trident-ec / sp2_spine_text_s0_lr2.json
farguney's picture
Upload sp2_spine_text_s0_lr2.json with huggingface_hub
240d0c8 verified
Raw
History Blame
5.04 kB
{
"screen": "SP-2-SPINE-LM",
"seed": 0,
"steps": 3000,
"batch": 32,
"block": 512,
"lrs": [
0.0003
],
"text_mb": 24.0,
"d_spine": 384,
"layers_spine": 8,
"n_spine": 12018432,
"tf_mult": 1.0,
"d_transformer": 400,
"layers_transformer": 6,
"n_transformer": 11962800,
"device": "cuda",
"thresholds": {
"pass": 1.05,
"concern": 1.15
},
"corpora": {
"text": {
"spine_lr0.0003": {
"best_bpb": 1.6920182704925537,
"history": [
{
"step": 0,
"train_bits": 8.10444164619865,
"eval_bpb": 6.368494033813477,
"best": 6.368494033813477,
"kBps": 10.9
},
{
"step": 250,
"train_bits": 2.0672142651058736,
"eval_bpb": 2.1344122886657715,
"best": 2.1344122886657715,
"kBps": 21.1
},
{
"step": 500,
"train_bits": 1.9160518339565165,
"eval_bpb": 1.9972063302993774,
"best": 1.9972063302993774,
"kBps": 21.1
},
{
"step": 750,
"train_bits": 1.7798878378062541,
"eval_bpb": 1.9081671237945557,
"best": 1.9081671237945557,
"kBps": 21.1
},
{
"step": 1000,
"train_bits": 1.7443163221629978,
"eval_bpb": 1.852670669555664,
"best": 1.852670669555664,
"kBps": 21.1
},
{
"step": 1250,
"train_bits": 1.6570211962554025,
"eval_bpb": 1.8103423118591309,
"best": 1.8103423118591309,
"kBps": 21.1
},
{
"step": 1500,
"train_bits": 1.6759471990522512,
"eval_bpb": 1.7894508838653564,
"best": 1.7894508838653564,
"kBps": 21.1
},
{
"step": 1750,
"train_bits": 1.6142092110108217,
"eval_bpb": 1.745488166809082,
"best": 1.745488166809082,
"kBps": 21.1
},
{
"step": 2000,
"train_bits": 1.6002025999592528,
"eval_bpb": 1.721181035041809,
"best": 1.721181035041809,
"kBps": 21.1
},
{
"step": 2250,
"train_bits": 1.5390813417038176,
"eval_bpb": 1.7037999629974365,
"best": 1.7037999629974365,
"kBps": 21.1
},
{
"step": 2500,
"train_bits": 1.4838240039122006,
"eval_bpb": 1.6920182704925537,
"best": 1.6920182704925537,
"kBps": 21.1
},
{
"step": 2750,
"train_bits": 1.5799485471324417,
"eval_bpb": 1.6952470541000366,
"best": 1.6920182704925537,
"kBps": 21.1
},
{
"step": 2999,
"train_bits": 1.4944028287494229,
"eval_bpb": 1.6929237842559814,
"best": 1.6920182704925537,
"kBps": 21.1
}
],
"wall_s": 2334.0
},
"tf_lr0.0003": {
"best_bpb": 2.3104419708251953,
"history": [
{
"step": 0,
"train_bits": 8.184333090969252,
"eval_bpb": 6.767765045166016,
"best": 6.767765045166016,
"kBps": 16.1
},
{
"step": 250,
"train_bits": 3.5008781530327293,
"eval_bpb": 3.5642881393432617,
"best": 3.5642881393432617,
"kBps": 26.8
},
{
"step": 500,
"train_bits": 3.384189988214124,
"eval_bpb": 3.3779335021972656,
"best": 3.3779335021972656,
"kBps": 26.8
},
{
"step": 750,
"train_bits": 2.96575207108129,
"eval_bpb": 3.0074915885925293,
"best": 3.0074915885925293,
"kBps": 26.9
},
{
"step": 1000,
"train_bits": 2.7209486981728017,
"eval_bpb": 2.7959744930267334,
"best": 2.7959744930267334,
"kBps": 26.9
},
{
"step": 1250,
"train_bits": 2.580237824699028,
"eval_bpb": 2.642259120941162,
"best": 2.642259120941162,
"kBps": 26.9
},
{
"step": 1500,
"train_bits": 2.475455330869052,
"eval_bpb": 2.528750419616699,
"best": 2.528750419616699,
"kBps": 26.9
},
{
"step": 1750,
"train_bits": 2.421410126800716,
"eval_bpb": 2.442403554916382,
"best": 2.442403554916382,
"kBps": 26.9
},
{
"step": 2000,
"train_bits": 2.330143061571444,
"eval_bpb": 2.388734817504883,
"best": 2.388734817504883,
"kBps": 26.9
},
{
"step": 2250,
"train_bits": 2.2836532273381622,
"eval_bpb": 2.3470757007598877,
"best": 2.3470757007598877,
"kBps": 26.9
},
{
"step": 2500,
"train_bits": 2.298321971595553,
"eval_bpb": 2.3256967067718506,
"best": 2.3256967067718506,
"kBps": 26.9
},
{
"step": 2750,
"train_bits": 2.2844876871601487,
"eval_bpb": 2.3124146461486816,
"best": 2.3124146461486816,
"kBps": 26.9
},
{
"step": 2999,
"train_bits": 2.240674246939594,
"eval_bpb": 2.3104419708251953,
"best": 2.3104419708251953,
"kBps": 26.9
}
],
"wall_s": 1828.9
},
"spine_best": 1.6920182704925537,
"tf_best": 2.3104419708251953,
"ratio": 0.7323353245215823,
"verdict": "PASS"
}
},
"partial": false,
"verdict": "PASS"
}