lnp_ml/models/final/optuna_trials.json

562 lines
12 KiB
JSON

[
{
"number": 0,
"value": 2.9030250906944275,
"params": {
"dropout": 0.249816047538945,
"lr": 0.0007969454818643932,
"weight_decay": 0.008471801418819975,
"backbone_lr_ratio": 0.15751320499779725,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 16
},
"user_attrs": {
"epoch_mean": 10,
"fold_best_epochs": [
7,
11,
13
],
"fold_val_losses": [
3.346657862265905,
2.727404405673345,
2.635013004144033
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 1,
"value": 4.1861379808849755,
"params": {
"dropout": 0.18493564427131048,
"lr": 2.3102018878452926e-05,
"weight_decay": 5.415244119402538e-05,
"backbone_lr_ratio": 0.04059611610484305,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 32
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
20
],
"fold_val_losses": [
4.24272620677948,
4.012106723255581,
4.303581012619866
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 2,
"value": 3.5133936824621976,
"params": {
"dropout": 0.1798695128633439,
"lr": 0.00010677482709481354,
"weight_decay": 0.00234238498471129,
"backbone_lr_ratio": 0.012385137298860933,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 1,
"llm_lora_r": 32
},
"user_attrs": {
"epoch_mean": 19,
"fold_best_epochs": [
18,
20,
20
],
"fold_val_losses": [
3.6705237362119885,
3.35566207435396,
3.513995236820645
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 3,
"value": 4.693677743275961,
"params": {
"dropout": 0.27606099749584057,
"lr": 1.7541893487450798e-05,
"weight_decay": 0.0009565499215943821,
"backbone_lr_ratio": 0.011715937392307063,
"moe_n_experts": 2,
"moe_top_k": 1,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 16
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
20
],
"fold_val_losses": [
4.833255463176304,
4.549637953440349,
4.698139813211229
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 4,
"value": 3.2253691399538957,
"params": {
"dropout": 0.4757995766256756,
"lr": 0.0006161049539380963,
"weight_decay": 0.002463768595899745,
"backbone_lr_ratio": 0.697828126512603,
"moe_n_experts": 4,
"moe_top_k": 1,
"moe_expert_hidden_mult": 1,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 13,
"fold_best_epochs": [
7,
13,
19
],
"fold_val_losses": [
3.5457335942321353,
3.1522699031564922,
2.9781039224730597
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 5,
"value": 4.624290854842574,
"params": {
"dropout": 0.31707843326329943,
"lr": 1.913588048769229e-05,
"weight_decay": 0.016172900811143146,
"backbone_lr_ratio": 0.01409617514981587,
"moe_n_experts": 2,
"moe_top_k": 1,
"moe_expert_hidden_mult": 1,
"llm_lora_r": 16
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
20
],
"fold_val_losses": [
4.722254474957784,
4.507743954658508,
4.642874134911431
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 6,
"value": 4.131107242019088,
"params": {
"dropout": 0.24338629141770907,
"lr": 1.70505392602693e-05,
"weight_decay": 0.028340904295147733,
"backbone_lr_ratio": 0.17643967683381545,
"moe_n_experts": 2,
"moe_top_k": 1,
"moe_expert_hidden_mult": 1,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
20
],
"fold_val_losses": [
4.218653745121426,
3.9478391938739352,
4.226828787061903
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 7,
"value": 3.0308966758074583,
"params": {
"dropout": 0.38529791488919807,
"lr": 0.0003323304206226791,
"weight_decay": 0.0017583640270008513,
"backbone_lr_ratio": 0.3482846706526883,
"moe_n_experts": 4,
"moe_top_k": 1,
"moe_expert_hidden_mult": 1,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 14,
"fold_best_epochs": [
8,
15,
19
],
"fold_val_losses": [
3.5245150327682495,
2.85073944595125,
2.7174355487028756
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 8,
"value": 3.889537049664391,
"params": {
"dropout": 0.4630265895704372,
"lr": 3.151987295193886e-05,
"weight_decay": 0.00043805807679056546,
"backbone_lr_ratio": 0.3244160088734159,
"moe_n_experts": 8,
"moe_top_k": 1,
"moe_expert_hidden_mult": 1,
"llm_lora_r": 16
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
20
],
"fold_val_losses": [
4.002199537224239,
3.755847387843662,
3.9105642239252725
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 9,
"value": 2.9054288014217664,
"params": {
"dropout": 0.17462802355441434,
"lr": 0.0006097025297491432,
"weight_decay": 0.0014367095138664223,
"backbone_lr_ratio": 0.4119839624605187,
"moe_n_experts": 2,
"moe_top_k": 1,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 13,
"fold_best_epochs": [
10,
11,
17
],
"fold_val_losses": [
3.114242351717419,
2.8655375705824957,
2.736506481965383
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 10,
"value": 3.2267529169718423,
"params": {
"dropout": 0.11691600370687807,
"lr": 0.00013112992007873722,
"weight_decay": 0.07553503645583189,
"backbone_lr_ratio": 0.05739049944367051,
"moe_n_experts": 8,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 16
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
19
],
"fold_val_losses": [
3.380473878648546,
3.0783044166035123,
3.2214804556634693
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 11,
"value": 2.8969200551509857,
"params": {
"dropout": 0.19323529834582587,
"lr": 0.000980484886752915,
"weight_decay": 0.00014367509113739864,
"backbone_lr_ratio": 0.13093310334961422,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 11,
"fold_best_epochs": [
11,
6,
17
],
"fold_val_losses": [
3.2465109096633062,
2.925493541691038,
2.5187557140986123
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 12,
"value": 2.9026516896707037,
"params": {
"dropout": 0.34287370961329033,
"lr": 0.0009954383775681419,
"weight_decay": 1.2791436634859164e-05,
"backbone_lr_ratio": 0.11798172454857601,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 14,
"fold_best_epochs": [
8,
20,
15
],
"fold_val_losses": [
3.2743234402603574,
2.9065700074036918,
2.527061621348063
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 13,
"value": 3.0620501229056605,
"params": {
"dropout": 0.35582764453212956,
"lr": 0.00025048271616507544,
"weight_decay": 1.0326835200348013e-05,
"backbone_lr_ratio": 0.08476798466510083,
"moe_n_experts": 8,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 19,
"fold_best_epochs": [
16,
20,
20
],
"fold_val_losses": [
3.3088861074712543,
2.8279807567596436,
3.049283504486084
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 14,
"value": 3.0037260243186243,
"params": {
"dropout": 0.37640619632798594,
"lr": 0.000983992977213618,
"weight_decay": 8.560558113400214e-05,
"backbone_lr_ratio": 0.029140753854155613,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 13,
"fold_best_epochs": [
12,
16,
11
],
"fold_val_losses": [
3.3224475847350226,
2.900958306259579,
2.7877721819612713
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 15,
"value": 3.057953337828318,
"params": {
"dropout": 0.4243012196347461,
"lr": 0.00031858611917698923,
"weight_decay": 1.457235563612646e-05,
"backbone_lr_ratio": 0.13443508445661562,
"moe_n_experts": 4,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 18,
"fold_best_epochs": [
18,
18,
17
],
"fold_val_losses": [
3.3081277741326227,
2.9643009768591986,
2.9014312624931335
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 16,
"value": 3.5674776236216226,
"params": {
"dropout": 0.10439567569750857,
"lr": 5.047116346834323e-05,
"weight_decay": 0.00012030457421620538,
"backbone_lr_ratio": 0.08403842125743101,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 32
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
19
],
"fold_val_losses": [
3.61348729663425,
3.4652820825576782,
3.6236634916729398
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 17,
"value": 3.2637910324114343,
"params": {
"dropout": 0.3121830815678499,
"lr": 0.00019221762216693134,
"weight_decay": 3.3001861372005635e-05,
"backbone_lr_ratio": 0.026623228883155377,
"moe_n_experts": 2,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 19,
"fold_best_epochs": [
20,
19,
18
],
"fold_val_losses": [
3.35848887430297,
3.204534967740377,
3.228349255190955
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 18,
"value": 2.932276170562815,
"params": {
"dropout": 0.20651560054419343,
"lr": 0.0004218332710981938,
"weight_decay": 0.00022119130962064922,
"backbone_lr_ratio": 0.2167274117185541,
"moe_n_experts": 8,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 15,
"fold_best_epochs": [
12,
16,
17
],
"fold_val_losses": [
3.367874321010378,
2.7223513391282825,
2.706602851549784
]
},
"state": "TrialState.COMPLETE"
},
{
"number": 19,
"value": 3.2775669528378386,
"params": {
"dropout": 0.34109501751685645,
"lr": 5.26451532204125e-05,
"weight_decay": 1.8996419945400298e-05,
"backbone_lr_ratio": 0.6387255624731281,
"moe_n_experts": 4,
"moe_top_k": 2,
"moe_expert_hidden_mult": 2,
"llm_lora_r": 8
},
"user_attrs": {
"epoch_mean": 20,
"fold_best_epochs": [
20,
20,
20
],
"fold_val_losses": [
3.412749081850052,
3.104751957787408,
3.3151998188760547
]
},
"state": "TrialState.COMPLETE"
}
]