added transformer on 1M events

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/.gitattributes

@@ -0,0 +1,4 @@
+plots_checkpoint*/** filter=lfs diff=lfs merge=lfs -text
+preds_checkpoint*/** filter=lfs diff=lfs merge=lfs -text
+runs/** filter=lfs diff=lfs merge=lfs -text
+checkpoints/** filter=lfs diff=lfs merge=lfs -text

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/checkpoints/checkpoint-10-2.326306.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7ccf57c386b27378589988bb8138990fc4b6167044fe1307cb3679bbaded995
+size 631766830

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_1.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.1809584355065391, "Regression_eta": 0.00021549460627068675, "Regression_sin_phi": 0.00014699551204219778, "Regression_cos_phi": 0.00014780340473040274, "Regression_energy": 0.18014646687387992, "Classification_binary": 3.0907964776412356, "Classification": 0.028157859394667924, "ispu": 6.739435331836497e-05, "MET": 7.605959448071343, "Sliced_Wasserstein_Loss": 52.211901561699946, "Total": 3.4806370941073563}, "valid": {"Regression_pt": 0.1560728563476249, "Regression_eta": 0.00021242698756363145, "Regression_sin_phi": 0.0001405400746236246, "Regression_cos_phi": 0.00014247594063863839, "Regression_energy": 0.1566179217763721, "Classification_binary": 2.670871168556756, "Classification": 0.02148569404371055, "ispu": 2.2155036000330433e-10, "MET": 5.439039180351984, "Sliced_Wasserstein_Loss": 47.2730354659249, "Total": 3.005548655317214}, "epoch_train_time": 3617.600632429123, "epoch_valid_time": 1040.5482785701752, "epoch_total_time": 4658.148911952972}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_10.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.09964167915079898, "Regression_eta": 0.0001397088009824035, "Regression_sin_phi": 9.003577124118236e-05, "Regression_cos_phi": 9.014389176766688e-05, "Regression_energy": 0.09931737714547065, "Classification_binary": 2.034971738191244, "Classification": 0.015284989958107318, "ispu": 3.0009371938361737e-15, "MET": 3.5108390270310634, "Sliced_Wasserstein_Loss": 38.934281127325484, "Total": 2.2495388038167774}, "valid": {"Regression_pt": 0.1060637693118447, "Regression_eta": 0.0001479885827363768, "Regression_sin_phi": 9.52059819352197e-05, "Regression_cos_phi": 9.531670998185283e-05, "Regression_energy": 0.10615338592512973, "Classification_binary": 2.0983671231411147, "Classification": 0.015383389718751672, "ispu": 0.0, "MET": 3.5345486018241146, "Sliced_Wasserstein_Loss": 38.800698084947044, "Total": 2.3263056077484756}, "epoch_train_time": 3630.1035368442535, "epoch_valid_time": 1033.9881041049957, "epoch_total_time": 4664.091645956039}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_2.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.1406924939029431, "Regression_eta": 0.00018237418008554631, "Regression_sin_phi": 0.00012451623930298726, "Regression_cos_phi": 0.00012471466712281044, "Regression_energy": 0.14004935899230883, "Classification_binary": 2.4912077706306537, "Classification": 0.019820465124351292, "ispu": 2.6400675580940582e-11, "MET": 4.647695785863629, "Sliced_Wasserstein_Loss": 46.33445340501792, "Total": 2.7922045344235364}, "valid": {"Regression_pt": 0.13530600194664733, "Regression_eta": 0.00017553599133036973, "Regression_sin_phi": 0.0001195417542731531, "Regression_cos_phi": 0.00012011609384053734, "Regression_energy": 0.13530850943568123, "Classification_binary": 2.4270329049427626, "Classification": 0.01862096128405588, "ispu": 1.129222483740377e-13, "MET": 4.420098444019471, "Sliced_Wasserstein_Loss": 44.3398951535252, "Total": 2.7166894708863807}, "epoch_train_time": 3628.9076433181763, "epoch_valid_time": 1031.0589447021484, "epoch_total_time": 4659.966592311859}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_3.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.12660502980524088, "Regression_eta": 0.0001660489063455009, "Regression_sin_phi": 0.00011171370281825641, "Regression_cos_phi": 0.00011188287258066808, "Regression_energy": 0.1261319682793032, "Classification_binary": 2.348539806547619, "Classification": 0.018299353450287485, "ispu": 2.962162467579577e-15, "MET": 4.243572255291006, "Sliced_Wasserstein_Loss": 43.912663743812935, "Total": 2.619972011542072}, "valid": {"Regression_pt": 0.12495328495690462, "Regression_eta": 0.0001672271790957979, "Regression_sin_phi": 0.00011214603392647028, "Regression_cos_phi": 0.00011173559278425105, "Regression_energy": 0.12492733505737134, "Classification_binary": 2.33832316685835, "Classification": 0.01784987085909082, "ispu": 0.0, "MET": 4.331618768053921, "Sliced_Wasserstein_Loss": 43.28514830961806, "Total": 2.6064431184470953}, "epoch_train_time": 3631.964106321335, "epoch_valid_time": 1034.2509853839874, "epoch_total_time": 4666.215092658997}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_4.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.11904703567694999, "Regression_eta": 0.00015782828745245425, "Regression_sin_phi": 0.00010465266334662606, "Regression_cos_phi": 0.00010478338867268234, "Regression_energy": 0.11857049476166474, "Classification_binary": 2.2651194676459294, "Classification": 0.017436679749587305, "ispu": 0.0, "MET": 4.037996098416965, "Sliced_Wasserstein_Loss": 42.513486196449904, "Total": 2.520537481065455}, "valid": {"Regression_pt": 0.12028810160930646, "Regression_eta": 0.00016301201411947485, "Regression_sin_phi": 0.00010597006935633044, "Regression_cos_phi": 0.00010645244801836472, "Regression_energy": 0.12029852336060368, "Classification_binary": 2.270865443123462, "Classification": 0.01701379630914331, "ispu": 0.0, "MET": 3.9944956436557186, "Sliced_Wasserstein_Loss": 42.347049855568635, "Total": 2.52884364635712}, "epoch_train_time": 3629.249868631363, "epoch_valid_time": 1041.715768814087, "epoch_total_time": 4670.965641975403}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_5.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.11393740657735428, "Regression_eta": 0.00015272132201617154, "Regression_sin_phi": 0.00010009012565736499, "Regression_cos_phi": 0.00010030442363350318, "Regression_energy": 0.11347376232078853, "Classification_binary": 2.204909947569978, "Classification": 0.016820606466730538, "ispu": 0.0, "MET": 3.891115364716675, "Sliced_Wasserstein_Loss": 41.56457213261649, "Total": 2.449494934342465}, "valid": {"Regression_pt": 0.11554292674955868, "Regression_eta": 0.00015717767146330818, "Regression_sin_phi": 0.00010437786585508954, "Regression_cos_phi": 0.00010657853157849086, "Regression_energy": 0.11557501477328956, "Classification_binary": 2.221271280223601, "Classification": 0.01655093930016817, "ispu": 0.0, "MET": 4.140659269016797, "Sliced_Wasserstein_Loss": 41.132960441853, "Total": 2.4693139760752114}, "epoch_train_time": 3634.312597990036, "epoch_valid_time": 1034.1657779216766, "epoch_total_time": 4668.478380680084}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_6.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.10986073942065626, "Regression_eta": 0.0001487721760416625, "Regression_sin_phi": 9.663510106899445e-05, "Regression_cos_phi": 9.673728046003798e-05, "Regression_energy": 0.10941738812950695, "Classification_binary": 2.1566410250256016, "Classification": 0.016338832143590522, "ispu": 0.0, "MET": 3.768947879330944, "Sliced_Wasserstein_Loss": 40.79141865079365, "Total": 2.3925969595387437}, "valid": {"Regression_pt": 0.112498247369644, "Regression_eta": 0.00015528222207807787, "Regression_sin_phi": 0.00010116737105954301, "Regression_cos_phi": 0.00010034016929259973, "Regression_energy": 0.11251851623933481, "Classification_binary": 2.1800310259976463, "Classification": 0.016144455085020593, "ispu": 0.0, "MET": 3.7187006860489995, "Sliced_Wasserstein_Loss": 40.621088317107095, "Total": 2.421554459318498}, "epoch_train_time": 3629.0600349903107, "epoch_valid_time": 1035.2439005374908, "epoch_total_time": 4664.303944826126}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_7.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.10637912076372888, "Regression_eta": 0.00014547484587190663, "Regression_sin_phi": 9.403404348802803e-05, "Regression_cos_phi": 9.414262180748318e-05, "Regression_energy": 0.1059651046833664, "Classification_binary": 2.1153307144990614, "Classification": 0.015953127771010676, "ispu": 0.0, "MET": 3.6691988287250386, "Sliced_Wasserstein_Loss": 40.15710072111282, "Total": 2.343960680150196}, "valid": {"Regression_pt": 0.10971037665828608, "Regression_eta": 0.00015167487525960348, "Regression_sin_phi": 9.770144418406463e-05, "Regression_cos_phi": 9.772997432001468e-05, "Regression_energy": 0.10966003098002701, "Classification_binary": 2.146385120225741, "Classification": 0.01581318574601142, "ispu": 0.0, "MET": 3.6153520180271745, "Sliced_Wasserstein_Loss": 39.532547207660215, "Total": 2.381919624946507}, "epoch_train_time": 3633.2319824695587, "epoch_valid_time": 1031.886385679245, "epoch_total_time": 4665.11837387085}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_8.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.1034678719438044, "Regression_eta": 0.00014290572189393558, "Regression_sin_phi": 9.217529212234166e-05, "Regression_cos_phi": 9.227262908107369e-05, "Regression_energy": 0.10308563914757211, "Classification_binary": 2.079565925552569, "Classification": 0.015644156173911797, "ispu": 0.0, "MET": 3.5928636099377025, "Sliced_Wasserstein_Loss": 39.62373325226148, "Total": 2.3020950007467142}, "valid": {"Regression_pt": 0.1075918879804349, "Regression_eta": 0.00014891935132003102, "Regression_sin_phi": 9.639054374209161e-05, "Regression_cos_phi": 9.628614793171204e-05, "Regression_energy": 0.10750513800174856, "Classification_binary": 2.1245576371429338, "Classification": 0.01563770232563339, "ispu": 0.0, "MET": 3.5618814860383012, "Sliced_Wasserstein_Loss": 39.40250213972398, "Total": 2.3556396771691452}, "epoch_train_time": 3629.116830587387, "epoch_valid_time": 1037.676260471344, "epoch_total_time": 4666.79309296608}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/history/epoch_9.json

@@ -0,0 +1 @@
+{"train": {"Regression_pt": 0.10115222374231951, "Regression_eta": 0.00014090951055893618, "Regression_sin_phi": 9.083634941503443e-05, "Regression_cos_phi": 9.092952554432233e-05, "Regression_energy": 0.10080205549821855, "Classification_binary": 2.052080166463987, "Classification": 0.015418790187988697, "ispu": 2.8972061186000455e-15, "MET": 3.537445063150708, "Sliced_Wasserstein_Loss": 39.19917541389316, "Total": 2.2697752656169996}, "valid": {"Regression_pt": 0.10639774203746857, "Regression_eta": 0.0001481339372108617, "Regression_sin_phi": 9.552524652763703e-05, "Regression_cos_phi": 9.549498825873157e-05, "Regression_energy": 0.10664135373900048, "Classification_binary": 2.103942942922863, "Classification": 0.015437925972759475, "ispu": 3.545539454911835e-15, "MET": 3.588870426340002, "Sliced_Wasserstein_Loss": 38.846381191826254, "Total": 2.332753615799187}, "epoch_train_time": 3637.356245994568, "epoch_valid_time": 1033.6469557285309, "epoch_total_time": 4671.0032081604}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/hyperparameters.json

@@ -0,0 +1 @@
+{"num_mlpf_params": 52630547, "train": true, "test": null, "make_plots": null, "comet": true, "save_attention": true, "dataset": "clic", "sort_data": false, "data_dir": "/scratch/persistent/joosep/tensorflow_datasets", "gpus": 1, "gpu_batch_multiplier": 256, "load": null, "finetune": null, "num_epochs": 10, "patience": 20, "lr": 0.0001, "lr_schedule": "cosinedecay", "conv_type": "attention", "ntrain": 1000000, "ntest": 2000, "nvalid": null, "num_workers": 8, "prefetch_factor": 100, "checkpoint_freq": 1, "comet_name": "particleflow-pt", "comet_offline": false, "comet_step_freq": 100, "dtype": "bfloat16", "val_freq": null, "model": {"trainable": "all", "learned_representation_mode": "last", "input_encoding": "split", "pt_mode": "direct-elemtype-split", "eta_mode": "linear", "sin_phi_mode": "linear", "cos_phi_mode": "linear", "energy_mode": "direct-elemtype-split", "gnn_lsh": {"conv_type": "gnn_lsh", "embedding_dim": 1024, "width": 1024, "num_convs": 3, "activation": "elu", "bin_size": 32, "max_num_bins": 200, "distance_dim": 128, "layernorm": true, "num_node_messages": 2, "ffn_dist_hidden_dim": 128, "ffn_dist_num_layers": 2}, "attention": {"conv_type": "attention", "num_convs": 3, "dropout_ff": 0.0, "dropout_conv_id_mha": 0.0, "dropout_conv_id_ff": 0.0, "dropout_conv_reg_mha": 0.0, "dropout_conv_reg_ff": 0.0, "activation": "relu", "head_dim": 32, "num_heads": 32, "attention_type": "flash", "use_pre_layernorm": true}, "mamba": {"conv_type": "mamba", "embedding_dim": 128, "width": 128, "num_convs": 2, "dropout": 0.0, "activation": "elu", "num_heads": 2, "d_state": 16, "d_conv": 4, "expand": 2}}, "lr_schedule_config": {"onecycle": {"pct_start": 0.3}}, "raytune": {"local_dir": null, "sched": null, "search_alg": null, "default_metric": "val_loss", "default_mode": "min", "asha": {"max_t": 200, "reduction_factor": 4, "brackets": 1, "grace_period": 10}, "hyperband": {"max_t": 200, "reduction_factor": 4}, "hyperopt": {"n_random_steps": 10}, "nevergrad": {"n_random_steps": 10}}, "train_dataset": {"clic": {"physical": {"batch_size": 1, "samples": {"clic_edm_qq_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}, "clic_edm_ttbar_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}, "clic_edm_ww_fullhad_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}}}}}, "valid_dataset": {"clic": {"physical": {"batch_size": 1, "samples": {"clic_edm_qq_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}, "clic_edm_ttbar_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}, "clic_edm_ww_fullhad_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}}}}}, "test_dataset": {"clic_edm_qq_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}, "clic_edm_ttbar_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}, "clic_edm_ww_fullhad_pf": {"version": "2.5.0", "splits": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]}}, "enabled_test_datasets": ["clic_edm_qq_pf"]}

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/model_kwargs.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2784c409dcad68c684a808b0e9c311e9eb8ee97cd7c92d4d4f1771878dd87f5
+size 553

clic/clusters/v2.3.0/pyg-clic_20250208_095515_010468/overridden_config.yaml

@@ -0,0 +1,225 @@
+checkpoint_freq: 1
+comet: true
+comet_name: particleflow-pt
+comet_offline: false
+comet_step_freq: 100
+conv_type: attention
+data_dir: /scratch/persistent/joosep/tensorflow_datasets
+dataset: clic
+dtype: bfloat16
+enabled_test_datasets:
+- clic_edm_qq_pf
+finetune: null
+gpu_batch_multiplier: 256
+gpus: 1
+load: null
+lr: 0.0001
+lr_schedule: cosinedecay
+lr_schedule_config:
+  onecycle:
+    pct_start: 0.3
+make_plots: null
+model:
+  attention:
+    activation: relu
+    attention_type: flash
+    conv_type: attention
+    dropout_conv_id_ff: 0.0
+    dropout_conv_id_mha: 0.0
+    dropout_conv_reg_ff: 0.0
+    dropout_conv_reg_mha: 0.0
+    dropout_ff: 0.0
+    head_dim: 32
+    num_convs: 3
+    num_heads: 32
+    use_pre_layernorm: true
+  cos_phi_mode: linear
+  energy_mode: direct-elemtype-split
+  eta_mode: linear
+  gnn_lsh:
+    activation: elu
+    bin_size: 32
+    conv_type: gnn_lsh
+    distance_dim: 128
+    embedding_dim: 1024
+    ffn_dist_hidden_dim: 128
+    ffn_dist_num_layers: 2
+    layernorm: true
+    max_num_bins: 200
+    num_convs: 3
+    num_node_messages: 2
+    width: 1024
+  input_encoding: split
+  learned_representation_mode: last
+  mamba:
+    activation: elu
+    conv_type: mamba
+    d_conv: 4
+    d_state: 16
+    dropout: 0.0
+    embedding_dim: 128
+    expand: 2
+    num_convs: 2
+    num_heads: 2
+    width: 128
+  pt_mode: direct-elemtype-split
+  sin_phi_mode: linear
+  trainable: all
+ntest: 2000
+ntrain: 1000000
+num_epochs: 10
+num_workers: 8
+nvalid: null
+patience: 20
+prefetch_factor: 100
+raytune:
+  asha:
+    brackets: 1
+    grace_period: 10
+    max_t: 200
+    reduction_factor: 4
+  default_metric: val_loss
+  default_mode: min
+  hyperband:
+    max_t: 200
+    reduction_factor: 4
+  hyperopt:
+    n_random_steps: 10
+  local_dir: null
+  nevergrad:
+    n_random_steps: 10
+  sched: null
+  search_alg: null
+save_attention: true
+sort_data: false
+test: null
+test_dataset:
+  clic_edm_qq_pf:
+    splits:
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    version: 2.5.0
+  clic_edm_ttbar_pf:
+    splits:
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    version: 2.5.0
+  clic_edm_ww_fullhad_pf:
+    splits:
+    - 1
+    - 2
+    - 3
+    - 4
+    - 5
+    - 6
+    - 7
+    - 8
+    - 9
+    - 10
+    version: 2.5.0
+train: true
+train_dataset:
+  clic:
+    physical:
+      batch_size: 1
+      samples:
+        clic_edm_qq_pf:
+          splits:
+          - 1
+          - 2
+          - 3
+          - 4
+          - 5
+          - 6
+          - 7
+          - 8
+          - 9
+          - 10
+          version: 2.5.0
+        clic_edm_ttbar_pf:
+          splits:
+          - 1
+          - 2
+          - 3
+          - 4
+          - 5
+          - 6
+          - 7
+          - 8
+          - 9
+          - 10
+          version: 2.5.0
+        clic_edm_ww_fullhad_pf:
+          splits:
+          - 1
+          - 2
+          - 3
+          - 4
+          - 5
+          - 6
+          - 7
+          - 8
+          - 9
+          - 10
+          version: 2.5.0
+val_freq: null
+valid_dataset:
+  clic:
+    physical:
+      batch_size: 1
+      samples:
+        clic_edm_qq_pf:
+          splits:
+          - 1
+          - 2
+          - 3
+          - 4
+          - 5
+          - 6
+          - 7
+          - 8
+          - 9
+          - 10
+          version: 2.5.0
+        clic_edm_ttbar_pf:
+          splits:
+          - 1
+          - 2
+          - 3
+          - 4
+          - 5
+          - 6
+          - 7
+          - 8
+          - 9
+          - 10
+          version: 2.5.0
+        clic_edm_ww_fullhad_pf:
+          splits:
+          - 1
+          - 2
+          - 3
+          - 4
+          - 5
+          - 6
+          - 7
+          - 8
+          - 9
+          - 10
+          version: 2.5.0