Upload folder using huggingface_hub

parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0/seed_97/checkpoints/checkpoints.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e82c9a56a0e4634589608ad47787a72cd37ccc446e22221e28880f2be3177328
+size 130

parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0/seed_97/config.yaml

@@ -0,0 +1,135 @@
+seed: 0
+runs_folder: /media/god/Second/Runs
+huggingface:
+  repo_id: kclauw/grokking-oinformation
+synergymask: false
+sparsity_sampling: 10
+scheduler: null
+device: cpu
+test_seed: 123
+evaluate_oinformation: true
+evaluate_sparsity: false
+evaluate_norms: false
+evaluate_subnetworks: false
+evaluate_lottery_ticket: false
+create_grid_search_file: false
+create_plots: true
+store_local: true
+experiment_type: train_single
+unique_plot_path: fcn_relu/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0
+store_on_scratch: true
+oinformation:
+  sampling_rate: 1
+  max_batch_greedy: 40
+  layer: fc2_post
+  njobs: 16
+  layers:
+  - fc3_post
+  train: true
+  features_type:
+  - train
+  max_batch_exhaustive: 3
+loss:
+  _target_: model.neural_network.MyHingeLoss
+train:
+  lr: 0.2
+  train_batch_size: 32
+  num_workers: 6
+  eval_batch_size: 32
+  max_epochs: 600
+regularization:
+  weight_decay:
+    name: wd
+    value: 0.01
+    type: null
+  dropout:
+    name: do
+    value: 0
+    type: dropout
+  batchnorm:
+    name: bn
+    value: false
+    type: batchnorm
+  oinfo_syn:
+    name: syn
+    value: 0
+    type: null
+  layernorm:
+    name: ln
+    value: false
+    type: layernorm
+  oinfo_red:
+    name: red
+    value: 0
+wandb:
+  enabled: false
+  project: grokking_replica
+unique_filename: parity_n_40_k_3_N_1600_100#fcn_relu_20_20_20_0.00_default#sgd_bs_32_lr_0.20#wd_0.01_do_0_no_bn_syn_0_no_ln_red_0
+regularization_name: null
+dataset:
+  train_samples: 1600
+  test_samples: 100
+  parameters:
+    _target_: dataloaders.sparse_parity.parity
+    'n': 40
+    k: 3
+  name: parity_n_${dataset.parameters.n}_k_${dataset.parameters.k}_N_${dataset.train_samples}_${dataset.test_samples}
+model:
+  parameters:
+    _target_: src.model.neural_network.DynamicFFN
+    activation: relu
+    initialization: default
+    layers:
+      fc1: 20
+      fc2: 20
+      fc3: 20
+    initialization_scale: 0.0
+    constrained_norm: false
+  name: fcn_${model.parameters.activation}
+  layer_names:
+  - fc1
+  - fc2
+  - fc3
+optimizer:
+  name: sgd
+  parameters:
+    _target_: torch.optim.SGD
+    weight_decay: ${regularization.weight_decay.value}
+paths:
+  root_dir: .
+  data_dir: ${paths.root_dir}/data/
+  log_dir: ${paths.root_dir}/runs/
+  output_dir: /kyukon/scratch/gent/433/vsc43397/oinformation-grokking/runs/parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0
+  plot_dir: ./plots/1_baseline_datasize_vs_lr_small_100_neurons
+  plot_dir_all: ./plots/1_baseline_datasize_vs_lr_small_100_neurons/all/fcn_relu/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0
+  work_dir: ${hydra:runtime.cwd}
+experiment_name: 1_baseline_datasize_vs_lr_small_100_neurons
+seeds:
+- 97
+evaluate_generalizing_models_only: false
+remove_files: true
+plots:
+  experiment1:
+    title:
+    - train.max_epochs
+    type: heatmap
+    x: dataset.train_samples
+    x_label: dataset size
+    y_label: learning rate
+    'y':
+    - test_acc
+    - synergy
+    - redundancy
+grid_search:
+  dataset.train_samples:
+    scale: null
+    min: 800
+    max: 2000
+    num: 4
+    type: integer
+  train.lr:
+    min: 0.01
+    max: 0.2
+    num: 6
+    type: uniform
+experiment_oinfo_title: null

parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0/seed_97/epoch_results_600_seed_97.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04899b9643aa498d22194a67e97bf855a34258f60da00c6523511bfc3ed8741e
+size 53705

parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0/seed_97/oinformation_results_train_600.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9c3f48435004c6d6320dabd4edb7e892da37160e8a3282ffcf995f75c3acfa5
+size 2853528

parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0/seed_97/optimal_oinformation_results_train_600.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f696a6990be07fa609db2909b9e9d82a42096598f500dbd90ad5c54987a1f89
+size 371997

parity_n_40_k_3_N_1600_100/fcn_relu_20_20_20_0.00_default/sgd_bs_32_lr_0.20/wd_0.01_do_0_no_bn_syn_0_no_ln_red_0/seed_97/optimal_results_seed_600_97.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33d1a4931adb924dfe0bfb605dff4cfd3fd77a3ed6fe5ca0692991552f30aa3b
+size 97