Lottolabs's picture
Add files using upload-large-folder tool
8e6b0e0 verified
Raw History Blame Contribute Delete
9.57 kB
{
"ref_ppl": {
"chat": 8.5587,
"code": 5.6556,
"long-book": 12.1535,
"short-all": 7.1382
},
"gsm8k_bf16_cpu_acc": 0.98,
"rows": [
{
"config": "bf16",
"plan": "plans/bf16.json",
"weights": "IT",
"read_GB": 23.8446,
"floor_ms@128": 46.66,
"chat": {
"ppl_delta_pct": -0.77,
"dnll": -0.0077,
"se": 0.0014,
"top1": 0.975,
"kl": 0.0095,
"kl_p99": 0.103
},
"code": {
"ppl_delta_pct": -0.05,
"dnll": -0.0005,
"se": 0.0014,
"top1": 0.9843,
"kl": 0.0139,
"kl_p99": 0.061
},
"long-book": {
"ppl_delta_pct": -0.14,
"dnll": -0.0014,
"se": 0.0013,
"top1": 0.9764,
"kl": 0.0048,
"kl_p99": 0.053
},
"short-all": {
"ppl_delta_pct": -0.45,
"dnll": -0.0045,
"se": 0.001,
"top1": 0.9791,
"kl": 0.0114,
"kl_p99": 0.081
}
},
{
"config": "a-bfp8",
"plan": "plans/a-bfp8.json",
"weights": "IT",
"read_GB": 12.6675,
"floor_ms@128": 24.83,
"chat": {
"ppl_delta_pct": -0.29,
"dnll": -0.0029,
"se": 0.0015,
"top1": 0.9713,
"kl": 0.0122,
"kl_p99": 0.139
},
"code": {
"ppl_delta_pct": 1.01,
"dnll": 0.01,
"se": 0.0014,
"top1": 0.9827,
"kl": 0.0057,
"kl_p99": 0.08
},
"long-book": {
"ppl_delta_pct": 1.61,
"dnll": 0.016,
"se": 0.0015,
"top1": 0.9723,
"kl": 0.007,
"kl_p99": 0.068
},
"short-all": {
"ppl_delta_pct": 0.28,
"dnll": 0.0028,
"se": 0.001,
"top1": 0.9763,
"kl": 0.0093,
"kl_p99": 0.105
}
},
{
"config": "a-lm4",
"plan": "plans/a-lm4.json",
"weights": "IT",
"read_GB": 12.1641,
"floor_ms@128": 23.84,
"chat": {
"ppl_delta_pct": 0.61,
"dnll": 0.0061,
"se": 0.0018,
"top1": 0.955,
"kl": 0.0208,
"kl_p99": 0.244
},
"code": {
"ppl_delta_pct": 1.98,
"dnll": 0.0196,
"se": 0.0022,
"top1": 0.9703,
"kl": 0.0169,
"kl_p99": 0.289
},
"long-book": {
"ppl_delta_pct": 2.55,
"dnll": 0.0252,
"se": 0.0023,
"top1": 0.9605,
"kl": 0.0149,
"kl_p99": 0.128
},
"short-all": {
"ppl_delta_pct": 1.21,
"dnll": 0.012,
"se": 0.0014,
"top1": 0.9617,
"kl": 0.0191,
"kl_p99": 0.264
}
},
{
"config": "f-dn4",
"plan": "plans/f-dn4.json",
"weights": "IT",
"read_GB": 11.2519,
"floor_ms@128": 22.06,
"chat": {
"ppl_delta_pct": 2.61,
"dnll": 0.0257,
"se": 0.004,
"top1": 0.9102,
"kl": 0.0942,
"kl_p99": 1.456
},
"code": {
"ppl_delta_pct": 2.43,
"dnll": 0.024,
"se": 0.005,
"top1": 0.9333,
"kl": 0.101,
"kl_p99": 1.949
},
"long-book": {
"ppl_delta_pct": -0.64,
"dnll": -0.0065,
"se": 0.0049,
"top1": 0.912,
"kl": 0.0732,
"kl_p99": 0.819
},
"short-all": {
"ppl_delta_pct": 2.53,
"dnll": 0.025,
"se": 0.0031,
"top1": 0.9203,
"kl": 0.0972,
"kl_p99": 1.65
}
},
{
"config": "f-gu4",
"plan": "plans/f-gu4.json",
"weights": "IT",
"read_GB": 9.8363,
"floor_ms@128": 19.3,
"chat": {
"ppl_delta_pct": 1.56,
"dnll": 0.0155,
"se": 0.004,
"top1": 0.9021,
"kl": 0.0962,
"kl_p99": 1.461
},
"code": {
"ppl_delta_pct": 2.43,
"dnll": 0.024,
"se": 0.0051,
"top1": 0.9286,
"kl": 0.1111,
"kl_p99": 2.099
},
"long-book": {
"ppl_delta_pct": 9.73,
"dnll": 0.0929,
"se": 0.0053,
"top1": 0.9053,
"kl": 0.0848,
"kl_p99": 0.863
},
"short-all": {
"ppl_delta_pct": 1.94,
"dnll": 0.0192,
"se": 0.0032,
"top1": 0.9137,
"kl": 0.1028,
"kl_p99": 1.702
}
},
{
"config": "g-rank3072",
"plan": "plans/g-rank3072.json",
"weights": "IT",
"read_GB": 11.6058,
"floor_ms@128": 22.75,
"chat": {
"ppl_delta_pct": -0.04,
"dnll": -0.0004,
"se": 0.0031,
"top1": 0.9308,
"kl": 0.0548,
"kl_p99": 0.837
},
"code": {
"ppl_delta_pct": 2.05,
"dnll": 0.0203,
"se": 0.0041,
"top1": 0.9478,
"kl": 0.0688,
"kl_p99": 1.243
},
"long-book": {
"ppl_delta_pct": 7.3,
"dnll": 0.0704,
"se": 0.0037,
"top1": 0.9422,
"kl": 0.036,
"kl_p99": 0.353
},
"short-all": {
"ppl_delta_pct": 0.87,
"dnll": 0.0087,
"se": 0.0025,
"top1": 0.9382,
"kl": 0.061,
"kl_p99": 1.024
}
},
{
"config": "g-rank2048",
"plan": "plans/g-rank2048.json",
"weights": "IT",
"read_GB": 10.5441,
"floor_ms@128": 20.68,
"chat": {
"ppl_delta_pct": 1.25,
"dnll": 0.0125,
"se": 0.0037,
"top1": 0.9126,
"kl": 0.082,
"kl_p99": 1.28
},
"code": {
"ppl_delta_pct": 1.82,
"dnll": 0.018,
"se": 0.0048,
"top1": 0.9349,
"kl": 0.0971,
"kl_p99": 1.813
},
"long-book": {
"ppl_delta_pct": 8.86,
"dnll": 0.0849,
"se": 0.0048,
"top1": 0.9169,
"kl": 0.0671,
"kl_p99": 0.713
},
"short-all": {
"ppl_delta_pct": 1.5,
"dnll": 0.0149,
"se": 0.003,
"top1": 0.9224,
"kl": 0.0886,
"kl_p99": 1.542
}
},
{
"config": "b-mlp4",
"plan": "plans/b-mlp4.json",
"weights": "IT",
"read_GB": 8.4207,
"floor_ms@128": 16.53,
"chat": {
"ppl_delta_pct": 5.17,
"dnll": 0.0504,
"se": 0.0052,
"top1": 0.8706,
"kl": 0.1747,
"kl_p99": 2.743
},
"code": {
"ppl_delta_pct": 3.91,
"dnll": 0.0383,
"se": 0.0066,
"top1": 0.9083,
"kl": 0.1758,
"kl_p99": 3.069
},
"long-book": {
"ppl_delta_pct": 9.91,
"dnll": 0.0945,
"se": 0.007,
"top1": 0.8769,
"kl": 0.1502,
"kl_p99": 1.589
},
"short-all": {
"ppl_delta_pct": 4.61,
"dnll": 0.0451,
"se": 0.0041,
"top1": 0.8871,
"kl": 0.1752,
"kl_p99": 2.848
}
},
{
"config": "c-mlp4-lm4",
"plan": "plans/c-mlp4-lm4.json",
"weights": "IT",
"read_GB": 7.9174,
"floor_ms@128": 15.55,
"chat": {
"ppl_delta_pct": 6.21,
"dnll": 0.0603,
"se": 0.0054,
"top1": 0.8667,
"kl": 0.186,
"kl_p99": 2.85
},
"code": {
"ppl_delta_pct": 4.99,
"dnll": 0.0487,
"se": 0.0069,
"top1": 0.9057,
"kl": 0.1882,
"kl_p99": 3.352
},
"long-book": {
"ppl_delta_pct": 10.8,
"dnll": 0.1026,
"se": 0.0072,
"top1": 0.874,
"kl": 0.1582,
"kl_p99": 1.699
},
"short-all": {
"ppl_delta_pct": 5.67,
"dnll": 0.0552,
"se": 0.0043,
"top1": 0.8838,
"kl": 0.1869,
"kl_p99": 3.089
}
},
{
"config": "e-all4",
"plan": "plans/e-all4.json",
"weights": "IT",
"read_GB": 7.2096,
"floor_ms@128": 14.17,
"chat": {
"ppl_delta_pct": 7.24,
"dnll": 0.0699,
"se": 0.0068,
"top1": 0.8266,
"kl": 0.2946,
"kl_p99": 4.066
},
"code": {
"ppl_delta_pct": 8.32,
"dnll": 0.0799,
"se": 0.0085,
"top1": 0.8736,
"kl": 0.2975,
"kl_p99": 4.776
},
"long-book": {
"ppl_delta_pct": 19.16,
"dnll": 0.1753,
"se": 0.0107,
"top1": 0.8009,
"kl": 0.3806,
"kl_p99": 3.648
},
"short-all": {
"ppl_delta_pct": 7.71,
"dnll": 0.0743,
"se": 0.0053,
"top1": 0.8472,
"kl": 0.2959,
"kl_p99": 4.394
}
},
{
"config": "e-all4-lm4",
"plan": "plans/e-all4-lm4.json",
"weights": "IT",
"read_GB": 6.7063,
"floor_ms@128": 13.18,
"chat": {
"ppl_delta_pct": 8.48,
"dnll": 0.0814,
"se": 0.0069,
"top1": 0.824,
"kl": 0.3089,
"kl_p99": 4.273
},
"code": {
"ppl_delta_pct": 9.76,
"dnll": 0.0931,
"se": 0.0087,
"top1": 0.8709,
"kl": 0.3114,
"kl_p99": 5.03
},
"long-book": {
"ppl_delta_pct": 19.92,
"dnll": 0.1816,
"se": 0.0108,
"top1": 0.7996,
"kl": 0.3899,
"kl_p99": 3.703
},
"short-all": {
"ppl_delta_pct": 9.04,
"dnll": 0.0865,
"se": 0.0055,
"top1": 0.8446,
"kl": 0.31,
"kl_p99": 4.618
}
},
{
"config": "d-qat-a",
"plan": "plans/a-bfp8.json",
"weights": "QAT",
"read_GB": 12.6675,
"floor_ms@128": 24.83,
"chat": {
"ppl_delta_pct": -2.86,
"dnll": -0.029,
"se": 0.0053,
"top1": 0.9074,
"kl": 0.1289,
"kl_p99": 2.267
},
"code": {
"ppl_delta_pct": -8.9,
"dnll": -0.0932,
"se": 0.0059,
"top1": 0.928,
"kl": 0.1018,
"kl_p99": 1.497
},
"long-book": {
"ppl_delta_pct": -4.87,
"dnll": -0.0499,
"se": 0.0054,
"top1": 0.9055,
"kl": 0.0771,
"kl_p99": 0.782
},
"short-all": {
"ppl_delta_pct": -5.55,
"dnll": -0.0571,
"se": 0.0039,
"top1": 0.9164,
"kl": 0.117,
"kl_p99": 2.004
}
},
{
"config": "d-qat-b",
"plan": "plans/b-mlp4.json",
"weights": "QAT",
"read_GB": 8.4207,
"floor_ms@128": 16.53,
"chat": {
"ppl_delta_pct": 0.98,
"dnll": 0.0098,
"se": 0.0062,
"top1": 0.8658,
"kl": 0.2104,
"kl_p99": 3.608
},
"code": {
"ppl_delta_pct": -0.85,
"dnll": -0.0086,
"se": 0.0075,
"top1": 0.8987,
"kl": 0.1855,
"kl_p99": 3.064
},
"long-book": {
"ppl_delta_pct": 6.68,
"dnll": 0.0647,
"se": 0.0074,
"top1": 0.8719,
"kl": 0.1512,
"kl_p99": 1.517
},
"short-all": {
"ppl_delta_pct": 0.18,
"dnll": 0.0018,
"se": 0.0048,
"top1": 0.8802,
"kl": 0.1995,
"kl_p99": 3.406
}
}
],
"gsm8k_50": {
"a-lm4": {
"acc": 0.98,
"agree": 0.96
},
"a-bfp8": {
"acc": 1.0,
"agree": 0.98
},
"b-mlp4": {
"acc": 0.98,
"agree": 0.96
}
},
"pick": "a-lm4"
}