diff --git a/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/checkpoint_final.pt b/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..4ad43b7b9f0b3f3315ee0bfa2a0dc668854b7b6e --- /dev/null +++ b/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e64ecdc33e3260309257bef47efb37bad92e7e4a5eb5295375bc23fda7be1e2 +size 438019989 diff --git a/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/metrics_history.json b/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..c69ff5738a5550ac26fd7cae40beebe8a6e6c30c --- /dev/null +++ b/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "mcc": 0.00 + }, + { + "epoch": 2, + "mcc": 0.00 + }, + { + "epoch": 3, + "mcc": 0.00 + } +] \ No newline at end of file diff --git a/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/params.json b/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/params.json new file mode 100644 index 0000000000000000000000000000000000000000..452b23fb320939614a7d42f9e0537e12132cd9c6 --- /dev/null +++ b/cola__SAM_Muon__lr2e-05__bs32__ep3__rho0.002/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.002, + "samRho": 0.002, + "asamRho": 0.002, + "esamRho": 0.002, + "looksamRho": 0.002, + "friendlysamRho": 0.002, + "fishersamRho": 0.002, + "mongesamRho": 0.002, + "gsamRho": 0.002, + "bayesiansamRho": 0.002, + "msamRho": 0.002, + "adamwSamRho": 0.002, + "adamSamRho": 0.002, + "adamwMsamRho": 0.002, + "adamwFriendlysamRho": 0.002, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "cola", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/cola__SAM__lr2e-05__bs32__ep3__rho0.002/checkpoint_final.pt b/cola__SAM__lr2e-05__bs32__ep3__rho0.002/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..fafce9d7c44c4d5da6ca450503f9af0d8455491c --- /dev/null +++ b/cola__SAM__lr2e-05__bs32__ep3__rho0.002/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6966b8c6c5229ea00168ec0ef1a5087b243b5a9b0e5732387c5302cfcc16998b +size 438019989 diff --git a/cola__SAM__lr2e-05__bs32__ep3__rho0.002/metrics_history.json b/cola__SAM__lr2e-05__bs32__ep3__rho0.002/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..f8f925ba13c41f93bf5e692eec413742adb857ce --- /dev/null +++ b/cola__SAM__lr2e-05__bs32__ep3__rho0.002/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "mcc": 4.191 + }, + { + "epoch": 2, + "mcc": -2.07 + }, + { + "epoch": 3, + "mcc": -2.07 + } +] \ No newline at end of file diff --git a/cola__SAM__lr2e-05__bs32__ep3__rho0.002/params.json b/cola__SAM__lr2e-05__bs32__ep3__rho0.002/params.json new file mode 100644 index 0000000000000000000000000000000000000000..f99338fe24107745082988a40f12e3cae2588c19 --- /dev/null +++ b/cola__SAM__lr2e-05__bs32__ep3__rho0.002/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.002, + "samRho": 0.002, + "asamRho": 0.002, + "esamRho": 0.002, + "looksamRho": 0.002, + "friendlysamRho": 0.002, + "fishersamRho": 0.002, + "mongesamRho": 0.002, + "gsamRho": 0.002, + "bayesiansamRho": 0.002, + "msamRho": 0.002, + "adamwSamRho": 0.002, + "adamSamRho": 0.002, + "adamwMsamRho": 0.002, + "adamwFriendlysamRho": 0.002, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "cola", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/cola__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/cola__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..440f66ed6aad122d4c3a851ba4590cc2eb9aeecd --- /dev/null +++ b/cola__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14b8edf27a50b5e2e8210b03ee1b852ab99c864752ec2a0003a99a3060a18084 +size 438019989 diff --git a/cola__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/cola__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..4097458097fbfaa57ef587038acb956ece342bd5 --- /dev/null +++ b/cola__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "mcc": 1.815 + }, + { + "epoch": 2, + "mcc": 1.815 + }, + { + "epoch": 3, + "mcc": 1.815 + } +] \ No newline at end of file diff --git a/cola__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/cola__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..76effd09b432e763671c8c164e9f02c4d56b93f2 --- /dev/null +++ b/cola__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "cola", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..db73c08431946b2631c49eed405a580cea2c443a --- /dev/null +++ b/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17e9f5f5a3d3849a252583b7fb6620bba91d177646f934c7152b694e113f3097 +size 438023061 diff --git a/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..d946b568ac9e50b923677e32beb5644aa3940ed8 --- /dev/null +++ b/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,20 @@ +[ + { + "epoch": 1, + "matched_accuracy": 66.623, + "mismatched_accuracy": 68.033, + "mnli_accuracy_mean": 67.328 + }, + { + "epoch": 2, + "matched_accuracy": 68.813, + "mismatched_accuracy": 70.179, + "mnli_accuracy_mean": 69.496 + }, + { + "epoch": 3, + "matched_accuracy": 69.791, + "mismatched_accuracy": 70.962, + "mnli_accuracy_mean": 70.376 + } +] \ No newline at end of file diff --git a/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json b/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..270409b2a344699d42dc0128b698a82340e9ac24 --- /dev/null +++ b/mnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "mnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..4b1276fba1bc79c3c3ad241a101d984e4294a526 --- /dev/null +++ b/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3745554ca0999b3436bc17b776218ff325eb06d836d50cf64f6e177bca40b2f +size 438023061 diff --git a/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..f59e0589d5c319ae8fe09b61fe5559038a20d429 --- /dev/null +++ b/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,20 @@ +[ + { + "epoch": 1, + "matched_accuracy": 42.649, + "mismatched_accuracy": 43.379, + "mnli_accuracy_mean": 43.014 + }, + { + "epoch": 2, + "matched_accuracy": 47.845, + "mismatched_accuracy": 49.247, + "mnli_accuracy_mean": 48.546 + }, + { + "epoch": 3, + "matched_accuracy": 49.292, + "mismatched_accuracy": 50.417, + "mnli_accuracy_mean": 49.855 + } +] \ No newline at end of file diff --git a/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/params.json b/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..124993f924319d3138b01604dd4df603997e26d9 --- /dev/null +++ b/mnli__SAM__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "mnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/mnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/mnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..c2c9254d837be95c73d6d14149a09bbd7d027fd8 --- /dev/null +++ b/mnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14e83e5cf89e24acadd580cce15adbdfece5de884d855088916bf85e93cdb4cc +size 438023061 diff --git a/mnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/mnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..8e6e4edc501499450c8059479c7c74e3b23fdfed --- /dev/null +++ b/mnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,20 @@ +[ + { + "epoch": 1, + "matched_accuracy": 43.617, + "mismatched_accuracy": 45.362, + "mnli_accuracy_mean": 44.489 + }, + { + "epoch": 2, + "matched_accuracy": 48.548, + "mismatched_accuracy": 50.376, + "mnli_accuracy_mean": 49.462 + }, + { + "epoch": 3, + "matched_accuracy": 49.71, + "mismatched_accuracy": 51.221, + "mnli_accuracy_mean": 50.465 + } +] \ No newline at end of file diff --git a/mnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/mnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..9b598ad429b4bf83dd089be162dcfb1064e02fad --- /dev/null +++ b/mnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "mnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..63ee35ac23d98d98af9baa1514945b6250ba6cee --- /dev/null +++ b/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4616778e98e335762523a8dfba87e9d3ef970d806a2282cd1c69784a08436b6a +size 438019989 diff --git a/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..856e3975c33246190c23aad8dbf531582a6be563 --- /dev/null +++ b/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "accuracy": 68.382, + "f1": 81.223 + }, + { + "epoch": 2, + "accuracy": 68.382, + "f1": 81.223 + }, + { + "epoch": 3, + "accuracy": 68.382, + "f1": 81.223 + } +] \ No newline at end of file diff --git a/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json b/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..f99f0d8bdf54fbd7ea3e227202279d73e8ffab41 --- /dev/null +++ b/mrpc__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "mrpc", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..8d5feb5010b6a7d125fa45597f763267e1042b45 --- /dev/null +++ b/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d29426cf54d9fe894382439ea33d4f823823c09f20426937ae04305780fbdf5c +size 438019989 diff --git a/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..0784119857a92deb9d8c87f362929eb669bf6bd2 --- /dev/null +++ b/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "accuracy": 68.137, + "f1": 81.05 + }, + { + "epoch": 2, + "accuracy": 68.382, + "f1": 81.223 + }, + { + "epoch": 3, + "accuracy": 68.382, + "f1": 81.223 + } +] \ No newline at end of file diff --git a/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/params.json b/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..020132823bbc4688a7653a40c7bece2291fb542c --- /dev/null +++ b/mrpc__SAM__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "mrpc", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..633a272bc41453e06bf7348735b13be7441e162c --- /dev/null +++ b/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5871ddc9ae62ff178690465754810966419b021786556ef6b4ac362f8837b2d9 +size 438019989 diff --git a/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..856e3975c33246190c23aad8dbf531582a6be563 --- /dev/null +++ b/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "accuracy": 68.382, + "f1": 81.223 + }, + { + "epoch": 2, + "accuracy": 68.382, + "f1": 81.223 + }, + { + "epoch": 3, + "accuracy": 68.382, + "f1": 81.223 + } +] \ No newline at end of file diff --git a/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..4d6a9d1af0add2ff5fef8a6fbc54cd79350988c8 --- /dev/null +++ b/mrpc__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "mrpc", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..87690328c0dcd8b5ba1504411d75623d326d4fe5 --- /dev/null +++ b/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dde3d7cac1c35cf1f30f06708ee04c2128faeb81184036d34013276088f1b04 +size 438019989 diff --git a/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..adae425ed952cc598cca36e3b4eba32f5ee4e79a --- /dev/null +++ b/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 79.791 + }, + { + "epoch": 2, + "accuracy": 80.432 + }, + { + "epoch": 3, + "accuracy": 80.981 + } +] \ No newline at end of file diff --git a/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json b/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..db1b05ca6f88e0665868e3ec2b0f0f02a1646e4a --- /dev/null +++ b/qnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "qnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4e71c9c7d49e951bdb16db47557cd3b06df9967 --- /dev/null +++ b/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02b876fcd8afd89aa7c65133ada0d2ab06acb3ec0414f9be91bdb0d673292c80 +size 438019989 diff --git a/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..e82a052f45caf54069dced68aca5c911d461aeeb --- /dev/null +++ b/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 69.028 + }, + { + "epoch": 2, + "accuracy": 70.913 + }, + { + "epoch": 3, + "accuracy": 71.572 + } +] \ No newline at end of file diff --git a/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/params.json b/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..7672368647868531f921b479ff10c4902d091440 --- /dev/null +++ b/qnli__SAM__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "qnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/qnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/qnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..0384e079552b535f71389be429de40c9d1701f56 --- /dev/null +++ b/qnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00e43002ce2e69bcfeb4f6f098ee458c3cc8aa41ebd0987fff38c7d0fe251e7c +size 438019989 diff --git a/qnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/qnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..b5ba520361ea6809667060b016ab6244564064b3 --- /dev/null +++ b/qnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 68.9 + }, + { + "epoch": 2, + "accuracy": 71.682 + }, + { + "epoch": 3, + "accuracy": 72.305 + } +] \ No newline at end of file diff --git a/qnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/qnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..aae30040ee74be17a49f1f73c0674f5da0fef4b2 --- /dev/null +++ b/qnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "qnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..0d48b962f85c393cde300a245b3b8e8d53a0a4c9 --- /dev/null +++ b/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:584b8028b067f59aa6297402ac078381058e98ca23acf6d4810a03dbcbd10795 +size 438019989 diff --git a/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..72f07292a2fa6dedd34f413702bc878b3717e2f1 --- /dev/null +++ b/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "accuracy": 80.759, + "f1": 74.723 + }, + { + "epoch": 2, + "accuracy": 81.959, + "f1": 76.102 + }, + { + "epoch": 3, + "accuracy": 82.29, + "f1": 76.452 + } +] \ No newline at end of file diff --git a/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json b/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..13f8856a6139a5dcd02144992e2275d654b4c709 --- /dev/null +++ b/qqp__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "qqp", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..bd5b7542e28a9cd1d41883c642c4ce3f4f1f4e2b --- /dev/null +++ b/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af8ad46e1ee33dabe90a14bc6d48709d462bb05d408e5130fa3c9a94697b7ef7 +size 438019989 diff --git a/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..c92b1e425be38f4d673164b0f856082ebaf9298c --- /dev/null +++ b/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "accuracy": 70.994, + "f1": 66.906 + }, + { + "epoch": 2, + "accuracy": 73.26, + "f1": 70.156 + }, + { + "epoch": 3, + "accuracy": 73.876, + "f1": 70.638 + } +] \ No newline at end of file diff --git a/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/params.json b/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..dda41163f305d1433e6946f5152b1b1bb843aa31 --- /dev/null +++ b/qqp__SAM__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "qqp", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/qqp__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/qqp__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..f5b209ec755b954ad738bb55f6cf54a5ff2f0a45 --- /dev/null +++ b/qqp__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae328d07cd59416db72a8b1d3fdb1658350b4f6a5292ae668d141ab4b1e11332 +size 438019989 diff --git a/qqp__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/qqp__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..bc719dcd873d4984be7ce42cab8401d3984e0e96 --- /dev/null +++ b/qqp__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "accuracy": 70.952, + "f1": 65.703 + }, + { + "epoch": 2, + "accuracy": 73.208, + "f1": 70.291 + }, + { + "epoch": 3, + "accuracy": 73.93, + "f1": 70.908 + } +] \ No newline at end of file diff --git a/qqp__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/qqp__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..bec59d1e73a574d6b7cb5695a4d3b887708edf48 --- /dev/null +++ b/qqp__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "qqp", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt b/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..4e1d28506d0472294755cf17e3c233bec56f10f0 --- /dev/null +++ b/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:944f38c4f637c231bf5280993d262e8930e575a8334701d12305983e5bee8914 +size 438019989 diff --git a/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json b/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..88dcacc3bba3564ea5b42283f6b64114e70035a9 --- /dev/null +++ b/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 47.292 + }, + { + "epoch": 2, + "accuracy": 47.292 + }, + { + "epoch": 3, + "accuracy": 47.292 + } +] \ No newline at end of file diff --git a/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json b/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json new file mode 100644 index 0000000000000000000000000000000000000000..9b938c8fa88799527d023725e66c9930b2e4d029 --- /dev/null +++ b/rte__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.004, + "samRho": 0.004, + "asamRho": 0.004, + "esamRho": 0.004, + "looksamRho": 0.004, + "friendlysamRho": 0.004, + "fishersamRho": 0.004, + "mongesamRho": 0.004, + "gsamRho": 0.004, + "bayesiansamRho": 0.004, + "msamRho": 0.004, + "adamwSamRho": 0.004, + "adamSamRho": 0.004, + "adamwMsamRho": 0.004, + "adamwFriendlysamRho": 0.004, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "rte", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/rte__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt b/rte__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..3efdc6a55da721559e1389a4b3b9dbc961af4572 --- /dev/null +++ b/rte__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36aa572affaf64f610267f60042ec71f861250fedf711fbd6c93cbbfc907d466 +size 438019989 diff --git a/rte__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json b/rte__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..88dcacc3bba3564ea5b42283f6b64114e70035a9 --- /dev/null +++ b/rte__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 47.292 + }, + { + "epoch": 2, + "accuracy": 47.292 + }, + { + "epoch": 3, + "accuracy": 47.292 + } +] \ No newline at end of file diff --git a/rte__SAM__lr2e-05__bs32__ep3__rho0.004/params.json b/rte__SAM__lr2e-05__bs32__ep3__rho0.004/params.json new file mode 100644 index 0000000000000000000000000000000000000000..122e425bbadc46be1b0dde9a89f5610a13f987ef --- /dev/null +++ b/rte__SAM__lr2e-05__bs32__ep3__rho0.004/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.004, + "samRho": 0.004, + "asamRho": 0.004, + "esamRho": 0.004, + "looksamRho": 0.004, + "friendlysamRho": 0.004, + "fishersamRho": 0.004, + "mongesamRho": 0.004, + "gsamRho": 0.004, + "bayesiansamRho": 0.004, + "msamRho": 0.004, + "adamwSamRho": 0.004, + "adamSamRho": 0.004, + "adamwMsamRho": 0.004, + "adamwFriendlysamRho": 0.004, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "rte", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/rte__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/rte__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..a682467dcd8fa0244f6fb5c8f3095c4ffb7c7603 --- /dev/null +++ b/rte__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5b101488f8b75f52bcf14a019f8eb5ac6917a58f9f908eed2c0d540d3aa8a38 +size 438019989 diff --git a/rte__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/rte__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..831955673596b256c790aaa75f9bd26397057bb2 --- /dev/null +++ b/rte__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 53.069 + }, + { + "epoch": 2, + "accuracy": 53.791 + }, + { + "epoch": 3, + "accuracy": 54.152 + } +] \ No newline at end of file diff --git a/rte__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/rte__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..16e17864a1e73ee47b1e6d972e9959d83452e54a --- /dev/null +++ b/rte__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "rte", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..3738650783cfc7cb5770cb942421ea67d2db48dc --- /dev/null +++ b/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e044da9e3d0bf6cfb423bbc51fceebe2300cbfce5d4bea824e04b19a0bbd0588 +size 438019989 diff --git a/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..954511f0b68a430ddb0b9816af4947988b51cdf0 --- /dev/null +++ b/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 87.271 + }, + { + "epoch": 2, + "accuracy": 88.417 + }, + { + "epoch": 3, + "accuracy": 88.417 + } +] \ No newline at end of file diff --git a/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json b/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..d0f982a6bc891bdb9e86df7b0f29538f62c2fc72 --- /dev/null +++ b/sst2__SAM_Muon__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "sst2", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt b/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..4d5ccd079ec589d042f37f2c67ef338c47485685 --- /dev/null +++ b/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2bb8b26bc9e3a4aedd5aa0256212e5cc26db7e7d82363a75074a6c02045fc88 +size 438019989 diff --git a/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json b/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..79dbabc35ccea1dc85e8c42b9512c8f2b0d7903f --- /dev/null +++ b/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 51.032 + }, + { + "epoch": 2, + "accuracy": 52.982 + }, + { + "epoch": 3, + "accuracy": 54.702 + } +] \ No newline at end of file diff --git a/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/params.json b/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/params.json new file mode 100644 index 0000000000000000000000000000000000000000..8debd33196213fd6e9742ab9a965d6a5bca53e0d --- /dev/null +++ b/sst2__SAM__lr2e-05__bs32__ep3__rho0.005/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.005, + "samRho": 0.005, + "asamRho": 0.005, + "esamRho": 0.005, + "looksamRho": 0.005, + "friendlysamRho": 0.005, + "fishersamRho": 0.005, + "mongesamRho": 0.005, + "gsamRho": 0.005, + "bayesiansamRho": 0.005, + "msamRho": 0.005, + "adamwSamRho": 0.005, + "adamSamRho": 0.005, + "adamwMsamRho": 0.005, + "adamwFriendlysamRho": 0.005, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "sst2", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/sst2__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/sst2__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..e4ab7095399ffbe0302f9ed05e60e86cf4bb0b00 --- /dev/null +++ b/sst2__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51f73bcefd717e35f04e3b43cd57345ecd7fe2386b24a66c75f68cd2fe7ab6ec +size 438019989 diff --git a/sst2__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/sst2__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..1bc1452c6dfc7c45dbb3c1de13cd952f49748174 --- /dev/null +++ b/sst2__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 51.032 + }, + { + "epoch": 2, + "accuracy": 52.523 + }, + { + "epoch": 3, + "accuracy": 54.128 + } +] \ No newline at end of file diff --git a/sst2__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/sst2__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..123102ec19785241de784dc39af7b0b20120ae0b --- /dev/null +++ b/sst2__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "sst2", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt b/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..758076464ef2e8d3c7ac232aa1d16655c47c53a5 --- /dev/null +++ b/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:073d8e9184133bace92ff40f8a1db1e17ed4dbf502e86000d58131cb72d269d2 +size 438016917 diff --git a/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json b/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..55a430bc3ba37b57abeb93c33c096dab3eced6d4 --- /dev/null +++ b/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "pearson": 23.811, + "spearman": 47.492 + }, + { + "epoch": 2, + "pearson": 24.397, + "spearman": 65.806 + }, + { + "epoch": 3, + "pearson": 25.926, + "spearman": 72.837 + } +] \ No newline at end of file diff --git a/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json b/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json new file mode 100644 index 0000000000000000000000000000000000000000..2cc91c2e3a278db62069c68b39ef948bdadcddc6 --- /dev/null +++ b/stsb__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.004, + "samRho": 0.004, + "asamRho": 0.004, + "esamRho": 0.004, + "looksamRho": 0.004, + "friendlysamRho": 0.004, + "fishersamRho": 0.004, + "mongesamRho": 0.004, + "gsamRho": 0.004, + "bayesiansamRho": 0.004, + "msamRho": 0.004, + "adamwSamRho": 0.004, + "adamSamRho": 0.004, + "adamwMsamRho": 0.004, + "adamwFriendlysamRho": 0.004, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "stsb", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt b/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..8bf2bd6878469ea86afb70d17866db5fabdd1b3b --- /dev/null +++ b/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac29cf5cf4e5d0ae978a79032646c888f110ed7d8192f1c104d96cc06ba46506 +size 438016917 diff --git a/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json b/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..44b7e04b831cfe200473740ce69a7bd7cba1ee4f --- /dev/null +++ b/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "pearson": 45.691, + "spearman": 63.906 + }, + { + "epoch": 2, + "pearson": 43.929, + "spearman": 65.14 + }, + { + "epoch": 3, + "pearson": 43.526, + "spearman": 65.452 + } +] \ No newline at end of file diff --git a/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/params.json b/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/params.json new file mode 100644 index 0000000000000000000000000000000000000000..db785d836c2759ab5c2fa5bc3b73274aad360d66 --- /dev/null +++ b/stsb__SAM__lr2e-05__bs32__ep3__rho0.004/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.004, + "samRho": 0.004, + "asamRho": 0.004, + "esamRho": 0.004, + "looksamRho": 0.004, + "friendlysamRho": 0.004, + "fishersamRho": 0.004, + "mongesamRho": 0.004, + "gsamRho": 0.004, + "bayesiansamRho": 0.004, + "msamRho": 0.004, + "adamwSamRho": 0.004, + "adamSamRho": 0.004, + "adamwMsamRho": 0.004, + "adamwFriendlysamRho": 0.004, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "stsb", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/stsb__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/stsb__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..61eb178bc12f0ad5171dd91cd83dfc4c58f9fbdf --- /dev/null +++ b/stsb__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e87601d9a9e8e30a757e72c8d43a12a2c21550fa97ab980e9ff393b3a2c819a0 +size 438016917 diff --git a/stsb__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/stsb__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..e0cd196e15fbfcd2c4af8b99b86a09a08c9c7a01 --- /dev/null +++ b/stsb__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,17 @@ +[ + { + "epoch": 1, + "pearson": 36.141, + "spearman": 42.274 + }, + { + "epoch": 2, + "pearson": 38.231, + "spearman": 58.721 + }, + { + "epoch": 3, + "pearson": 38.49, + "spearman": 61.073 + } +] \ No newline at end of file diff --git a/stsb__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/stsb__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..646730f048b2a2f4e03499abfc1df79e544d5f59 --- /dev/null +++ b/stsb__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "stsb", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt b/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..80dd246c6c700696410ba130d64405cf5f1368d9 --- /dev/null +++ b/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac91f7f844946ec3d5c34f8c327a82e2f665d19d54951d14e9a4d967e612abf0 +size 438019989 diff --git a/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json b/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..7ec84477085196e8e2119eed9590d9e86375ad67 --- /dev/null +++ b/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 56.338 + }, + { + "epoch": 2, + "accuracy": 56.338 + }, + { + "epoch": 3, + "accuracy": 56.338 + } +] \ No newline at end of file diff --git a/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json b/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json new file mode 100644 index 0000000000000000000000000000000000000000..0cafed33367cb44e1b730e69c2dedbd4c606c1cf --- /dev/null +++ b/wnli__SAM_Muon__lr2e-05__bs32__ep3__rho0.004/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM_Muon", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.004, + "samRho": 0.004, + "asamRho": 0.004, + "esamRho": 0.004, + "looksamRho": 0.004, + "friendlysamRho": 0.004, + "fishersamRho": 0.004, + "mongesamRho": 0.004, + "gsamRho": 0.004, + "bayesiansamRho": 0.004, + "msamRho": 0.004, + "adamwSamRho": 0.004, + "adamSamRho": 0.004, + "adamwMsamRho": 0.004, + "adamwFriendlysamRho": 0.004, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "wnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt b/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..8b9e82f521f718d7adefd8322a2da10bcb504a2c --- /dev/null +++ b/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ce8fd3ccbbe01e1975bbf45cb840a44a747179fe202d2befac85ee3029d4179 +size 438019989 diff --git a/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json b/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..7ec84477085196e8e2119eed9590d9e86375ad67 --- /dev/null +++ b/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 56.338 + }, + { + "epoch": 2, + "accuracy": 56.338 + }, + { + "epoch": 3, + "accuracy": 56.338 + } +] \ No newline at end of file diff --git a/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/params.json b/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/params.json new file mode 100644 index 0000000000000000000000000000000000000000..822ff6be601227b8b869a4f71c9422443f3fdd48 --- /dev/null +++ b/wnli__SAM__lr2e-05__bs32__ep3__rho0.004/params.json @@ -0,0 +1,99 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SAM", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.004, + "samRho": 0.004, + "asamRho": 0.004, + "esamRho": 0.004, + "looksamRho": 0.004, + "friendlysamRho": 0.004, + "fishersamRho": 0.004, + "mongesamRho": 0.004, + "gsamRho": 0.004, + "bayesiansamRho": 0.004, + "msamRho": 0.004, + "adamwSamRho": 0.004, + "adamSamRho": 0.004, + "adamwMsamRho": 0.004, + "adamwFriendlysamRho": 0.004, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "wnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file diff --git a/wnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt b/wnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt new file mode 100644 index 0000000000000000000000000000000000000000..de11f56ab7ce5a7172a9be3656ac55b70ef98c4e --- /dev/null +++ b/wnli__SGD__lr2e-05__bs32__ep3__rho_na/checkpoint_final.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2531931219e17e69add2035718ebd9b5b157d7ef38fb82898f89bdf9eb8d69b8 +size 438019989 diff --git a/wnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json b/wnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json new file mode 100644 index 0000000000000000000000000000000000000000..7ec84477085196e8e2119eed9590d9e86375ad67 --- /dev/null +++ b/wnli__SGD__lr2e-05__bs32__ep3__rho_na/metrics_history.json @@ -0,0 +1,14 @@ +[ + { + "epoch": 1, + "accuracy": 56.338 + }, + { + "epoch": 2, + "accuracy": 56.338 + }, + { + "epoch": 3, + "accuracy": 56.338 + } +] \ No newline at end of file diff --git a/wnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json b/wnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json new file mode 100644 index 0000000000000000000000000000000000000000..f28b1feb1ae37702d4ff1044d94c5692943c95b3 --- /dev/null +++ b/wnli__SGD__lr2e-05__bs32__ep3__rho_na/params.json @@ -0,0 +1,98 @@ +{ + "label_smoothing": 0.0, + "ESAMbeta": 0.5, + "ESAMgamma": 0.5, + "lookSAMk": 5, + "lookSAMalpha": 0.7, + "lookSAMLayerwise": false, + "optimizer": "SGD", + "weightDecay": 0.01, + "momentum": 0.9, + "rho": 0.05, + "samRho": 0.05, + "asamRho": 0.05, + "esamRho": 0.05, + "looksamRho": 0.05, + "friendlysamRho": 0.05, + "fishersamRho": 0.05, + "mongesamRho": 0.05, + "gsamRho": 0.05, + "bayesiansamRho": 0.05, + "msamRho": 0.05, + "adamwSamRho": 0.05, + "adamwMsamRho": 0.05, + "adamwFriendlysamRho": 0.05, + "sigma": 1.0, + "lmbda": 0.9, + "fisherEta": 1.0, + "gsamAlpha": 0.01, + "gsamAdaptive": false, + "bayesMSharpness": 8, + "bayesDelta": 10.0, + "bayesBeta1": 0.9, + "bayesBeta2": 0.999, + "bayesSInit": 1.0, + "bayesDamping": 1e-08, + "bayesEps": 1e-12, + "nesterov": false, + "grad_clip_norm": 1.0, + "rndSeed": true, + "deterministic": false, + "glue_task": "wnli", + "glue_output_root": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "widthFactor": 4, + "ViTPatchsize": 32, + "ViTSize": "S", + "pyramidAlpha": 270, + "pyramidBottleneck": true, + "model": "WRN", + "dropout": 0.0, + "BN": true, + "depth": 16, + "dataThreads": 4, + "dataDir": "/home/hungpv/.datasets", + "batchSize": 32, + "dataset": "CIFAR100", + "imageSize": 0, + "flip": false, + "crop": false, + "cut": false, + "cutoutProp": 0.5, + "randAugment": false, + "randAugment_magnitude": 10, + "mixup": false, + "mixupProp": 0.2, + "normalize": "std", + "truncate": true, + "verbose": true, + "metrics": [ + "learningRate", + "loss", + "accuracy" + ], + "logEach": 50, + "learningRate": 2e-05, + "lrScheduler": [ + "const" + ], + "LRScheduler_step_steps": 3.0, + "LRScheduler_step_gamma": 0.2, + "LRScheduler_exp_maxDecay": 0.01, + "LRScheduler_cos_periods": 0.5, + "LRScheduler_StartRamp_epochs": 1.0, + "LRScheduler_WRN_T0": 50.0, + "LRScheduler_WRN_Tmult": 1.2, + "rhoScheduler": [ + "const" + ], + "rhoScheduler_StartRamp_epochs": 8.0, + "rhoScheduler_StartJump_epochs": 8.0, + "saveCheckpoint": false, + "keepLastCheckpoint": false, + "checkpointsList": [], + "saveCheckpointInterval": 0, + "logDir": "/mnt/hungpv/projects/SAM_Variant/logs_nlp", + "logSubDir": "glue_bert", + "epochs": 3, + "contin": false +} \ No newline at end of file