"model.language_model.layers.41.mlp.down_proj.weight": {
"global_scale": 1773.5257568359375
},
"model.language_model.layers.41.mlp.gate_proj.weight": {
"global_scale": 3780.923095703125
},
"model.language_model.layers.41.mlp.up_proj.weight": {
"global_scale": 5173.89501953125
},
"model.language_model.layers.41.self_attn.k_proj.weight": {
"global_scale": 6813.1484375
},
"model.language_model.layers.41.self_attn.o_proj.weight": {
"global_scale": 5687.0078125
},
"model.language_model.layers.41.self_attn.q_proj.weight": {
"global_scale": 7168.0
},
"model.language_model.layers.41.self_attn.v_proj.weight": {
"global_scale": 14188.2060546875
},
"model.language_model.layers.42.mlp.down_proj.weight": {
"global_scale": 2191.490478515625
},
"model.language_model.layers.42.mlp.gate_proj.weight": {
"global_scale": 4221.64404296875
},
"model.language_model.layers.42.mlp.up_proj.weight": {
"global_scale": 5173.89501953125
},
"model.language_model.layers.42.self_attn.k_proj.weight": {
"global_scale": 9426.4111328125
},
"model.language_model.layers.42.self_attn.o_proj.weight": {
"global_scale": 3018.105224609375
},
"model.language_model.layers.42.self_attn.q_proj.weight": {
"global_scale": 6371.5556640625
},
"model.language_model.layers.42.self_attn.v_proj.weight": {
"global_scale": 11864.275390625
},
"model.language_model.layers.43.mlp.down_proj.weight": {
"global_scale": 1162.37841796875
},
"model.language_model.layers.43.mlp.gate_proj.weight": {
"global_scale": 4096.0
},
"model.language_model.layers.43.mlp.up_proj.weight": {
"global_scale": 1439.5982666015625
},
"model.language_model.layers.43.self_attn.k_proj.weight": {
"global_scale": 8340.9453125
},
"model.language_model.layers.43.self_attn.o_proj.weight": {
"global_scale": 3185.77783203125
},
"model.language_model.layers.43.self_attn.q_proj.weight": {
"global_scale": 7688.58056640625
},
"model.language_model.layers.43.self_attn.v_proj.weight": {
"global_scale": 8879.0712890625
},
"model.language_model.layers.44.mlp.down_proj.weight": {
"global_scale": 1943.8643798828125
},
"model.language_model.layers.44.mlp.gate_proj.weight": {
"global_scale": 3292.478271484375
},
"model.language_model.layers.44.mlp.up_proj.weight": {
"global_scale": 6491.7734375
},
"model.language_model.layers.44.self_attn.k_proj.weight": {
"global_scale": 6915.859375
},
"model.language_model.layers.44.self_attn.o_proj.weight": {
"global_scale": 2356.602783203125
},
"model.language_model.layers.44.self_attn.q_proj.weight": {
"global_scale": 6713.44384765625
},
"model.language_model.layers.44.self_attn.v_proj.weight": {
"global_scale": 16581.396484375
},
"model.language_model.layers.45.mlp.down_proj.weight": {
"global_scale": 1943.8643798828125
},
"model.language_model.layers.45.mlp.gate_proj.weight": {
"global_scale": 4071.763427734375
},
"model.language_model.layers.45.mlp.up_proj.weight": {
"global_scale": 3200.59521484375
},
"model.language_model.layers.45.self_attn.k_proj.weight": {
"global_scale": 5376.0
},
"model.language_model.layers.45.self_attn.o_proj.weight": {
"global_scale": 2035.8817138671875
},
"model.language_model.layers.45.self_attn.q_proj.weight": {
"global_scale": 7561.84619140625
},
"model.language_model.layers.45.self_attn.v_proj.weight": {
"global_scale": 13972.1416015625
},
"model.language_model.layers.46.mlp.down_proj.weight": {
"global_scale": 1859.805419921875
},
"model.language_model.layers.46.mlp.gate_proj.weight": {
"global_scale": 3719.61083984375
},
"model.language_model.layers.46.mlp.up_proj.weight": {
"global_scale": 5906.67822265625
},
"model.language_model.layers.46.self_attn.k_proj.weight": {
"global_scale": 10794.1650390625
},
"model.language_model.layers.46.self_attn.o_proj.weight": {
"global_scale": 3440.639892578125
},
"model.language_model.layers.46.self_attn.q_proj.weight": {
"global_scale": 7021.71435546875
},
"model.language_model.layers.46.self_attn.v_proj.weight": {
"global_scale": 12019.703125
},
"model.language_model.layers.47.mlp.down_proj.weight": {
"global_scale": 2048.0
},
"model.language_model.layers.47.mlp.gate_proj.weight": {
"global_scale": 4120.52734375
},
"model.language_model.layers.47.mlp.up_proj.weight": {
"global_scale": 5710.60595703125
},
"model.language_model.layers.47.self_attn.k_proj.weight": {
"global_scale": 9624.16796875
},
"model.language_model.layers.47.self_attn.o_proj.weight": {
"global_scale": 5252.8857421875
},
"model.language_model.layers.47.self_attn.q_proj.weight": {
"global_scale": 7021.71435546875
},
"model.language_model.layers.47.self_attn.v_proj.weight": {
"global_scale": 9624.16796875
},
"model.language_model.layers.48.mlp.down_proj.weight": {
"global_scale": 2763.566162109375
},
"model.language_model.layers.48.mlp.gate_proj.weight": {
"global_scale": 3127.8544921875
},
"model.language_model.layers.48.mlp.up_proj.weight": {
"global_scale": 6553.60009765625
},
"model.language_model.layers.48.self_attn.k_proj.weight": {
"global_scale": 9175.0400390625
},
"model.language_model.layers.48.self_attn.o_proj.weight": {
"global_scale": 2548.6220703125
},
"model.language_model.layers.48.self_attn.q_proj.weight": {
"global_scale": 8048.28076171875
},
"model.language_model.layers.48.self_attn.v_proj.weight": {
"global_scale": 11967.443359375
},
"model.language_model.layers.49.mlp.down_proj.weight": {
"global_scale": 2719.87353515625
},
"model.language_model.layers.49.mlp.gate_proj.weight": {
"global_scale": 3142.136962890625
},
"model.language_model.layers.49.mlp.up_proj.weight": {
"global_scale": 5640.39306640625
},
"model.language_model.layers.49.self_attn.k_proj.weight": {
"global_scale": 9760.6806640625
},
"model.language_model.layers.49.self_attn.o_proj.weight": {
"global_scale": 3865.8876953125
},
"model.language_model.layers.49.self_attn.q_proj.weight": {
"global_scale": 10426.181640625
},
"model.language_model.layers.49.self_attn.v_proj.weight": {
"global_scale": 13297.1591796875
},
"model.language_model.layers.5.mlp.down_proj.weight": {
"global_scale": 3389.79296875
},
"model.language_model.layers.5.mlp.gate_proj.weight": {
"global_scale": 7479.65234375
},
"model.language_model.layers.5.mlp.up_proj.weight": {
"global_scale": 4300.80029296875
},
"model.language_model.layers.5.self_attn.k_proj.weight": {
"global_scale": 6950.7880859375
},
"model.language_model.layers.5.self_attn.o_proj.weight": {
"global_scale": 5549.4189453125
},
"model.language_model.layers.5.self_attn.q_proj.weight": {
"global_scale": 11374.015625
},
"model.language_model.layers.5.self_attn.v_proj.weight": {
"global_scale": 14563.5546875
},
"model.language_model.layers.6.mlp.down_proj.weight": {
"global_scale": 1131.7894287109375
},
"model.language_model.layers.6.mlp.gate_proj.weight": {
"global_scale": 6009.8515625
},
"model.language_model.layers.6.mlp.up_proj.weight": {
"global_scale": 3004.92578125
},
"model.language_model.layers.6.self_attn.k_proj.weight": {
"global_scale": 8340.9453125
},
"model.language_model.layers.6.self_attn.o_proj.weight": {
"global_scale": 3932.159912109375
},
"model.language_model.layers.6.self_attn.q_proj.weight": {
"global_scale": 8548.173828125
},
"model.language_model.layers.6.self_attn.v_proj.weight": {
"global_scale": 16096.5615234375
},
"model.language_model.layers.7.mlp.down_proj.weight": {
"global_scale": 2110.822021484375
},
"model.language_model.layers.7.mlp.gate_proj.weight": {
"global_scale": 4986.43505859375
},
"model.language_model.layers.7.mlp.up_proj.weight": {
"global_scale": 2548.6220703125
},
"model.language_model.layers.7.self_attn.k_proj.weight": {
"global_scale": 6401.1904296875
},
"model.language_model.layers.7.self_attn.o_proj.weight": {
"global_scale": 2567.641845703125
},
"model.language_model.layers.7.self_attn.q_proj.weight": {
"global_scale": 9236.6171875
},
"model.language_model.layers.7.self_attn.v_proj.weight": {
"global_scale": 16681.890625
},
"model.language_model.layers.8.mlp.down_proj.weight": {
"global_scale": 2278.569580078125
},
"model.language_model.layers.8.mlp.gate_proj.weight": {
"global_scale": 5983.7216796875
},
"model.language_model.layers.8.mlp.up_proj.weight": {
"global_scale": 9362.28515625
},
"model.language_model.layers.8.self_attn.k_proj.weight": {
"global_scale": 4845.9716796875
},
"model.language_model.layers.8.self_attn.o_proj.weight": {
"global_scale": 3679.828857421875
},
"model.language_model.layers.8.self_attn.q_proj.weight": {
"global_scale": 6779.5859375
},
"model.language_model.layers.8.self_attn.v_proj.weight": {
"global_scale": 13361.708984375
},
"model.language_model.layers.9.mlp.down_proj.weight": {
"global_scale": 1670.213623046875
},
"model.language_model.layers.9.mlp.gate_proj.weight": {
"global_scale": 4411.0771484375
},
"model.language_model.layers.9.mlp.up_proj.weight": {
"global_scale": 6986.07080078125
},
"model.language_model.layers.9.self_attn.k_proj.weight": {
"global_scale": 3932.159912109375
},
"model.language_model.layers.9.self_attn.o_proj.weight": {
"global_scale": 2831.802490234375
},
"model.language_model.layers.9.self_attn.q_proj.weight": {
"global_scale": 9299.02734375
},
"model.language_model.layers.9.self_attn.v_proj.weight": {
"global_scale": 15291.7333984375
}
},
"dense_keys": [
"model.language_model.embed_tokens.weight",
"model.language_model.layers.0.input_layernorm.weight",
"model.language_model.layers.0.post_attention_layernorm.weight",
"model.language_model.layers.0.self_attn.k_norm.weight",
"model.language_model.layers.0.self_attn.q_norm.weight",
"model.language_model.layers.1.input_layernorm.weight",
"model.language_model.layers.1.post_attention_layernorm.weight",
"model.language_model.layers.1.self_attn.k_norm.weight",
"model.language_model.layers.1.self_attn.q_norm.weight",
"model.language_model.layers.10.input_layernorm.weight",
"model.language_model.layers.10.post_attention_layernorm.weight",
"model.language_model.layers.10.self_attn.k_norm.weight",
"model.language_model.layers.10.self_attn.q_norm.weight",
"model.language_model.layers.11.input_layernorm.weight",
"model.language_model.layers.11.post_attention_layernorm.weight",
"model.language_model.layers.11.self_attn.k_norm.weight",
"model.language_model.layers.11.self_attn.q_norm.weight",
"model.language_model.layers.12.input_layernorm.weight",
"model.language_model.layers.12.post_attention_layernorm.weight",
"model.language_model.layers.12.self_attn.k_norm.weight",
"model.language_model.layers.12.self_attn.q_norm.weight",
"model.language_model.layers.13.input_layernorm.weight",
"model.language_model.layers.13.post_attention_layernorm.weight",
"model.language_model.layers.13.self_attn.k_norm.weight",
"model.language_model.layers.13.self_attn.q_norm.weight",
"model.language_model.layers.14.input_layernorm.weight",
"model.language_model.layers.14.post_attention_layernorm.weight",
"model.language_model.layers.14.self_attn.k_norm.weight",
"model.language_model.layers.14.self_attn.q_norm.weight",
"model.language_model.layers.15.input_layernorm.weight",
"model.language_model.layers.15.post_attention_layernorm.weight",
"model.language_model.layers.15.self_attn.k_norm.weight",
"model.language_model.layers.15.self_attn.q_norm.weight",
"model.language_model.layers.16.input_layernorm.weight",
"model.language_model.layers.16.post_attention_layernorm.weight",
"model.language_model.layers.16.self_attn.k_norm.weight",
"model.language_model.layers.16.self_attn.q_norm.weight",
"model.language_model.layers.17.input_layernorm.weight",
"model.language_model.layers.17.post_attention_layernorm.weight",
"model.language_model.layers.17.self_attn.k_norm.weight",
"model.language_model.layers.17.self_attn.q_norm.weight",
"model.language_model.layers.18.input_layernorm.weight",
"model.language_model.layers.18.post_attention_layernorm.weight",
"model.language_model.layers.18.self_attn.k_norm.weight",
"model.language_model.layers.18.self_attn.q_norm.weight",
"model.language_model.layers.19.input_layernorm.weight",
"model.language_model.layers.19.post_attention_layernorm.weight",
"model.language_model.layers.19.self_attn.k_norm.weight",
"model.language_model.layers.19.self_attn.q_norm.weight",
"model.language_model.layers.2.input_layernorm.weight",
"model.language_model.layers.2.post_attention_layernorm.weight",
"model.language_model.layers.2.self_attn.k_norm.weight",
"model.language_model.layers.2.self_attn.q_norm.weight",
"model.language_model.layers.20.input_layernorm.weight",
"model.language_model.layers.20.post_attention_layernorm.weight",
"model.language_model.layers.20.self_attn.k_norm.weight",
"model.language_model.layers.20.self_attn.q_norm.weight",
"model.language_model.layers.21.input_layernorm.weight",
"model.language_model.layers.21.post_attention_layernorm.weight",
"model.language_model.layers.21.self_attn.k_norm.weight",
"model.language_model.layers.21.self_attn.q_norm.weight",
"model.language_model.layers.22.input_layernorm.weight",
"model.language_model.layers.22.post_attention_layernorm.weight",
"model.language_model.layers.22.self_attn.k_norm.weight",
"model.language_model.layers.22.self_attn.q_norm.weight",
"model.language_model.layers.23.input_layernorm.weight",
"model.language_model.layers.23.post_attention_layernorm.weight",
"model.language_model.layers.23.self_attn.k_norm.weight",
"model.language_model.layers.23.self_attn.q_norm.weight",
"model.language_model.layers.24.input_layernorm.weight",
"model.language_model.layers.24.post_attention_layernorm.weight",
"model.language_model.layers.24.self_attn.k_norm.weight",
"model.language_model.layers.24.self_attn.q_norm.weight",
"model.language_model.layers.25.input_layernorm.weight",
"model.language_model.layers.25.post_attention_layernorm.weight",
"model.language_model.layers.25.self_attn.k_norm.weight",
"model.language_model.layers.25.self_attn.q_norm.weight",
"model.language_model.layers.26.input_layernorm.weight",
"model.language_model.layers.26.post_attention_layernorm.weight",
"model.language_model.layers.26.self_attn.k_norm.weight",
"model.language_model.layers.26.self_attn.q_norm.weight",
"model.language_model.layers.27.input_layernorm.weight",
"model.language_model.layers.27.post_attention_layernorm.weight",
"model.language_model.layers.27.self_attn.k_norm.weight",
"model.language_model.layers.27.self_attn.q_norm.weight",
"model.language_model.layers.28.input_layernorm.weight",
"model.language_model.layers.28.post_attention_layernorm.weight",
"model.language_model.layers.28.self_attn.k_norm.weight",
"model.language_model.layers.28.self_attn.q_norm.weight",
"model.language_model.layers.29.input_layernorm.weight",
"model.language_model.layers.29.post_attention_layernorm.weight",
"model.language_model.layers.29.self_attn.k_norm.weight",
"model.language_model.layers.29.self_attn.q_norm.weight",
"model.language_model.layers.3.input_layernorm.weight",
"model.language_model.layers.3.post_attention_layernorm.weight",
"model.language_model.layers.3.self_attn.k_norm.weight",
"model.language_model.layers.3.self_attn.q_norm.weight",
"model.language_model.layers.30.input_layernorm.weight",
"model.language_model.layers.30.post_attention_layernorm.weight",
"model.language_model.layers.30.self_attn.k_norm.weight",
"model.language_model.layers.30.self_attn.q_norm.weight",
"model.language_model.layers.31.input_layernorm.weight",
"model.language_model.layers.31.post_attention_layernorm.weight",
"model.language_model.layers.31.self_attn.k_norm.weight",
"model.language_model.layers.31.self_attn.q_norm.weight",
"model.language_model.layers.32.input_layernorm.weight",
"model.language_model.layers.32.post_attention_layernorm.weight",
"model.language_model.layers.32.self_attn.k_norm.weight",
"model.language_model.layers.32.self_attn.q_norm.weight",
"model.language_model.layers.33.input_layernorm.weight",
"model.language_model.layers.33.post_attention_layernorm.weight",
"model.language_model.layers.33.self_attn.k_norm.weight",
"model.language_model.layers.33.self_attn.q_norm.weight",
"model.language_model.layers.34.input_layernorm.weight",
"model.language_model.layers.34.post_attention_layernorm.weight",
"model.language_model.layers.34.self_attn.k_norm.weight",
"model.language_model.layers.34.self_attn.q_norm.weight",
"model.language_model.layers.35.input_layernorm.weight",
"model.language_model.layers.35.post_attention_layernorm.weight",
"model.language_model.layers.35.self_attn.k_norm.weight",
"model.language_model.layers.35.self_attn.q_norm.weight",
"model.language_model.layers.36.input_layernorm.weight",
"model.language_model.layers.36.post_attention_layernorm.weight",
"model.language_model.layers.36.self_attn.k_norm.weight",
"model.language_model.layers.36.self_attn.q_norm.weight",
"model.language_model.layers.37.input_layernorm.weight",
"model.language_model.layers.37.post_attention_layernorm.weight",
"model.language_model.layers.37.self_attn.k_norm.weight",
"model.language_model.layers.37.self_attn.q_norm.weight",
"model.language_model.layers.38.input_layernorm.weight",
"model.language_model.layers.38.post_attention_layernorm.weight",
"model.language_model.layers.38.self_attn.k_norm.weight",
"model.language_model.layers.38.self_attn.q_norm.weight",
"model.language_model.layers.39.input_layernorm.weight",
"model.language_model.layers.39.post_attention_layernorm.weight",
"model.language_model.layers.39.self_attn.k_norm.weight",
"model.language_model.layers.39.self_attn.q_norm.weight",
"model.language_model.layers.4.input_layernorm.weight",
"model.language_model.layers.4.post_attention_layernorm.weight",
"model.language_model.layers.4.self_attn.k_norm.weight",
"model.language_model.layers.4.self_attn.q_norm.weight",
"model.language_model.layers.40.input_layernorm.weight",
"model.language_model.layers.40.post_attention_layernorm.weight",
"model.language_model.layers.40.self_attn.k_norm.weight",
"model.language_model.layers.40.self_attn.q_norm.weight",
"model.language_model.layers.41.input_layernorm.weight",
"model.language_model.layers.41.post_attention_layernorm.weight",
"model.language_model.layers.41.self_attn.k_norm.weight",
"model.language_model.layers.41.self_attn.q_norm.weight",
"model.language_model.layers.42.input_layernorm.weight",
"model.language_model.layers.42.post_attention_layernorm.weight",
"model.language_model.layers.42.self_attn.k_norm.weight",
"model.language_model.layers.42.self_attn.q_norm.weight",
"model.language_model.layers.43.input_layernorm.weight",
"model.language_model.layers.43.post_attention_layernorm.weight",
"model.language_model.layers.43.self_attn.k_norm.weight",
"model.language_model.layers.43.self_attn.q_norm.weight",
"model.language_model.layers.44.input_layernorm.weight",
"model.language_model.layers.44.post_attention_layernorm.weight",
"model.language_model.layers.44.self_attn.k_norm.weight",
"model.language_model.layers.44.self_attn.q_norm.weight",
"model.language_model.layers.45.input_layernorm.weight",
"model.language_model.layers.45.post_attention_layernorm.weight",
"model.language_model.layers.45.self_attn.k_norm.weight",
"model.language_model.layers.45.self_attn.q_norm.weight",
"model.language_model.layers.46.input_layernorm.weight",
"model.language_model.layers.46.post_attention_layernorm.weight",
"model.language_model.layers.46.self_attn.k_norm.weight",
"model.language_model.layers.46.self_attn.q_norm.weight",
"model.language_model.layers.47.input_layernorm.weight",
"model.language_model.layers.47.post_attention_layernorm.weight",
"model.language_model.layers.47.self_attn.k_norm.weight",
"model.language_model.layers.47.self_attn.q_norm.weight",
"model.language_model.layers.48.input_layernorm.weight",
"model.language_model.layers.48.post_attention_layernorm.weight",
"model.language_model.layers.48.self_attn.k_norm.weight",
"model.language_model.layers.48.self_attn.q_norm.weight",
"model.language_model.layers.49.input_layernorm.weight",
"model.language_model.layers.49.post_attention_layernorm.weight",
"model.language_model.layers.49.self_attn.k_norm.weight",
"model.language_model.layers.49.self_attn.q_norm.weight",
"model.language_model.layers.5.input_layernorm.weight",
"model.language_model.layers.5.post_attention_layernorm.weight",
"model.language_model.layers.5.self_attn.k_norm.weight",
"model.language_model.layers.5.self_attn.q_norm.weight",
"model.language_model.layers.6.input_layernorm.weight",
"model.language_model.layers.6.post_attention_layernorm.weight",
"model.language_model.layers.6.self_attn.k_norm.weight",
"model.language_model.layers.6.self_attn.q_norm.weight",
"model.language_model.layers.7.input_layernorm.weight",
"model.language_model.layers.7.post_attention_layernorm.weight",
"model.language_model.layers.7.self_attn.k_norm.weight",
"model.language_model.layers.7.self_attn.q_norm.weight",
"model.language_model.layers.8.input_layernorm.weight",
"model.language_model.layers.8.post_attention_layernorm.weight",
"model.language_model.layers.8.self_attn.k_norm.weight",
"model.language_model.layers.8.self_attn.q_norm.weight",
"model.language_model.layers.9.input_layernorm.weight",
"model.language_model.layers.9.post_attention_layernorm.weight",
"model.language_model.layers.9.self_attn.k_norm.weight",
"model.language_model.layers.9.self_attn.q_norm.weight"
],
"source_dir": "/Users/aryank/os/FastH3-NVFP4-encoder"
}