Tribewarez/pot-o-pathfinder-tiny-v1
Text Generation • Updated
Datasets:
challenge stringlengths 76 103 | optimal_path stringlengths 54 67 | difficulty float64 1.86 3.64 | source stringclasses 1
value |
|---|---|---|---|
tensor:shape=[32,32];dtype=float32;target_mml=0.53;ops:matmul,quant8,lowrank,relu,prune0.4 | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.48 | 1.9159 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.28;ops:prune0.4,quant4,quant8,relu,transpose,matmul | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.26 | 2.8365 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float16;target_mml=0.3;ops:transpose,gelu,prune0.2,matmul,prune0.4,relu | matmul -> gelu -> quant:4bit -> prune:0.3 -> score:0.28 | 2.737 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=int8;target_mml=0.32;ops:matmul,transpose,gelu,lowrank,prune0.2 | matmul -> gelu -> quant:8bit -> prune:0.4 -> score:0.3 | 2.6439 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float32;target_mml=0.49;ops:relu,gelu,prune0.4,lowrank,quant4,matmul | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.3 -> score:0.46 | 2.0291 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float16;target_mml=0.45;ops:quant8,relu,transpose,matmul | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.2 -> score:0.41 | 2.152 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=int8;target_mml=0.23;ops:transpose,gelu,lowrank,prune0.2 | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.21 | 3.1203 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.28;ops:prune0.2,gelu,quant8,transpose,prune0.4 | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.4 -> score:0.26 | 2.8365 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float16;target_mml=0.33;ops:prune0.4,quant4,quant8,matmul,relu | matmul[lowrank:16] -> relu -> quant:8bit -> prune:0.3 -> score:0.28 | 2.5995 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float16;target_mml=0.23;ops:quant8,relu,transpose,prune0.4,prune0.2,quant4 | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.4 -> score:0.21 | 3.1203 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float32;target_mml=0.53;ops:lowrank,transpose,relu,prune0.2,matmul | matmul -> relu -> quant:4bit -> prune:0.2 -> score:0.5 | 1.9159 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float16;target_mml=0.16;ops:relu,transpose,lowrank,quant8,matmul | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.15 | 3.6439 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.19;ops:transpose,matmul,quant4,lowrank | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.17 | 3.3959 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float32;target_mml=0.5;ops:relu,transpose,prune0.2,prune0.4,gelu | matmul -> relu -> quant:4bit -> prune:0.2 -> score:0.48 | 2 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float32;target_mml=0.48;ops:prune0.4,gelu,relu,quant4,prune0.2,transpose | matmul[lowrank:32] -> gelu -> quant:8bit -> prune:0.2 -> score:0.46 | 2.0589 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.16;ops:matmul,quant4,prune0.4,transpose | matmul -> relu -> quant:4bit -> prune:0.4 -> score:0.15 | 3.6439 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float32;target_mml=0.26;ops:transpose,relu,quant8,gelu,matmul,lowrank | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.4 -> score:0.24 | 2.9434 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=int8;target_mml=0.41;ops:gelu,prune0.2,quant8,relu,matmul | matmul[lowrank:32] -> gelu -> quant:8bit -> prune:0.4 -> score:0.38 | 2.2863 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.41;ops:quant4,prune0.4,prune0.2,gelu,lowrank | matmul -> gelu -> quant:4bit -> prune:0.2 -> score:0.36 | 2.2863 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float16;target_mml=0.17;ops:lowrank,transpose,quant8,relu,matmul,prune0.4 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.15 | 3.5564 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float16;target_mml=0.38;ops:prune0.2,matmul,gelu,quant8 | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.3 -> score:0.35 | 2.3959 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float32;target_mml=0.33;ops:transpose,lowrank,relu,quant4,prune0.4 | matmul -> relu -> quant:8bit -> prune:0.3 -> score:0.28 | 2.5995 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float32;target_mml=0.23;ops:prune0.2,quant8,relu,transpose | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.4 -> score:0.21 | 3.1203 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.54;ops:quant8,relu,lowrank,quant4 | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.3 -> score:0.52 | 1.889 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.37;ops:prune0.2,transpose,gelu,lowrank,quant8,relu | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.4 -> score:0.36 | 2.4344 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=int8;target_mml=0.39;ops:gelu,quant4,prune0.4,matmul | matmul -> relu -> quant:4bit -> prune:0.2 -> score:0.36 | 2.3585 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.54;ops:quant4,lowrank,prune0.2,prune0.4,gelu,quant8 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.3 -> score:0.52 | 1.889 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=int8;target_mml=0.18;ops:quant4,relu,quant8,matmul,prune0.4,prune0.2 | matmul -> relu -> quant:4bit -> prune:0.4 -> score:0.16 | 3.4739 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=int8;target_mml=0.2;ops:quant8,prune0.2,quant4,transpose,lowrank | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.18 | 3.3219 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float32;target_mml=0.39;ops:gelu,prune0.2,transpose,prune0.4 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.3 -> score:0.37 | 2.3585 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float16;target_mml=0.5;ops:gelu,quant8,prune0.2,relu | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.46 | 2 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float16;target_mml=0.43;ops:matmul,quant8,prune0.2,transpose | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.2 -> score:0.41 | 2.2176 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.33;ops:quant4,prune0.2,gelu,transpose,quant8 | matmul -> relu -> quant:4bit -> prune:0.4 -> score:0.32 | 2.5995 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.4;ops:relu,quant4,prune0.4,lowrank | matmul -> gelu -> quant:8bit -> prune:0.2 -> score:0.35 | 2.3219 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float16;target_mml=0.18;ops:transpose,relu,quant4,lowrank | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.4 -> score:0.16 | 3.4739 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.35;ops:quant8,lowrank,prune0.2,relu,matmul,transpose | matmul -> relu -> quant:8bit -> prune:0.4 -> score:0.34 | 2.5146 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float32;target_mml=0.41;ops:relu,gelu,quant4,matmul | matmul[lowrank:32] -> gelu -> quant:8bit -> prune:0.3 -> score:0.39 | 2.2863 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.44;ops:quant4,quant8,prune0.4,relu | matmul -> relu -> quant:8bit -> prune:0.4 -> score:0.42 | 2.1844 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.21;ops:relu,quant8,lowrank,prune0.4,gelu | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.19 | 3.2515 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=int8;target_mml=0.29;ops:quant4,matmul,prune0.2,quant8,relu | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.27 | 2.7859 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float32;target_mml=0.4;ops:gelu,quant8,transpose,lowrank,prune0.2 | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.2 -> score:0.36 | 2.3219 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float16;target_mml=0.52;ops:prune0.4,gelu,prune0.2,relu | matmul -> gelu -> quant:8bit -> prune:0.4 -> score:0.48 | 1.9434 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float16;target_mml=0.4;ops:quant4,prune0.2,matmul,lowrank,prune0.4 | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.2 -> score:0.35 | 2.3219 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float32;target_mml=0.25;ops:prune0.2,quant8,matmul,relu,gelu | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.23 | 3 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float32;target_mml=0.3;ops:prune0.4,quant8,gelu,relu | matmul -> gelu -> quant:8bit -> prune:0.2 -> score:0.28 | 2.737 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float32;target_mml=0.51;ops:quant8,relu,transpose,lowrank | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.2 -> score:0.47 | 1.9714 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float32;target_mml=0.2;ops:transpose,matmul,quant8,quant4,relu | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.18 | 3.3219 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.35;ops:gelu,matmul,quant4,prune0.4,lowrank,quant8 | matmul -> relu -> quant:8bit -> prune:0.2 -> score:0.32 | 2.5146 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float16;target_mml=0.25;ops:prune0.2,matmul,quant4,transpose,lowrank,gelu | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.23 | 3 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=int8;target_mml=0.32;ops:prune0.2,transpose,quant4,relu | matmul[lowrank:16] -> relu -> quant:8bit -> prune:0.4 -> score:0.29 | 2.6439 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.18;ops:gelu,matmul,prune0.2,transpose,quant4,lowrank | matmul -> relu -> quant:4bit -> prune:0.4 -> score:0.16 | 3.4739 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float16;target_mml=0.38;ops:lowrank,prune0.2,quant8,quant4 | matmul -> gelu -> quant:8bit -> prune:0.2 -> score:0.35 | 2.3959 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=int8;target_mml=0.54;ops:matmul,lowrank,quant4,relu,gelu | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.3 -> score:0.5 | 1.889 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=int8;target_mml=0.55;ops:lowrank,relu,quant4,gelu | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.3 -> score:0.53 | 1.8625 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float32;target_mml=0.52;ops:transpose,matmul,relu,quant4,quant8,prune0.4 | matmul[lowrank:32] -> relu -> quant:8bit -> prune:0.2 -> score:0.51 | 1.9434 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.42;ops:quant4,transpose,quant8,relu | matmul -> gelu -> quant:4bit -> prune:0.2 -> score:0.4 | 2.2515 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float32;target_mml=0.25;ops:matmul,lowrank,prune0.2,relu,prune0.4,quant4 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.23 | 3 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float16;target_mml=0.27;ops:prune0.4,matmul,quant8,prune0.2 | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.4 -> score:0.25 | 2.889 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float32;target_mml=0.26;ops:quant4,matmul,transpose,lowrank,prune0.4,quant8 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.24 | 2.9434 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float16;target_mml=0.44;ops:quant4,quant8,gelu,transpose | matmul[lowrank:16] -> relu -> quant:8bit -> prune:0.4 -> score:0.43 | 2.1844 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.27;ops:prune0.4,lowrank,quant8,transpose,gelu | matmul -> relu -> quant:4bit -> prune:0.4 -> score:0.25 | 2.889 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float32;target_mml=0.44;ops:relu,transpose,lowrank,quant4 | matmul -> gelu -> quant:4bit -> prune:0.3 -> score:0.41 | 2.1844 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.38;ops:gelu,transpose,matmul,quant4 | matmul -> relu -> quant:8bit -> prune:0.4 -> score:0.34 | 2.3959 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=int8;target_mml=0.24;ops:prune0.2,quant8,quant4,relu,prune0.4 | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.4 -> score:0.22 | 3.0589 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.18;ops:lowrank,prune0.2,quant4,prune0.4 | matmul -> relu -> quant:4bit -> prune:0.4 -> score:0.16 | 3.4739 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float16;target_mml=0.37;ops:gelu,quant4,prune0.2,lowrank,matmul,prune0.4 | matmul[lowrank:32] -> gelu -> quant:8bit -> prune:0.2 -> score:0.35 | 2.4344 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float32;target_mml=0.41;ops:prune0.4,matmul,transpose,gelu,lowrank | matmul[lowrank:16] -> gelu -> quant:8bit -> prune:0.3 -> score:0.38 | 2.2863 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=int8;target_mml=0.44;ops:relu,prune0.4,lowrank,gelu,transpose,quant4 | matmul -> gelu -> quant:8bit -> prune:0.4 -> score:0.43 | 2.1844 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float32;target_mml=0.43;ops:gelu,quant4,prune0.4,prune0.2,relu | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.2 -> score:0.4 | 2.2176 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.37;ops:gelu,prune0.4,matmul,prune0.2,quant4,transpose | matmul -> gelu -> quant:8bit -> prune:0.2 -> score:0.34 | 2.4344 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=int8;target_mml=0.48;ops:lowrank,quant4,prune0.2,quant8 | matmul -> relu -> quant:4bit -> prune:0.3 -> score:0.45 | 2.0589 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.48;ops:prune0.4,quant4,quant8,gelu,matmul | matmul -> relu -> quant:4bit -> prune:0.2 -> score:0.45 | 2.0589 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=int8;target_mml=0.27;ops:quant4,prune0.4,relu,matmul,transpose | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.4 -> score:0.25 | 2.889 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float16;target_mml=0.31;ops:lowrank,quant4,gelu,quant8,prune0.4 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.3 -> score:0.28 | 2.6897 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=int8;target_mml=0.53;ops:gelu,quant4,relu,prune0.4,matmul,transpose | matmul[lowrank:32] -> gelu -> quant:8bit -> prune:0.3 -> score:0.51 | 1.9159 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=float16;target_mml=0.17;ops:prune0.2,transpose,quant8,matmul,relu | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.15 | 3.5564 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=int8;target_mml=0.21;ops:prune0.4,lowrank,quant4,transpose | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.19 | 3.2515 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.25;ops:quant8,lowrank,transpose,prune0.4,relu | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.23 | 3 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=int8;target_mml=0.16;ops:prune0.2,prune0.4,matmul,gelu,quant8,relu | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.15 | 3.6439 | PoT-O/TribeWarez |
tensor:shape=[32,32];dtype=float32;target_mml=0.41;ops:transpose,quant8,matmul,relu | matmul -> relu -> quant:8bit -> prune:0.3 -> score:0.37 | 2.2863 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float16;target_mml=0.21;ops:relu,prune0.4,quant4,matmul | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.19 | 3.2515 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=int8;target_mml=0.51;ops:quant8,matmul,relu,gelu,lowrank,prune0.4 | matmul[lowrank:32] -> relu -> quant:4bit -> prune:0.4 -> score:0.46 | 1.9714 | PoT-O/TribeWarez |
tensor:shape=[64,64];dtype=int8;target_mml=0.32;ops:prune0.4,gelu,quant8,relu,prune0.2,transpose | matmul[lowrank:32] -> gelu -> quant:4bit -> prune:0.3 -> score:0.28 | 2.6439 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float16;target_mml=0.33;ops:quant4,matmul,quant8,transpose,gelu,prune0.4 | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.2 -> score:0.28 | 2.5995 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=float16;target_mml=0.5;ops:prune0.4,lowrank,matmul,quant8 | matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.2 -> score:0.45 | 2 | PoT-O/TribeWarez |
tensor:shape=[32,64];dtype=float16;target_mml=0.34;ops:quant8,prune0.2,transpose,matmul,gelu | matmul[lowrank:32] -> gelu -> quant:8bit -> prune:0.4 -> score:0.3 | 2.5564 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float16;target_mml=0.49;ops:transpose,relu,gelu,quant8 | matmul -> relu -> quant:8bit -> prune:0.2 -> score:0.46 | 2.0291 | PoT-O/TribeWarez |
tensor:shape=[64,128];dtype=float16;target_mml=0.39;ops:relu,transpose,quant4,gelu,lowrank | matmul[lowrank:16] -> gelu -> quant:8bit -> prune:0.2 -> score:0.35 | 2.3585 | PoT-O/TribeWarez |
tensor:shape=[16,32];dtype=float32;target_mml=0.49;ops:quant4,matmul,quant8,relu,transpose,prune0.4 | matmul -> gelu -> quant:4bit -> prune:0.4 -> score:0.44 | 2.0291 | PoT-O/TribeWarez |
tensor:shape=[8,256];dtype=int8;target_mml=0.24;ops:gelu,lowrank,transpose,quant4,quant8 | matmul[lowrank:16] -> gelu -> quant:4bit -> prune:0.4 -> score:0.22 | 3.0589 | PoT-O/TribeWarez |
synthetic-pot-o-challenges-v1
Tiny synthetic starter dataset for training PoT-O (Proof of Tensor Optimizations) pathfinder models.
Format (JSONL)
{"challenge": "tensor:shape=[32,64];dtype=float16;target_mml=0.42;ops:matmul,lowrank,gelu,quant4,prune0.3,transpose", "optimal_path": "path: matmul[lowrank:16] -> relu -> quant:4bit -> prune:0.35 -> score:0.418"}
- challenge: Text encoding of tensor properties & allowed operations
- optimal_path: Heuristic "good" optimization sequence + predicted score (MML-inspired efficiency)
Dataset Details
- Size: 100 examples (v1 – expand in future versions)
- Train/Val Split: 90/10
- Generation: Rule-based synthetic (random shapes/dtypes/targets + simple heuristics for paths). Not from real traces yet.
- Intended use: Fine-tuning tiny models (e.g.
Tribewarez/pot-o-pathfinder-tiny-v1) to predict better tensor transformation paths for low-power PoT-O miners.
Next Iterations
- Add real tensor traces from ai3-lib
- More diverse challenges
- Verified optimal paths via solvers
- 500+ examples with varied op combinations
- Create v2 with real matrix compression benchmarks
MIT licensed • Tribewarez guild • Live beta • 2026
Link to Model
In your model README.yaml add:
datasets:
- Tribewarez/synthetic-pot-o-challenges-v1
- Downloads last month
- 19
Models trained or fine-tuned on Tribewarez/synthetic-pot-o-challenges-v1