| { | |
| "original_model_path": "C:\\Users\\drbaph\\Documents\\ComfyUI\\models\\breezetts2\\drbaph_Breeze-TTS-2-comfyui\\model.safetensors", | |
| "quantized_model_path": "C:\\Users\\drbaph\\Documents\\ComfyUI\\models\\breezetts2\\drbaph_Breeze-TTS-2-comfyui\\model-int8-text-encoder.safetensors", | |
| "comfy_kitchen_version": "0.2.31", | |
| "pytorch_version": "2.10.0+cu130", | |
| "cuda_version": "13.0", | |
| "method": "comfy_kitchen TensorWiseINT8Layout.quantize(is_weight=True, per_channel=True, convrot=True)", | |
| "quantized_tensors": 182, | |
| "quantized_params": 697761792, | |
| "unquantized": "backbone_model.layers.*, depth_decoder.model.layers.* (all decode-time linears stay bf16)", | |
| "original_size_bytes": 6966557274, | |
| "output_size_bytes": 6270699712, | |
| "elapsed_seconds": 6.4934158000396565, | |
| "worst_rel_l2": 0.009073183871805668, | |
| "median_rel_l2": 0.007955125533044338, | |
| "worst_cosine": 0.999958872795105, | |
| "per_tensor": [ | |
| { | |
| "tensor": "text_encoder.layers.0.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00020656641572713852, | |
| "mean_abs_err": 2.797946217469871e-05, | |
| "rel_l2": 0.008970697410404682, | |
| "cosine": 0.9999597668647766 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.0.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00023933500051498413, | |
| "mean_abs_err": 2.4346587451873347e-05, | |
| "rel_l2": 0.007968038320541382, | |
| "cosine": 0.9999681711196899 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.0.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00019369833171367645, | |
| "mean_abs_err": 2.592776218079962e-05, | |
| "rel_l2": 0.007964438758790493, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.0.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026987865567207336, | |
| "mean_abs_err": 3.885104524670169e-05, | |
| "rel_l2": 0.007972605526447296, | |
| "cosine": 0.9999682307243347 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.0.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.000321744941174984, | |
| "mean_abs_err": 4.160729076829739e-05, | |
| "rel_l2": 0.008121548220515251, | |
| "cosine": 0.9999669790267944 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.0.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00032746046781539917, | |
| "mean_abs_err": 3.8418460462708026e-05, | |
| "rel_l2": 0.007994755171239376, | |
| "cosine": 0.9999680519104004 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.0.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00027776695787906647, | |
| "mean_abs_err": 4.1382445488125086e-05, | |
| "rel_l2": 0.008023402653634548, | |
| "cosine": 0.9999678134918213 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0002707187086343765, | |
| "mean_abs_err": 2.859158303181175e-05, | |
| "rel_l2": 0.008981716819107533, | |
| "cosine": 0.9999595880508423 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00020155683159828186, | |
| "mean_abs_err": 2.3744967620586976e-05, | |
| "rel_l2": 0.007947574369609356, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002052774652838707, | |
| "mean_abs_err": 2.4912131266319193e-05, | |
| "rel_l2": 0.007934706285595894, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00022666295990347862, | |
| "mean_abs_err": 3.6654055293183774e-05, | |
| "rel_l2": 0.00790654867887497, | |
| "cosine": 0.9999688267707825 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00041987933218479156, | |
| "mean_abs_err": 4.223711584927514e-05, | |
| "rel_l2": 0.008434656076133251, | |
| "cosine": 0.9999644756317139 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004361348692327738, | |
| "mean_abs_err": 3.88821390515659e-05, | |
| "rel_l2": 0.007985027506947517, | |
| "cosine": 0.9999681711196899 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.1.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024056341499090195, | |
| "mean_abs_err": 3.949280653614551e-05, | |
| "rel_l2": 0.00797144416719675, | |
| "cosine": 0.9999682307243347 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0004994622431695461, | |
| "mean_abs_err": 2.8997241315664724e-05, | |
| "rel_l2": 0.008962404914200306, | |
| "cosine": 0.9999599456787109 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00022587087005376816, | |
| "mean_abs_err": 2.7649817639030516e-05, | |
| "rel_l2": 0.007903611287474632, | |
| "cosine": 0.9999687671661377 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024119787849485874, | |
| "mean_abs_err": 2.419894553895574e-05, | |
| "rel_l2": 0.007940232753753662, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00032904092222452164, | |
| "mean_abs_err": 3.916457717423327e-05, | |
| "rel_l2": 0.00793897733092308, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005854489281773567, | |
| "mean_abs_err": 4.4737575080944225e-05, | |
| "rel_l2": 0.008155359886586666, | |
| "cosine": 0.9999668002128601 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000441916286945343, | |
| "mean_abs_err": 4.080178041476756e-05, | |
| "rel_l2": 0.00792653113603592, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.10.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002703205682337284, | |
| "mean_abs_err": 4.131305468035862e-05, | |
| "rel_l2": 0.007881417870521545, | |
| "cosine": 0.9999690055847168 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005869949236512184, | |
| "mean_abs_err": 3.110179750365205e-05, | |
| "rel_l2": 0.00893147848546505, | |
| "cosine": 0.99996018409729 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000225776806473732, | |
| "mean_abs_err": 2.7423178835306317e-05, | |
| "rel_l2": 0.007921411655843258, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024015316739678383, | |
| "mean_abs_err": 2.6978041205438785e-05, | |
| "rel_l2": 0.007935961708426476, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00040703057311475277, | |
| "mean_abs_err": 3.809263216680847e-05, | |
| "rel_l2": 0.008052492514252663, | |
| "cosine": 0.9999675750732422 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0007184594869613647, | |
| "mean_abs_err": 4.587099829223007e-05, | |
| "rel_l2": 0.008229357190430164, | |
| "cosine": 0.9999661445617676 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00045102741569280624, | |
| "mean_abs_err": 4.250429992680438e-05, | |
| "rel_l2": 0.00794568657875061, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.11.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026520818937569857, | |
| "mean_abs_err": 4.3999643821734935e-05, | |
| "rel_l2": 0.007959697395563126, | |
| "cosine": 0.9999683499336243 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0007534623146057129, | |
| "mean_abs_err": 2.9220409487606958e-05, | |
| "rel_l2": 0.009023528546094894, | |
| "cosine": 0.9999592900276184 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024057773407548666, | |
| "mean_abs_err": 2.772843618004117e-05, | |
| "rel_l2": 0.007928523235023022, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00021462119184434414, | |
| "mean_abs_err": 2.567176125012338e-05, | |
| "rel_l2": 0.007913559675216675, | |
| "cosine": 0.9999687075614929 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003492487594485283, | |
| "mean_abs_err": 3.951948383473791e-05, | |
| "rel_l2": 0.00808818731456995, | |
| "cosine": 0.9999673366546631 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0004827156662940979, | |
| "mean_abs_err": 4.38933348050341e-05, | |
| "rel_l2": 0.008042722940444946, | |
| "cosine": 0.9999676942825317 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00035160500556230545, | |
| "mean_abs_err": 4.113037357456051e-05, | |
| "rel_l2": 0.007980970665812492, | |
| "cosine": 0.9999681711196899 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.12.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002672271803021431, | |
| "mean_abs_err": 4.229570913594216e-05, | |
| "rel_l2": 0.007950587198138237, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00034677982330322266, | |
| "mean_abs_err": 2.9670321964658797e-05, | |
| "rel_l2": 0.008940483443439007, | |
| "cosine": 0.9999600052833557 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00023941253311932087, | |
| "mean_abs_err": 2.6375944798928685e-05, | |
| "rel_l2": 0.007955878973007202, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00020468980073928833, | |
| "mean_abs_err": 2.5755580281838775e-05, | |
| "rel_l2": 0.007935441099107265, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000318639911711216, | |
| "mean_abs_err": 4.031285425298847e-05, | |
| "rel_l2": 0.007978199981153011, | |
| "cosine": 0.9999681711196899 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005072522908449173, | |
| "mean_abs_err": 4.5973491069162264e-05, | |
| "rel_l2": 0.008308728225529194, | |
| "cosine": 0.999965488910675 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004725456237792969, | |
| "mean_abs_err": 4.243482180754654e-05, | |
| "rel_l2": 0.00795433297753334, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.13.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00029289652593433857, | |
| "mean_abs_err": 4.36946538684424e-05, | |
| "rel_l2": 0.007921375334262848, | |
| "cosine": 0.9999687075614929 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006862860172986984, | |
| "mean_abs_err": 3.043674041691702e-05, | |
| "rel_l2": 0.008945777080953121, | |
| "cosine": 0.9999599456787109 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002442979603074491, | |
| "mean_abs_err": 2.7715392207028344e-05, | |
| "rel_l2": 0.007929785177111626, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00020622694864869118, | |
| "mean_abs_err": 2.670797402970493e-05, | |
| "rel_l2": 0.007926554419100285, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000349557027220726, | |
| "mean_abs_err": 4.0669183363206685e-05, | |
| "rel_l2": 0.007911005057394505, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006493804976344109, | |
| "mean_abs_err": 4.713579619419761e-05, | |
| "rel_l2": 0.008153656497597694, | |
| "cosine": 0.9999667406082153 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004902724176645279, | |
| "mean_abs_err": 4.406982407090254e-05, | |
| "rel_l2": 0.007987537421286106, | |
| "cosine": 0.9999681115150452 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.14.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002967715263366699, | |
| "mean_abs_err": 4.568686199490912e-05, | |
| "rel_l2": 0.007957719266414642, | |
| "cosine": 0.9999683499336243 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005684010684490204, | |
| "mean_abs_err": 3.092829865636304e-05, | |
| "rel_l2": 0.008936441503465176, | |
| "cosine": 0.9999601244926453 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002698830794543028, | |
| "mean_abs_err": 2.9062468456686474e-05, | |
| "rel_l2": 0.007948662154376507, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002498086541891098, | |
| "mean_abs_err": 2.774826680251863e-05, | |
| "rel_l2": 0.007927238009870052, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00027378741651773453, | |
| "mean_abs_err": 4.029141928185709e-05, | |
| "rel_l2": 0.007906126789748669, | |
| "cosine": 0.9999687671661377 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00045468308962881565, | |
| "mean_abs_err": 4.5985823817318305e-05, | |
| "rel_l2": 0.008228728547692299, | |
| "cosine": 0.9999662637710571 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0005461055552586913, | |
| "mean_abs_err": 4.145556886214763e-05, | |
| "rel_l2": 0.007897666655480862, | |
| "cosine": 0.9999688863754272 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.15.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002879721578210592, | |
| "mean_abs_err": 4.496155816013925e-05, | |
| "rel_l2": 0.007910206913948059, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006656097248196602, | |
| "mean_abs_err": 3.11288058583159e-05, | |
| "rel_l2": 0.008936969563364983, | |
| "cosine": 0.9999601244926453 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002878764644265175, | |
| "mean_abs_err": 2.9656692277058028e-05, | |
| "rel_l2": 0.007939698174595833, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024717487394809723, | |
| "mean_abs_err": 2.8428970836102962e-05, | |
| "rel_l2": 0.007927916944026947, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003598378971219063, | |
| "mean_abs_err": 4.066798646817915e-05, | |
| "rel_l2": 0.007962486706674099, | |
| "cosine": 0.9999683499336243 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005106199532747269, | |
| "mean_abs_err": 4.5656950533157215e-05, | |
| "rel_l2": 0.008158673532307148, | |
| "cosine": 0.9999667406082153 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00045864470303058624, | |
| "mean_abs_err": 4.319646177464165e-05, | |
| "rel_l2": 0.007907072082161903, | |
| "cosine": 0.9999687671661377 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.16.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00027803145349025726, | |
| "mean_abs_err": 4.6640219807159156e-05, | |
| "rel_l2": 0.007905792444944382, | |
| "cosine": 0.9999687075614929 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006273156031966209, | |
| "mean_abs_err": 3.232168819522485e-05, | |
| "rel_l2": 0.008961471728980541, | |
| "cosine": 0.9999598860740662 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026919133961200714, | |
| "mean_abs_err": 3.183860098943114e-05, | |
| "rel_l2": 0.00795331597328186, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002784421667456627, | |
| "mean_abs_err": 2.979859164042864e-05, | |
| "rel_l2": 0.007923610508441925, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000310712493956089, | |
| "mean_abs_err": 4.081440420122817e-05, | |
| "rel_l2": 0.007954413071274757, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006721913814544678, | |
| "mean_abs_err": 4.759236253448762e-05, | |
| "rel_l2": 0.00829317793250084, | |
| "cosine": 0.9999656677246094 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0005942909047007561, | |
| "mean_abs_err": 4.918796548736282e-05, | |
| "rel_l2": 0.007860478013753891, | |
| "cosine": 0.9999690651893616 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.17.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00033246632665395737, | |
| "mean_abs_err": 5.000580131309107e-05, | |
| "rel_l2": 0.007866171188652515, | |
| "cosine": 0.9999691247940063 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005804561078548431, | |
| "mean_abs_err": 3.215258402633481e-05, | |
| "rel_l2": 0.00896071083843708, | |
| "cosine": 0.9999598860740662 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002619265578687191, | |
| "mean_abs_err": 3.0102950404398143e-05, | |
| "rel_l2": 0.007947666570544243, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00022509577684104443, | |
| "mean_abs_err": 2.8918775569763966e-05, | |
| "rel_l2": 0.007940365001559258, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000308517599478364, | |
| "mean_abs_err": 4.346207424532622e-05, | |
| "rel_l2": 0.008007669821381569, | |
| "cosine": 0.9999679327011108 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0009103007614612579, | |
| "mean_abs_err": 4.8640500608598813e-05, | |
| "rel_l2": 0.008433609269559383, | |
| "cosine": 0.9999645352363586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0006566569209098816, | |
| "mean_abs_err": 4.761685340781696e-05, | |
| "rel_l2": 0.007961737923324108, | |
| "cosine": 0.9999681711196899 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.18.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003177933394908905, | |
| "mean_abs_err": 4.904352317680605e-05, | |
| "rel_l2": 0.007937918417155743, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006463476456701756, | |
| "mean_abs_err": 3.214738535461947e-05, | |
| "rel_l2": 0.008960945531725883, | |
| "cosine": 0.9999598264694214 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024938315618783236, | |
| "mean_abs_err": 3.0850747862132266e-05, | |
| "rel_l2": 0.00792153924703598, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002099168486893177, | |
| "mean_abs_err": 2.8655553251155652e-05, | |
| "rel_l2": 0.007940316572785378, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003219459322281182, | |
| "mean_abs_err": 4.42723321611993e-05, | |
| "rel_l2": 0.007933183573186398, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0008521135896444321, | |
| "mean_abs_err": 4.736207483801991e-05, | |
| "rel_l2": 0.009073183871805668, | |
| "cosine": 0.999958872795105 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0005671344697475433, | |
| "mean_abs_err": 5.0584625569172204e-05, | |
| "rel_l2": 0.007917975075542927, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.19.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002674043644219637, | |
| "mean_abs_err": 4.329692092142068e-05, | |
| "rel_l2": 0.007957505993545055, | |
| "cosine": 0.9999683499336243 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005183652974665165, | |
| "mean_abs_err": 2.818369830492884e-05, | |
| "rel_l2": 0.008924358524382114, | |
| "cosine": 0.99996018409729 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00022553279995918274, | |
| "mean_abs_err": 2.602654421934858e-05, | |
| "rel_l2": 0.007933931425213814, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00020707445219159126, | |
| "mean_abs_err": 2.5388018912053667e-05, | |
| "rel_l2": 0.007938358001410961, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00031521497294306755, | |
| "mean_abs_err": 3.5276400012662634e-05, | |
| "rel_l2": 0.008057917468249798, | |
| "cosine": 0.9999675154685974 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0003803437575697899, | |
| "mean_abs_err": 4.029880074085668e-05, | |
| "rel_l2": 0.008644454181194305, | |
| "cosine": 0.999962568283081 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00039730523712933064, | |
| "mean_abs_err": 3.8624624721705914e-05, | |
| "rel_l2": 0.008003915660083294, | |
| "cosine": 0.9999679327011108 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.2.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002723131328821182, | |
| "mean_abs_err": 3.8944253901718184e-05, | |
| "rel_l2": 0.008008047007024288, | |
| "cosine": 0.9999679923057556 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006491672247648239, | |
| "mean_abs_err": 3.229799767723307e-05, | |
| "rel_l2": 0.008937758393585682, | |
| "cosine": 0.9999600648880005 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00027403607964515686, | |
| "mean_abs_err": 3.074506821576506e-05, | |
| "rel_l2": 0.00795432273298502, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026377150788903236, | |
| "mean_abs_err": 2.891800249926746e-05, | |
| "rel_l2": 0.007955869659781456, | |
| "cosine": 0.9999683499336243 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003809090703725815, | |
| "mean_abs_err": 4.326660928199999e-05, | |
| "rel_l2": 0.007906608283519745, | |
| "cosine": 0.9999687075614929 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0008876696228981018, | |
| "mean_abs_err": 4.7957481001503766e-05, | |
| "rel_l2": 0.008227583952248096, | |
| "cosine": 0.9999661445617676 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0005388744175434113, | |
| "mean_abs_err": 4.741094380733557e-05, | |
| "rel_l2": 0.00797293335199356, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.20.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00030053011141717434, | |
| "mean_abs_err": 5.060202965978533e-05, | |
| "rel_l2": 0.00785363744944334, | |
| "cosine": 0.9999691247940063 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005132327787578106, | |
| "mean_abs_err": 3.251562884543091e-05, | |
| "rel_l2": 0.00896418560296297, | |
| "cosine": 0.9999598264694214 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003187693655490875, | |
| "mean_abs_err": 3.0183091439539567e-05, | |
| "rel_l2": 0.007937651127576828, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00021067215129733086, | |
| "mean_abs_err": 2.896980004152283e-05, | |
| "rel_l2": 0.007931210100650787, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000399609562009573, | |
| "mean_abs_err": 4.487497790250927e-05, | |
| "rel_l2": 0.00788166094571352, | |
| "cosine": 0.9999690055847168 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0008702920749783516, | |
| "mean_abs_err": 4.8526617320021614e-05, | |
| "rel_l2": 0.008211186155676842, | |
| "cosine": 0.9999662637710571 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0006546098738908768, | |
| "mean_abs_err": 4.890083073405549e-05, | |
| "rel_l2": 0.007904332131147385, | |
| "cosine": 0.9999687671661377 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.21.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003481737803667784, | |
| "mean_abs_err": 5.261986734694801e-05, | |
| "rel_l2": 0.007975711487233639, | |
| "cosine": 0.9999680519104004 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005333591252565384, | |
| "mean_abs_err": 3.2385069062002e-05, | |
| "rel_l2": 0.008977968245744705, | |
| "cosine": 0.9999597072601318 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003626309335231781, | |
| "mean_abs_err": 2.9717686629737727e-05, | |
| "rel_l2": 0.007956316694617271, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002301456406712532, | |
| "mean_abs_err": 2.9052991521894e-05, | |
| "rel_l2": 0.007950742729008198, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004382636398077011, | |
| "mean_abs_err": 4.31239532190375e-05, | |
| "rel_l2": 0.00788606982678175, | |
| "cosine": 0.9999690055847168 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.000702960416674614, | |
| "mean_abs_err": 4.8429985326947644e-05, | |
| "rel_l2": 0.008221814408898354, | |
| "cosine": 0.9999662041664124 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0006164312362670898, | |
| "mean_abs_err": 5.0034901505568996e-05, | |
| "rel_l2": 0.007989372126758099, | |
| "cosine": 0.9999681115150452 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.22.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00032069883309304714, | |
| "mean_abs_err": 5.201617022976279e-05, | |
| "rel_l2": 0.007993162609636784, | |
| "cosine": 0.9999681115150452 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006640143692493439, | |
| "mean_abs_err": 3.273362744948827e-05, | |
| "rel_l2": 0.008953742682933807, | |
| "cosine": 0.9999598860740662 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003001564182341099, | |
| "mean_abs_err": 3.0143761250656098e-05, | |
| "rel_l2": 0.007961604744195938, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002369892317801714, | |
| "mean_abs_err": 2.888010931201279e-05, | |
| "rel_l2": 0.007948333397507668, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004116203635931015, | |
| "mean_abs_err": 4.399473255034536e-05, | |
| "rel_l2": 0.007933514192700386, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0008253478445112705, | |
| "mean_abs_err": 5.497054007719271e-05, | |
| "rel_l2": 0.008845708332955837, | |
| "cosine": 0.9999608397483826 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000773110892623663, | |
| "mean_abs_err": 5.248463276075199e-05, | |
| "rel_l2": 0.007877490483224392, | |
| "cosine": 0.9999690055847168 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.23.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00033188890665769577, | |
| "mean_abs_err": 5.1554983656387776e-05, | |
| "rel_l2": 0.007833216339349747, | |
| "cosine": 0.9999693036079407 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005075366934761405, | |
| "mean_abs_err": 3.2519499654881656e-05, | |
| "rel_l2": 0.008956735953688622, | |
| "cosine": 0.9999598264694214 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00036054104566574097, | |
| "mean_abs_err": 2.9588718462036923e-05, | |
| "rel_l2": 0.007940402254462242, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024491280782967806, | |
| "mean_abs_err": 2.889705683628563e-05, | |
| "rel_l2": 0.007936762645840645, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00033783912658691406, | |
| "mean_abs_err": 4.379333768156357e-05, | |
| "rel_l2": 0.00795164704322815, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006743166595697403, | |
| "mean_abs_err": 4.934396565658972e-05, | |
| "rel_l2": 0.008342010900378227, | |
| "cosine": 0.9999651908874512 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0006714193150401115, | |
| "mean_abs_err": 4.97017172165215e-05, | |
| "rel_l2": 0.007908848114311695, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.24.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003778403624892235, | |
| "mean_abs_err": 5.1999606512254104e-05, | |
| "rel_l2": 0.007967795245349407, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00048340484499931335, | |
| "mean_abs_err": 3.3497504773549736e-05, | |
| "rel_l2": 0.009017049334943295, | |
| "cosine": 0.9999593496322632 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00032267882488667965, | |
| "mean_abs_err": 2.9604585506604053e-05, | |
| "rel_l2": 0.007950881496071815, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002512042410671711, | |
| "mean_abs_err": 2.9527474907808937e-05, | |
| "rel_l2": 0.007959000766277313, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004110862500965595, | |
| "mean_abs_err": 4.4241918658372015e-05, | |
| "rel_l2": 0.008024119772017002, | |
| "cosine": 0.9999677538871765 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0007032370194792747, | |
| "mean_abs_err": 5.101178612676449e-05, | |
| "rel_l2": 0.008469551801681519, | |
| "cosine": 0.9999642372131348 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0007719739805907011, | |
| "mean_abs_err": 5.182607492315583e-05, | |
| "rel_l2": 0.0079177375882864, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.25.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00033121462911367416, | |
| "mean_abs_err": 5.0364051276119426e-05, | |
| "rel_l2": 0.007952168583869934, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00045041926205158234, | |
| "mean_abs_err": 2.776979272312019e-05, | |
| "rel_l2": 0.009012777358293533, | |
| "cosine": 0.9999593496322632 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00019633164629340172, | |
| "mean_abs_err": 2.515634332667105e-05, | |
| "rel_l2": 0.007925971411168575, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00019867811352014542, | |
| "mean_abs_err": 2.415253584331367e-05, | |
| "rel_l2": 0.007924104109406471, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002787295961752534, | |
| "mean_abs_err": 3.568042302504182e-05, | |
| "rel_l2": 0.007939119823276997, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005836784839630127, | |
| "mean_abs_err": 4.3145690142409876e-05, | |
| "rel_l2": 0.00878596119582653, | |
| "cosine": 0.9999613761901855 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003864225000143051, | |
| "mean_abs_err": 3.779773396672681e-05, | |
| "rel_l2": 0.00792684406042099, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.3.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00025805551558732986, | |
| "mean_abs_err": 4.1465806134510785e-05, | |
| "rel_l2": 0.00786999799311161, | |
| "cosine": 0.9999690055847168 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0003335271030664444, | |
| "mean_abs_err": 2.8222735636518337e-05, | |
| "rel_l2": 0.008878986351191998, | |
| "cosine": 0.9999604821205139 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000200013630092144, | |
| "mean_abs_err": 2.624926491989754e-05, | |
| "rel_l2": 0.00792849063873291, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0001984627451747656, | |
| "mean_abs_err": 2.4439248591079377e-05, | |
| "rel_l2": 0.00793036725372076, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002502409042790532, | |
| "mean_abs_err": 3.472494427114725e-05, | |
| "rel_l2": 0.007939684204757214, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0006269481964409351, | |
| "mean_abs_err": 4.2735569877550006e-05, | |
| "rel_l2": 0.008397187106311321, | |
| "cosine": 0.9999647736549377 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00034347735345363617, | |
| "mean_abs_err": 3.841929355985485e-05, | |
| "rel_l2": 0.007910806685686111, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.4.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024229881819337606, | |
| "mean_abs_err": 3.859053322230466e-05, | |
| "rel_l2": 0.0079184053465724, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005854256451129913, | |
| "mean_abs_err": 2.9199924028944224e-05, | |
| "rel_l2": 0.008913605473935604, | |
| "cosine": 0.9999602437019348 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002213150728493929, | |
| "mean_abs_err": 2.7515470719663426e-05, | |
| "rel_l2": 0.00794124323874712, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002093187067657709, | |
| "mean_abs_err": 2.5714480216265656e-05, | |
| "rel_l2": 0.007949605584144592, | |
| "cosine": 0.999968409538269 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026402773801237345, | |
| "mean_abs_err": 3.13116870529484e-05, | |
| "rel_l2": 0.008024748414754868, | |
| "cosine": 0.9999678730964661 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0008032806217670441, | |
| "mean_abs_err": 4.272177829989232e-05, | |
| "rel_l2": 0.00882418267428875, | |
| "cosine": 0.9999611377716064 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004293927922844887, | |
| "mean_abs_err": 3.857156480080448e-05, | |
| "rel_l2": 0.007992702536284924, | |
| "cosine": 0.9999680519104004 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.5.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00028262659907341003, | |
| "mean_abs_err": 3.9967340853763744e-05, | |
| "rel_l2": 0.007882468402385712, | |
| "cosine": 0.9999688863754272 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0007869638502597809, | |
| "mean_abs_err": 3.033451321243774e-05, | |
| "rel_l2": 0.008966636843979359, | |
| "cosine": 0.9999598264694214 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002321251668035984, | |
| "mean_abs_err": 2.8264252250664867e-05, | |
| "rel_l2": 0.007936658337712288, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00020975619554519653, | |
| "mean_abs_err": 2.687431151571218e-05, | |
| "rel_l2": 0.00793658196926117, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0003668307326734066, | |
| "mean_abs_err": 3.830580317298882e-05, | |
| "rel_l2": 0.00800922978669405, | |
| "cosine": 0.9999679327011108 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0004354706034064293, | |
| "mean_abs_err": 4.333802644396201e-05, | |
| "rel_l2": 0.008297188207507133, | |
| "cosine": 0.9999656081199646 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000540098175406456, | |
| "mean_abs_err": 4.0013848774833605e-05, | |
| "rel_l2": 0.007958641275763512, | |
| "cosine": 0.9999682903289795 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.6.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002861279062926769, | |
| "mean_abs_err": 4.073204399901442e-05, | |
| "rel_l2": 0.007924570702016354, | |
| "cosine": 0.9999685883522034 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005717342719435692, | |
| "mean_abs_err": 3.029292929568328e-05, | |
| "rel_l2": 0.00893236044794321, | |
| "cosine": 0.9999601244926453 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002617649734020233, | |
| "mean_abs_err": 2.7768001018557698e-05, | |
| "rel_l2": 0.007922813296318054, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024792831391096115, | |
| "mean_abs_err": 2.615756238810718e-05, | |
| "rel_l2": 0.007923275232315063, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026348279789090157, | |
| "mean_abs_err": 3.8971546018728986e-05, | |
| "rel_l2": 0.008038604632019997, | |
| "cosine": 0.9999676942825317 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005475282669067383, | |
| "mean_abs_err": 4.106058622710407e-05, | |
| "rel_l2": 0.008104524575173855, | |
| "cosine": 0.9999671578407288 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00045084208250045776, | |
| "mean_abs_err": 3.926951831090264e-05, | |
| "rel_l2": 0.007955125533044338, | |
| "cosine": 0.9999683499336243 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.7.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002508438192307949, | |
| "mean_abs_err": 4.0381986764259636e-05, | |
| "rel_l2": 0.007943028584122658, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005675088614225388, | |
| "mean_abs_err": 3.0484201488434337e-05, | |
| "rel_l2": 0.008994260802865028, | |
| "cosine": 0.9999595880508423 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024007679894566536, | |
| "mean_abs_err": 2.7520332878339104e-05, | |
| "rel_l2": 0.00794390682131052, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00018938904395326972, | |
| "mean_abs_err": 2.6035202608909458e-05, | |
| "rel_l2": 0.007932097651064396, | |
| "cosine": 0.9999684691429138 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00024977908469736576, | |
| "mean_abs_err": 3.885027763317339e-05, | |
| "rel_l2": 0.008003909140825272, | |
| "cosine": 0.9999678134918213 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0005518754478543997, | |
| "mean_abs_err": 4.470529893296771e-05, | |
| "rel_l2": 0.008474930189549923, | |
| "cosine": 0.9999641180038452 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00039626751095056534, | |
| "mean_abs_err": 3.9939033740665764e-05, | |
| "rel_l2": 0.007926104590296745, | |
| "cosine": 0.9999686479568481 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.8.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002555013634264469, | |
| "mean_abs_err": 4.049383642268367e-05, | |
| "rel_l2": 0.007914336398243904, | |
| "cosine": 0.9999687075614929 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.mlp.down_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 6912 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.0004992559552192688, | |
| "mean_abs_err": 3.0718092602910474e-05, | |
| "rel_l2": 0.008930213749408722, | |
| "cosine": 0.99996018409729 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.mlp.gate_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00023995828814804554, | |
| "mean_abs_err": 2.6906536731985398e-05, | |
| "rel_l2": 0.007929641753435135, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.mlp.up_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0002329070121049881, | |
| "mean_abs_err": 2.6256580895278603e-05, | |
| "rel_l2": 0.007917355746030807, | |
| "cosine": 0.9999687671661377 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.self_attn.k_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.0004360424354672432, | |
| "mean_abs_err": 3.7818590499227867e-05, | |
| "rel_l2": 0.007931231521070004, | |
| "cosine": 0.9999685287475586 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.self_attn.o_proj.weight", | |
| "shape": [ | |
| 1152, | |
| 1024 | |
| ], | |
| "group_size": 256, | |
| "max_abs_err": 0.00048078224062919617, | |
| "mean_abs_err": 4.2723873775685206e-05, | |
| "rel_l2": 0.008095338009297848, | |
| "cosine": 0.9999672770500183 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.self_attn.q_proj.weight", | |
| "shape": [ | |
| 1024, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.000507330521941185, | |
| "mean_abs_err": 4.0819038986228406e-05, | |
| "rel_l2": 0.007977495901286602, | |
| "cosine": 0.9999681711196899 | |
| }, | |
| { | |
| "tensor": "text_encoder.layers.9.self_attn.v_proj.weight", | |
| "shape": [ | |
| 256, | |
| 1152 | |
| ], | |
| "group_size": 64, | |
| "max_abs_err": 0.00026716198772192, | |
| "mean_abs_err": 4.0962644561659545e-05, | |
| "rel_l2": 0.007875950075685978, | |
| "cosine": 0.9999690055847168 | |
| } | |
| ] | |
| } |