devils-agent / models /v003-transformer /training-report.json
devildasdf's picture
Upload experimental BAIM code, research checkpoints and measured evaluations
795f737 verified
Raw History Blame Contribute Delete
8.12 kB
{
"architecture": {
"data": "datasets/synthetic-v1",
"output": "models/v003-transformer",
"encoder": "transformer",
"no_lexical": false,
"epochs": 16,
"width": 64,
"seed": 1729
},
"parameter_count": 163908,
"threads": 2,
"device": "cpu",
"training_seconds": 337.2460462000454,
"dataset_manifest": {
"train": {
"count": 2400,
"seed": 101,
"templates": [
"fieldset",
"grid",
"stack"
],
"sha256": "3f24899383d914465232cfaac27f654eae7217ed3fb37894cf07e551120c56e5"
},
"validation": {
"count": 480,
"seed": 202,
"templates": [
"fieldset",
"grid",
"stack"
],
"sha256": "1cf777e5703257e3471c3f67b2e421fd3de505de429fbf57223f452648c477a8"
},
"test": {
"count": 480,
"seed": 303,
"templates": [
"nested",
"table"
],
"sha256": "849ccce76126296a01ee5ef756f2790470dea0e87fad94802d66c6e43fee20ac"
},
"novel_wording": {
"count": 480,
"seed": 404,
"templates": [
"nested",
"table"
],
"sha256": "8b467567370321ad5fbe604b282b89de6ceaaa4b679621fa0f3685708fa6e1c2"
},
"limitations": "Single-step generated tasks; test layouts held out but vocabulary shared. No arbitrary-site claim."
},
"history": [
{
"epoch": 1,
"loss": 1.3394273837636177,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.01182039876584895,
"target_ece": 0.019772875006310642,
"candidate_recall": 1.0
}
},
{
"epoch": 2,
"loss": 0.014025603610973218,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 0.9979166388511658,
"joint_step_accuracy": 0.9979166388511658,
"action_ece": 0.0006074905395507812,
"target_ece": 0.007352764179813676,
"candidate_recall": 1.0
}
},
{
"epoch": 3,
"loss": 0.0025258230989014634,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.0003701448440551758,
"target_ece": 0.0042307451367378235,
"candidate_recall": 1.0
}
},
{
"epoch": 4,
"loss": 0.0009982854274934844,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.00026351213455200195,
"target_ece": 0.00313020299654454,
"candidate_recall": 1.0
}
},
{
"epoch": 5,
"loss": 0.0005972246516122189,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.00018608570098876953,
"target_ece": 0.0023641924490220845,
"candidate_recall": 1.0
}
},
{
"epoch": 6,
"loss": 0.0003768317979656307,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.00013107061386108398,
"target_ece": 0.0020344614749774337,
"candidate_recall": 1.0
}
},
{
"epoch": 7,
"loss": 0.00025686507146211195,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.00010484457015991211,
"target_ece": 0.0018213928560726345,
"candidate_recall": 1.0
}
},
{
"epoch": 8,
"loss": 0.000185561689202012,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 8.314847946166992e-05,
"target_ece": 0.001661624584812671,
"candidate_recall": 1.0
}
},
{
"epoch": 9,
"loss": 0.00014041483920759914,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 6.490945816040039e-05,
"target_ece": 0.0015167492674663663,
"candidate_recall": 1.0
}
},
{
"epoch": 10,
"loss": 0.0001094463490230019,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 5.501508712768555e-05,
"target_ece": 0.001387017488013953,
"candidate_recall": 1.0
}
},
{
"epoch": 11,
"loss": 8.679379004062962e-05,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 4.649162292480469e-05,
"target_ece": 0.001319906412390992,
"candidate_recall": 1.0
}
},
{
"epoch": 12,
"loss": 7.15753852235043e-05,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 4.0471553802490234e-05,
"target_ece": 0.0012643637601286173,
"candidate_recall": 1.0
}
},
{
"epoch": 13,
"loss": 5.9838949853560496e-05,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 3.415346145629883e-05,
"target_ece": 0.0012052194506395608,
"candidate_recall": 1.0
}
},
{
"epoch": 14,
"loss": 5.114553201373814e-05,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 3.0517578125e-05,
"target_ece": 0.00116775679634884,
"candidate_recall": 1.0
}
},
{
"epoch": 15,
"loss": 4.43891615903244e-05,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 2.7120113372802734e-05,
"target_ece": 0.0011370688735041767,
"candidate_recall": 1.0
}
},
{
"epoch": 16,
"loss": 3.890003048773475e-05,
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 2.4437904357910156e-05,
"target_ece": 0.0011151702201459557,
"candidate_recall": 1.0
}
}
],
"evaluation": {
"validation": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.0,
"target_ece": 0.0006519494054373354,
"candidate_recall": 1.0
},
"test": {
"samples": 480,
"action_accuracy": 1.0,
"target_accuracy": 1.0,
"joint_step_accuracy": 1.0,
"action_ece": 0.0,
"target_ece": 0.0007661207773708156,
"candidate_recall": 1.0
},
"novel_wording": {
"samples": 480,
"action_accuracy": 0.49166667461395264,
"target_accuracy": 0.49791666865348816,
"joint_step_accuracy": 0.4270833432674408,
"action_ece": 0.506439236109145,
"target_ece": 0.47587777813896537,
"candidate_recall": 1.0
}
},
"limitations": "Synthetic single-step action/target prediction; not arbitrary-site task success.",
"production_promoted": false
}