diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..549bd6fe217355d0cd931db66c8b3a87e2b69c40 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9391214ff1923b83752a8d3951b5d3401a5ec6d0246781854e0201f97f9f82b3 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..9423ce176e70afb463cc8f01e47419bc22a6c45a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8ff8dd214a728fe9583f1d2f40c92851b977c13963f43257431bd0afcbbad2c +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..945da3a7b272ab3b44a825e59f8b065014495684 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9bfe57805c7c5080f34d0a9b1d11f82f05e58fdce3169d5d4cab80a6b73d0ae +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..3b5d365b3dcb80432a024213246d76dcc9715d01 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:616a2c97b60a139f9cc9ddbc270586849cc31afc3bcce85ae5087b5fe4964eb3 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..ab4817080ea292a2bbc1bd5871848be297464dac --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4882a2e3ee217c356fc14775ece3eb957872f6040ed76f9dd5fdf02601715dd +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..518c6672ba8f0f9a890728f99c74b618d98dfc75 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3cbd61ffa32e5a09cab1b8df46c660a822400be159e93bb759a75e0c8897905 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..5d979968d4edb113f9e45d50041439f46b0cd032 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbf69e6bab55d931c75b4629ca7f94f6d85e67a577f60d56d3ad0356e939e732 +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..52415a4c988570aa14831ea365ffdca3c05fe928 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f01537df585cceb4f79ff0535872d97637653cd1913821410649382c6045014b +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..18268a6f12c0588b2e5b19f23039fb9a7dc457de --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/0_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 6.59199857711792, + "learning_rate": 2e-05, + "loss": 0.5286, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 4.61942720413208, + "learning_rate": 2e-05, + "loss": 0.97, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 4.521786212921143, + "learning_rate": 2e-05, + "loss": 0.3505, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 0.39112570881843567, + "learning_rate": 2e-05, + "loss": 0.115, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 3.506913900375366, + "learning_rate": 2e-05, + "loss": 0.669, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 7.119588375091553, + "learning_rate": 2e-05, + "loss": 0.6728, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 4.159899711608887, + "learning_rate": 2e-05, + "loss": 0.4872, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 5.473998069763184, + "learning_rate": 2e-05, + "loss": 1.3448, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 9.59962272644043, + "learning_rate": 2e-05, + "loss": 1.1658, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 5.037639617919922, + "learning_rate": 2e-05, + "loss": 0.437, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 3.210973024368286, + "learning_rate": 2e-05, + "loss": 0.2022, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 3.4891958236694336, + "learning_rate": 2e-05, + "loss": 0.3925, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 2.502493381500244, + "learning_rate": 2e-05, + "loss": 0.5882, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 0.41307923197746277, + "learning_rate": 2e-05, + "loss": 0.6127, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 2.3616442680358887, + "learning_rate": 2e-05, + "loss": 0.6964, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 5.551742076873779, + "learning_rate": 2e-05, + "loss": 0.2612, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 4.289650917053223, + "learning_rate": 2e-05, + "loss": 0.3091, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 13.974275588989258, + "learning_rate": 2e-05, + "loss": 0.8659, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.4527721405029297, + "learning_rate": 2e-05, + "loss": 0.7772, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 2.5927505493164062, + "learning_rate": 2e-05, + "loss": 0.162, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 0.6561378836631775, + "learning_rate": 2e-05, + "loss": 0.0605, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 4.918743133544922, + "learning_rate": 2e-05, + "loss": 0.7217, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 0.39737483859062195, + "learning_rate": 2e-05, + "loss": 0.1857, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 14.40439224243164, + "learning_rate": 2e-05, + "loss": 1.4849, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 0.6556407809257507, + "learning_rate": 2e-05, + "loss": 0.1272, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 9.867362976074219, + "learning_rate": 2e-05, + "loss": 0.4252, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 2.099479913711548, + "learning_rate": 2e-05, + "loss": 0.0928, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 4.53439474105835, + "learning_rate": 2e-05, + "loss": 0.3139, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 7.238653659820557, + "learning_rate": 2e-05, + "loss": 1.6104, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 1.539478063583374, + "learning_rate": 2e-05, + "loss": 0.1581, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 1.8394383192062378, + "learning_rate": 2e-05, + "loss": 0.4265, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 0.7109262943267822, + "learning_rate": 2e-05, + "loss": 0.1265, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 1.0773893594741821, + "learning_rate": 2e-05, + "loss": 0.2364, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 3.935823917388916, + "learning_rate": 2e-05, + "loss": 1.0373, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 7.714654922485352, + "learning_rate": 2e-05, + "loss": 0.6007, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 8.1398286819458, + "learning_rate": 2e-05, + "loss": 0.8645, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 7.816657066345215, + "learning_rate": 2e-05, + "loss": 2.3697, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 7.676031589508057, + "learning_rate": 2e-05, + "loss": 0.59, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 1.7220346927642822, + "learning_rate": 2e-05, + "loss": 0.9622, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 3.681994915008545, + "learning_rate": 2e-05, + "loss": 0.4974, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 7.639248847961426, + "learning_rate": 2e-05, + "loss": 0.9375, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 9.04971694946289, + "learning_rate": 2e-05, + "loss": 1.22, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 1.4018713235855103, + "learning_rate": 2e-05, + "loss": 0.0818, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 4.88040828704834, + "learning_rate": 2e-05, + "loss": 0.7108, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 2.9837918281555176, + "learning_rate": 2e-05, + "loss": 0.3657, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 1.0587419271469116, + "learning_rate": 2e-05, + "loss": 0.3112, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 9.619867324829102, + "learning_rate": 2e-05, + "loss": 2.6249, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 0.8182443380355835, + "learning_rate": 2e-05, + "loss": 1.052, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 5.4298481941223145, + "learning_rate": 2e-05, + "loss": 0.983, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 1.9551576375961304, + "learning_rate": 2e-05, + "loss": 0.3735, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 2722175404474368.0, + "train_loss": 0.6631992340087891, + "train_runtime": 215.6578, + "train_samples_per_second": 1.855, + "train_steps_per_second": 0.464 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 2722175404474368.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..1300ca21ecc4f0fef6e8098c58c06bebdc7c732a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88cec14609ad32c290c19fe99cc2626c1573cbc85f114434a313145999b97e84 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..d60f72dc78cc164db92e35c2abddc5df9f8c5e38 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df187ae799f263db8566f5385dc79e9c2623b1d6957e1928f94a77e61051ea6c +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..b8ce8b69055dda0df905b69cf6df66a2646d2313 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ae4134c8cf9bd49bd9e9d46343957346c41793b1be9d466e80cfecfcea6f24a +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..216db4e59d1452ecd3379bf1b32e0dc9506dfc37 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5289d21311312be0fde45820e0a508ba5cbfd66bb88d64b7ec96bbf4065de637 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..f12849d349cf2086bfd2755abdeeae5097977042 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c1da2662d2df70b5cf7a9cbac9949d8be681b8e616a3e9f8bd8f573d3ae40e3 +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..1966bdbb78b85b8c1a3561b54769336159522124 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1cf4c5a1502c71922c0c95fafe0bced2fd2386a6dcb782663f81d80fe38eae27 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..a0e50ddbc497ca251d346ec9455dfbd1651691a1 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11cabf8f804c392f7578109c0fc6aeb86b7f26cd200fd331d5756dd441870339 +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..bcef4d548da713e63413482186bf3bb3eed84ef0 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:058f762da3f1f0d383fcdfdbaa9c44ff0a75249cec5d1d061055f5c8bc46789d +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..4ed1201e819790b2e2c4bdfed4336959fd4e891e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/1_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 0.849377453327179, + "learning_rate": 2e-05, + "loss": 0.0279, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 0.02175198495388031, + "learning_rate": 2e-05, + "loss": 0.0578, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 0.5898083448410034, + "learning_rate": 2e-05, + "loss": 0.0414, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 0.41922903060913086, + "learning_rate": 2e-05, + "loss": 0.0256, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 0.0018654951127246022, + "learning_rate": 2e-05, + "loss": 0.0004, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 0.004117058124393225, + "learning_rate": 2e-05, + "loss": 0.1077, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 0.047437746077775955, + "learning_rate": 2e-05, + "loss": 0.0009, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 1.518896460533142, + "learning_rate": 2e-05, + "loss": 0.349, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 0.045308228582143784, + "learning_rate": 2e-05, + "loss": 0.0889, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 0.11829675734043121, + "learning_rate": 2e-05, + "loss": 0.0026, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 0.005238677840679884, + "learning_rate": 2e-05, + "loss": 0.4929, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 0.2110355943441391, + "learning_rate": 2e-05, + "loss": 0.008, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 1.4381576776504517, + "learning_rate": 2e-05, + "loss": 0.0896, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 16.31382942199707, + "learning_rate": 2e-05, + "loss": 0.2398, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 0.05612191930413246, + "learning_rate": 2e-05, + "loss": 0.0021, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 0.22438876330852509, + "learning_rate": 2e-05, + "loss": 0.0076, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.866287350654602, + "learning_rate": 2e-05, + "loss": 0.0595, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 0.045663584023714066, + "learning_rate": 2e-05, + "loss": 0.0009, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 0.2734965682029724, + "learning_rate": 2e-05, + "loss": 0.0132, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 0.5363508462905884, + "learning_rate": 2e-05, + "loss": 0.0329, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 0.382310688495636, + "learning_rate": 2e-05, + "loss": 0.0192, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 8.9566011428833, + "learning_rate": 2e-05, + "loss": 0.3838, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 1.7171825170516968, + "learning_rate": 2e-05, + "loss": 0.0176, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 1.6939023733139038, + "learning_rate": 2e-05, + "loss": 0.0231, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 0.0025089113041758537, + "learning_rate": 2e-05, + "loss": 0.0026, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 0.06536436825990677, + "learning_rate": 2e-05, + "loss": 0.2801, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 0.20245826244354248, + "learning_rate": 2e-05, + "loss": 0.0073, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 0.018265284597873688, + "learning_rate": 2e-05, + "loss": 0.0028, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 0.13041305541992188, + "learning_rate": 2e-05, + "loss": 0.1354, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 0.010534190572798252, + "learning_rate": 2e-05, + "loss": 1.0962, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 0.2692118287086487, + "learning_rate": 2e-05, + "loss": 0.0085, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 0.038933683186769485, + "learning_rate": 2e-05, + "loss": 0.0019, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 0.018930355086922646, + "learning_rate": 2e-05, + "loss": 0.0017, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 0.7976124286651611, + "learning_rate": 2e-05, + "loss": 0.0334, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 0.2943456470966339, + "learning_rate": 2e-05, + "loss": 0.0089, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 0.12224985659122467, + "learning_rate": 2e-05, + "loss": 0.0067, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 0.3579569458961487, + "learning_rate": 2e-05, + "loss": 0.0677, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 0.02683216519653797, + "learning_rate": 2e-05, + "loss": 0.0016, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 3.3340158462524414, + "learning_rate": 2e-05, + "loss": 0.2447, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 0.009762264788150787, + "learning_rate": 2e-05, + "loss": 0.0083, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 0.024613451212644577, + "learning_rate": 2e-05, + "loss": 0.0116, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 0.009135313332080841, + "learning_rate": 2e-05, + "loss": 0.0008, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 0.2523867189884186, + "learning_rate": 2e-05, + "loss": 0.4089, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 0.32133349776268005, + "learning_rate": 2e-05, + "loss": 0.0093, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 0.09947264194488525, + "learning_rate": 2e-05, + "loss": 0.0039, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 0.12009107321500778, + "learning_rate": 2e-05, + "loss": 0.0069, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 0.007163307163864374, + "learning_rate": 2e-05, + "loss": 0.0081, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 0.4114590883255005, + "learning_rate": 2e-05, + "loss": 0.066, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 0.04830743372440338, + "learning_rate": 2e-05, + "loss": 0.0046, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 0.04225878044962883, + "learning_rate": 2e-05, + "loss": 0.0019, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 2743212791824384.0, + "train_loss": 0.0904453244805336, + "train_runtime": 221.9331, + "train_samples_per_second": 1.802, + "train_steps_per_second": 0.451 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 2743212791824384.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..007a57d774c9c419fceb8b381a15666bce4df737 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9be8024aa2eece99882c85b8046475baa83ad6283113525243ad4265e9bf7576 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..7af33ff80f19fb450dab669692d702862b9ec8c8 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce08ca256ac2358ea58f9bf1b6e1e689255df91a4f52dbeb33536a4dcc474533 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..c60537dcb51ee63d74560d1d14ebbc07b61ff9a4 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c586d85bfc4ed525de5cc0bcd8e9bb1def55fc41b70014035e84fbbdfc08625b +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..57d2a2f558be24f3ae7e8f176b297dd523c598cc --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d608fe1ce0b07b208d805224fc9aa2685c3edeca580c162648006f6425393528 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..f61bea9c5c1e98e1a9e28daf5f2947ed2cfb88b3 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2072f36c89ed2416f35894c760f95c059d0947e40d2c851e6559e50ac1c610dd +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..ee24bfa47a8f77b8e98af48557aeae48f3144878 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99426839dd4d5f67e775bb10a4c52b99e22e14e2de2c8b8d162e9059a49513cc +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..f1ce8831f75ce4390020bea1c1d37c11c4462b19 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8586e15be816067f89397fd74efde646b7f477c27e24148b917e3bfa10b0d3e +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..77ddb5247427c50ac3a1c45560b7637e585cc3e3 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6dd9c04cdc272cfff8bddd92cc280c2385a2797d1a2f3a196861109207940d86 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..dfb22a5cbf6d54d86fd654d6903e9e3252e296ac --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/2_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 1.3100306987762451, + "learning_rate": 2e-05, + "loss": 0.3864, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 4.013766288757324, + "learning_rate": 2e-05, + "loss": 0.9063, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 2.3978607654571533, + "learning_rate": 2e-05, + "loss": 1.1349, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 4.109842300415039, + "learning_rate": 2e-05, + "loss": 0.825, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 2.427557945251465, + "learning_rate": 2e-05, + "loss": 0.5722, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 4.734681606292725, + "learning_rate": 2e-05, + "loss": 1.0887, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 0.10447093099355698, + "learning_rate": 2e-05, + "loss": 0.5304, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 2.675856828689575, + "learning_rate": 2e-05, + "loss": 0.9177, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 1.6946303844451904, + "learning_rate": 2e-05, + "loss": 0.4756, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 1.9363477230072021, + "learning_rate": 2e-05, + "loss": 0.7104, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 0.7683027982711792, + "learning_rate": 2e-05, + "loss": 0.1338, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 5.035861492156982, + "learning_rate": 2e-05, + "loss": 0.9689, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 3.8543343544006348, + "learning_rate": 2e-05, + "loss": 0.523, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 3.667877435684204, + "learning_rate": 2e-05, + "loss": 0.3602, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 3.0984058380126953, + "learning_rate": 2e-05, + "loss": 0.5042, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 1.3304017782211304, + "learning_rate": 2e-05, + "loss": 0.2943, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 2.7662222385406494, + "learning_rate": 2e-05, + "loss": 0.5895, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 0.6082299947738647, + "learning_rate": 2e-05, + "loss": 0.1146, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.9511778354644775, + "learning_rate": 2e-05, + "loss": 0.7031, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 3.04605770111084, + "learning_rate": 2e-05, + "loss": 0.3958, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 3.8611412048339844, + "learning_rate": 2e-05, + "loss": 0.8662, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 4.399799823760986, + "learning_rate": 2e-05, + "loss": 0.287, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 2.2531864643096924, + "learning_rate": 2e-05, + "loss": 0.3218, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 1.4663535356521606, + "learning_rate": 2e-05, + "loss": 1.0107, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 1.7521519660949707, + "learning_rate": 2e-05, + "loss": 0.124, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 8.896435737609863, + "learning_rate": 2e-05, + "loss": 0.8135, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 2.5376131534576416, + "learning_rate": 2e-05, + "loss": 0.511, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 6.814590930938721, + "learning_rate": 2e-05, + "loss": 1.0585, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 2.1924495697021484, + "learning_rate": 2e-05, + "loss": 0.1282, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 0.17645888030529022, + "learning_rate": 2e-05, + "loss": 0.0467, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 2.022798538208008, + "learning_rate": 2e-05, + "loss": 0.7755, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 3.513547897338867, + "learning_rate": 2e-05, + "loss": 1.0105, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 2.7813096046447754, + "learning_rate": 2e-05, + "loss": 0.3313, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 14.315655708312988, + "learning_rate": 2e-05, + "loss": 1.5528, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 3.795628309249878, + "learning_rate": 2e-05, + "loss": 0.5101, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 1.2288278341293335, + "learning_rate": 2e-05, + "loss": 1.374, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 4.386161804199219, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 3.3581271171569824, + "learning_rate": 2e-05, + "loss": 0.6263, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 1.225866436958313, + "learning_rate": 2e-05, + "loss": 0.1825, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 7.104559898376465, + "learning_rate": 2e-05, + "loss": 1.1801, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 13.105498313903809, + "learning_rate": 2e-05, + "loss": 1.7541, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 0.2902798056602478, + "learning_rate": 2e-05, + "loss": 0.0271, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 4.769597053527832, + "learning_rate": 2e-05, + "loss": 0.5574, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 3.8983893394470215, + "learning_rate": 2e-05, + "loss": 1.3119, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 3.6488382816314697, + "learning_rate": 2e-05, + "loss": 1.0188, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 0.29925721883773804, + "learning_rate": 2e-05, + "loss": 0.5366, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 3.2376351356506348, + "learning_rate": 2e-05, + "loss": 0.6949, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 0.63726806640625, + "learning_rate": 2e-05, + "loss": 0.2721, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 5.418401718139648, + "learning_rate": 2e-05, + "loss": 1.4491, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 2.0275309085845947, + "learning_rate": 2e-05, + "loss": 0.5931, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5059805727162368.0, + "train_loss": 0.6651638150215149, + "train_runtime": 277.116, + "train_samples_per_second": 1.443, + "train_steps_per_second": 0.361 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5059805727162368.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..380cad7670afc3c16d264901857180390f1a1ed6 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57fec58061a92ab1b1bf4a4c6012fc1397a46c90b77813f7cdd58f7cb3fe2ed3 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..014e794ef32e4414a151fac95568b206adf6d6d2 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3401ff987c93089fb123edc132faa3c3adae7422977093002adb8f8d9588d91 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..46839f81ebe0ec85919fefcb6b141623e20e47a4 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ac62e5e6fe65b335575f7ed7e9b1e3e428debec1fca2e10d9dc2123665b121b +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..65800587c952f70b7fdfa27d611642c27e877cd4 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6901f0e6d410fbd830c0624606aa6c27c103266b25c9e6b95ab4a281cc6bfc6 +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..ba0ca0f73dc57f39fcf3bf929565e47eb1f7bd21 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eaaf2162f9a52ec84d0a5306af230ee2e2524d8b3ab4f9007c0eb79f88a34f95 +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..969778914dd3670b7232dc020b43db7a862527a1 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa5e605e3d74782eebb64e65d6103a2c51abb8a7defe4b8c0cf0e4b3c764165a +size 302202238 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..22da55602b05f61ba6dff5c2ec114c8a85eb7abd --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2765067dfd32667733c18170c3f622d661e26844689a324651bb49e31c0531eb +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..17ac513d22a206ee70d8b9e6ab616238df8b7d18 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a34222b6b3c027b0a0631a8b22b43c25f1371bfd82e5daf1543cf536e3c8723 +size 302201386 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..20470f8c4760151d3ed3891d8998a3ef03cb8b70 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/3_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 4.537669658660889, + "learning_rate": 2e-05, + "loss": 1.832, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 1.7357145547866821, + "learning_rate": 2e-05, + "loss": 0.6956, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 4.840099334716797, + "learning_rate": 2e-05, + "loss": 1.1172, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 6.74650239944458, + "learning_rate": 2e-05, + "loss": 1.2902, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 5.235438346862793, + "learning_rate": 2e-05, + "loss": 0.8775, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 7.184224605560303, + "learning_rate": 2e-05, + "loss": 2.0581, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 12.11132526397705, + "learning_rate": 2e-05, + "loss": 2.1586, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 5.101541042327881, + "learning_rate": 2e-05, + "loss": 1.0109, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 4.515460014343262, + "learning_rate": 2e-05, + "loss": 0.7775, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 1.4222784042358398, + "learning_rate": 2e-05, + "loss": 0.935, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 4.021170139312744, + "learning_rate": 2e-05, + "loss": 0.8757, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 2.8799550533294678, + "learning_rate": 2e-05, + "loss": 0.809, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 0.9323534369468689, + "learning_rate": 2e-05, + "loss": 0.4597, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 2.9174280166625977, + "learning_rate": 2e-05, + "loss": 0.3155, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 3.026035785675049, + "learning_rate": 2e-05, + "loss": 0.6032, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 8.995691299438477, + "learning_rate": 2e-05, + "loss": 0.9903, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.847169041633606, + "learning_rate": 2e-05, + "loss": 1.3635, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 9.795331001281738, + "learning_rate": 2e-05, + "loss": 2.0643, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 1.7796157598495483, + "learning_rate": 2e-05, + "loss": 0.4105, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 4.427263259887695, + "learning_rate": 2e-05, + "loss": 0.8952, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 15.724753379821777, + "learning_rate": 2e-05, + "loss": 2.4695, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 6.148226261138916, + "learning_rate": 2e-05, + "loss": 1.2556, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 1.515079140663147, + "learning_rate": 2e-05, + "loss": 0.4941, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 1.410451889038086, + "learning_rate": 2e-05, + "loss": 0.2573, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 4.407503604888916, + "learning_rate": 2e-05, + "loss": 0.8896, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 8.208944320678711, + "learning_rate": 2e-05, + "loss": 2.3883, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 1.5512627363204956, + "learning_rate": 2e-05, + "loss": 0.6002, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 5.655065059661865, + "learning_rate": 2e-05, + "loss": 0.9759, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 2.2757630348205566, + "learning_rate": 2e-05, + "loss": 0.481, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 9.758919715881348, + "learning_rate": 2e-05, + "loss": 1.1768, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 4.106739521026611, + "learning_rate": 2e-05, + "loss": 0.7397, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 4.5539231300354, + "learning_rate": 2e-05, + "loss": 0.7899, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 3.6534392833709717, + "learning_rate": 2e-05, + "loss": 0.6942, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 4.095523357391357, + "learning_rate": 2e-05, + "loss": 1.1672, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 6.486148834228516, + "learning_rate": 2e-05, + "loss": 1.2393, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 8.842077255249023, + "learning_rate": 2e-05, + "loss": 1.7434, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 8.89773178100586, + "learning_rate": 2e-05, + "loss": 1.2772, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 18.22632598876953, + "learning_rate": 2e-05, + "loss": 3.3775, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 2.2898244857788086, + "learning_rate": 2e-05, + "loss": 0.7762, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 7.303566932678223, + "learning_rate": 2e-05, + "loss": 0.6705, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 4.615819454193115, + "learning_rate": 2e-05, + "loss": 0.4729, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 3.424020290374756, + "learning_rate": 2e-05, + "loss": 0.6039, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 2.209282636642456, + "learning_rate": 2e-05, + "loss": 2.1192, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 1.9839006662368774, + "learning_rate": 2e-05, + "loss": 0.9141, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 1.8658138513565063, + "learning_rate": 2e-05, + "loss": 0.9538, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 5.851518630981445, + "learning_rate": 2e-05, + "loss": 0.7171, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 3.3957180976867676, + "learning_rate": 2e-05, + "loss": 0.8128, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 1.1212002038955688, + "learning_rate": 2e-05, + "loss": 0.3877, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 3.886902093887329, + "learning_rate": 2e-05, + "loss": 0.8257, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 3.7381937503814697, + "learning_rate": 2e-05, + "loss": 0.7051, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 2779357248487424.0, + "train_loss": 1.0703019142150878, + "train_runtime": 215.987, + "train_samples_per_second": 1.852, + "train_steps_per_second": 0.463 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 2779357248487424.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..69b7e42742ce3892860d2634d8de15eacb1ac1b6 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69d4760f0c1959606186ed7e2dbc53f991bd01e88e3a73b20618c9bcbd214386 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..ea87e2b0616230d476c4d07bc9a02819c04d8cfd --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4447cf5a3e3ba4fc20c85906454383543302b54a45eb6230e015b355a7c620c +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..88f2ec9fe16325633e36f949f60510d2dff72709 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c4e788867c89e82a865f5c263bfc27372f9ad1c131ee5f73ba1257cb1358df0 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..0d3d3854114a9249e1eb79aafa60d1b8255e56d6 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4132f75da66ae6d2d1d4e9d126dcf9907beda51bc8637a5d299f80f47e1dc1b +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..352c913ea5c206179e5ed19ea15aa7a4c1e91bd5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83ae2b5822fb41da2f51fa255253ef5fe5e0742b6941f2e9d52243a28d00e24f +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..1662cc24ff85ea68f7e0fda0f9ef38e54d3fdf3e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea8bb07c77a2cfd707f77600756cadf906298b5c948248212f3d5d1787fded51 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..0408ebc1735b7be0580ab0c1b434b64c4ce3fc5f --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4d08d3691c5ba154f8ed2e0f222d7e52e43fe806519ae3751d1bec29a8d4105 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..ded68ee810b8a13bbf5e17b8f14ed37039fe2016 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e3655617efa0261b43cf077e8d338e5d7409043547ee972bf9bb851282182c3 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..26be992884fc8d5361058ca40860f08a0cdf0a7e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/4_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 3.9793167114257812, + "learning_rate": 2e-05, + "loss": 0.9603, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 2.5188374519348145, + "learning_rate": 2e-05, + "loss": 0.8116, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 1.3143935203552246, + "learning_rate": 2e-05, + "loss": 0.6426, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 1.7295637130737305, + "learning_rate": 2e-05, + "loss": 0.7914, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 0.7666940689086914, + "learning_rate": 2e-05, + "loss": 0.5176, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 1.7987310886383057, + "learning_rate": 2e-05, + "loss": 0.2534, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 2.515049695968628, + "learning_rate": 2e-05, + "loss": 0.5228, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 4.360015869140625, + "learning_rate": 2e-05, + "loss": 1.2806, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 1.8536200523376465, + "learning_rate": 2e-05, + "loss": 0.4718, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 2.595823287963867, + "learning_rate": 2e-05, + "loss": 0.5098, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 3.5519330501556396, + "learning_rate": 2e-05, + "loss": 1.5173, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 1.3497133255004883, + "learning_rate": 2e-05, + "loss": 0.9349, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 1.930624008178711, + "learning_rate": 2e-05, + "loss": 0.4916, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 3.4734158515930176, + "learning_rate": 2e-05, + "loss": 1.0893, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 2.4782814979553223, + "learning_rate": 2e-05, + "loss": 0.8819, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 3.521637439727783, + "learning_rate": 2e-05, + "loss": 0.9765, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.3706670999526978, + "learning_rate": 2e-05, + "loss": 0.4721, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 5.157536506652832, + "learning_rate": 2e-05, + "loss": 0.672, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 1.5482473373413086, + "learning_rate": 2e-05, + "loss": 0.3766, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 5.12284517288208, + "learning_rate": 2e-05, + "loss": 1.2043, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 2.4150655269622803, + "learning_rate": 2e-05, + "loss": 0.9744, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 2.1841928958892822, + "learning_rate": 2e-05, + "loss": 0.5246, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 5.618422985076904, + "learning_rate": 2e-05, + "loss": 1.2388, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 2.524454116821289, + "learning_rate": 2e-05, + "loss": 0.5944, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 1.7722740173339844, + "learning_rate": 2e-05, + "loss": 0.4012, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 4.978190898895264, + "learning_rate": 2e-05, + "loss": 1.5566, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 3.1553709506988525, + "learning_rate": 2e-05, + "loss": 0.945, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 0.8509775400161743, + "learning_rate": 2e-05, + "loss": 0.1817, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 3.637312173843384, + "learning_rate": 2e-05, + "loss": 0.7817, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 5.024761199951172, + "learning_rate": 2e-05, + "loss": 1.0443, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 1.6964753866195679, + "learning_rate": 2e-05, + "loss": 1.0813, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 2.6202216148376465, + "learning_rate": 2e-05, + "loss": 0.8394, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 2.0826215744018555, + "learning_rate": 2e-05, + "loss": 0.6085, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 1.6744186878204346, + "learning_rate": 2e-05, + "loss": 0.5914, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 3.6541945934295654, + "learning_rate": 2e-05, + "loss": 0.8268, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 2.257369041442871, + "learning_rate": 2e-05, + "loss": 0.5257, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 0.468746542930603, + "learning_rate": 2e-05, + "loss": 0.5201, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 2.3300929069519043, + "learning_rate": 2e-05, + "loss": 1.035, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 3.127511739730835, + "learning_rate": 2e-05, + "loss": 0.8523, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 2.3533546924591064, + "learning_rate": 2e-05, + "loss": 0.3409, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 2.2223410606384277, + "learning_rate": 2e-05, + "loss": 0.5418, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 2.9352030754089355, + "learning_rate": 2e-05, + "loss": 0.7653, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 0.8046193718910217, + "learning_rate": 2e-05, + "loss": 0.3562, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 1.5213556289672852, + "learning_rate": 2e-05, + "loss": 0.484, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 2.6822187900543213, + "learning_rate": 2e-05, + "loss": 0.8862, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 5.5709662437438965, + "learning_rate": 2e-05, + "loss": 0.9033, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 4.302905082702637, + "learning_rate": 2e-05, + "loss": 1.0037, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 2.346940279006958, + "learning_rate": 2e-05, + "loss": 0.4615, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 1.4154908657073975, + "learning_rate": 2e-05, + "loss": 0.2501, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 7.068972587585449, + "learning_rate": 2e-05, + "loss": 1.4809, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5838906972962816.0, + "train_loss": 0.7595100402832031, + "train_runtime": 275.7767, + "train_samples_per_second": 1.45, + "train_steps_per_second": 0.363 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5838906972962816.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..4b47512115460c34b88f3a60d9a5ad637a6cb41c --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7061a373da978d284035594b21fa90359b7f63246265529bf3bbd4424d47ab77 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..d2007c7b2b0a2a6b1bec9ef2a4813b407d860bb6 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea9f5a3d5256c4d387533a530ca53a00edf9a3dd5b539d359ef244972f191740 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..d29ca5daa7db1ca0b322393117de60872d948585 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0d2247c87168cd25007195226e5a2d0cc8da9ff1ad46d1e494b8a8a64bc353b +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..421b1b7bf72c6c2701411f2d173e45c14cf1643e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd9f2bd53bd05c8e603b94af226a34333ecaf4dc93ab7e920e7ef6ffed5709e6 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..02b627056d79ef4302ccf0dee39ab0b632d7a17a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa4f85fb161c9eb53205e83989ae34fdafca370bf153f995e1e6f4ac63320fdf +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..ee511d8eaa5d6f6977c747738084b4c4b67a9ab7 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2fdf1e51b4032f528b7d92e3a441d9d9c9d793d3e6a1bcc81416b61a4c773d39 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..815da86fac51b82adfb63f1ad67796272f38dc8e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a77b197fa97b232efc0a8a4946b08f2af65e75753f9eb1e4b34d0308c4342a4 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..7ee8f8f7000aa132a41d931a4e70dca6903c3b0e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b47bbdd284a0516732c42f4da39c714af2b1a8e7e3cec075a59de2d8f1b078a7 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..d1be0d119946fabbc438766cb8e1c6a3ca837cee --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/5_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 0.9817007780075073, + "learning_rate": 2e-05, + "loss": 0.1973, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 5.000643730163574, + "learning_rate": 2e-05, + "loss": 0.4885, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 1.4880672693252563, + "learning_rate": 2e-05, + "loss": 0.1501, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 0.1219613328576088, + "learning_rate": 2e-05, + "loss": 0.0744, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 0.30120301246643066, + "learning_rate": 2e-05, + "loss": 0.1931, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 0.4320571720600128, + "learning_rate": 2e-05, + "loss": 0.0543, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 0.07388892769813538, + "learning_rate": 2e-05, + "loss": 0.2326, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 0.17824463546276093, + "learning_rate": 2e-05, + "loss": 0.0421, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 0.045538149774074554, + "learning_rate": 2e-05, + "loss": 0.1021, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 1.0216370820999146, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 5.225408554077148, + "learning_rate": 2e-05, + "loss": 0.4183, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 1.8671045303344727, + "learning_rate": 2e-05, + "loss": 0.1205, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 0.47157877683639526, + "learning_rate": 2e-05, + "loss": 0.4838, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 7.5469441413879395, + "learning_rate": 2e-05, + "loss": 1.7558, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 0.03332682326436043, + "learning_rate": 2e-05, + "loss": 0.1481, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 4.355489253997803, + "learning_rate": 2e-05, + "loss": 0.1577, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.1134517192840576, + "learning_rate": 2e-05, + "loss": 0.0298, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 0.9689416289329529, + "learning_rate": 2e-05, + "loss": 0.0965, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 0.46283620595932007, + "learning_rate": 2e-05, + "loss": 0.0317, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 1.4462867975234985, + "learning_rate": 2e-05, + "loss": 0.1154, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 0.12069402635097504, + "learning_rate": 2e-05, + "loss": 0.0406, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 7.088903427124023, + "learning_rate": 2e-05, + "loss": 0.8975, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 0.9605217576026917, + "learning_rate": 2e-05, + "loss": 0.9944, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 3.018798828125, + "learning_rate": 2e-05, + "loss": 1.2334, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 5.704972743988037, + "learning_rate": 2e-05, + "loss": 0.3938, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 6.408511638641357, + "learning_rate": 2e-05, + "loss": 0.3957, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 0.08005835115909576, + "learning_rate": 2e-05, + "loss": 0.1802, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 7.279331684112549, + "learning_rate": 2e-05, + "loss": 0.3024, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 0.16459767520427704, + "learning_rate": 2e-05, + "loss": 0.0241, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 0.4274437427520752, + "learning_rate": 2e-05, + "loss": 0.0496, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 0.7548133134841919, + "learning_rate": 2e-05, + "loss": 0.0634, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 6.092557430267334, + "learning_rate": 2e-05, + "loss": 1.0826, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 1.127923607826233, + "learning_rate": 2e-05, + "loss": 0.1145, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 0.29472213983535767, + "learning_rate": 2e-05, + "loss": 0.0665, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 3.9681475162506104, + "learning_rate": 2e-05, + "loss": 0.7564, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 2.1758413314819336, + "learning_rate": 2e-05, + "loss": 0.1239, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 0.18620817363262177, + "learning_rate": 2e-05, + "loss": 0.5831, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 5.317977428436279, + "learning_rate": 2e-05, + "loss": 0.269, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 0.3147411346435547, + "learning_rate": 2e-05, + "loss": 1.0955, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 3.4752018451690674, + "learning_rate": 2e-05, + "loss": 0.3737, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 0.4545653164386749, + "learning_rate": 2e-05, + "loss": 0.0599, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 0.0749347060918808, + "learning_rate": 2e-05, + "loss": 0.7111, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 1.0164120197296143, + "learning_rate": 2e-05, + "loss": 0.0859, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 7.552253246307373, + "learning_rate": 2e-05, + "loss": 1.0359, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 0.4121910631656647, + "learning_rate": 2e-05, + "loss": 0.0295, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 7.473265171051025, + "learning_rate": 2e-05, + "loss": 0.6172, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 0.7466989159584045, + "learning_rate": 2e-05, + "loss": 0.0707, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 4.750700950622559, + "learning_rate": 2e-05, + "loss": 0.3368, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 0.14242853224277496, + "learning_rate": 2e-05, + "loss": 0.2927, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 5.459415435791016, + "learning_rate": 2e-05, + "loss": 1.9345, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5122038796648448.0, + "train_loss": 0.38509593725204466, + "train_runtime": 275.5399, + "train_samples_per_second": 1.452, + "train_steps_per_second": 0.363 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5122038796648448.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..16bc73285529af2bacbd97ed34dec90084079adc --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d869f75f7e4ea4cfba5002a15045d4fea412cbd1f18c87b01b532c6793bf6957 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..a44cba35f00d718d37ca216b48d28b6a4730e7d0 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d54838afd17936b9784bef34dca7d4fee653d4b69dadb1b7aec0019c3cb8003 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..d6f1875cd7af7b73a9e0b435e8a4e5d35b0fe089 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26ba2c9ace9e0661c5105b0efa1864ad6698b7df18de13c17efad5361cf10e12 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..4fa52600439b3f72794568cd4d3bb7a5143fc6ff --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:37b7226cf960533fad34c5f1a1c5748b11a499000c636d5fedda94ff065e104e +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..e47c0a6a122e7a3bcf3527c2e0cc0289533dc772 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1223c4abf0e0ae4ac4fa0f46252e641a5aef91bf48d8b37f3f501168ca876c1 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..3eea2d31321ed6e09e4d5b615e9adeb590593a02 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5cc4d2860e45f1c4a542b103527b5fa1a46808e9d9c274ceb82604c8ca8e816b +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..45ed7a1e269d0aa36905f4183b2c1fb9ef35ce1d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff3c605d6693a267ee5bd7ecb1da76e4eaba12410ba036124d91695804a3aa2b +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..923102c2da9048dbd0af50d03715a88ceca9459a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:380b1c6de9764bad66bbaa165f923dd1a93ce9acc6da07c438cd85cdf888c3c9 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..98b040f291ce131453744d56136807453e26f9d3 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/6_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 4.208325386047363, + "learning_rate": 2e-05, + "loss": 0.8797, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 5.241598129272461, + "learning_rate": 2e-05, + "loss": 0.8743, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 5.811727046966553, + "learning_rate": 2e-05, + "loss": 1.2891, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 1.7633750438690186, + "learning_rate": 2e-05, + "loss": 1.0135, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 7.0835371017456055, + "learning_rate": 2e-05, + "loss": 1.6237, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 2.144989252090454, + "learning_rate": 2e-05, + "loss": 0.3553, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 7.332456111907959, + "learning_rate": 2e-05, + "loss": 0.862, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 7.372664451599121, + "learning_rate": 2e-05, + "loss": 1.1755, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 5.060377597808838, + "learning_rate": 2e-05, + "loss": 1.7372, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 2.078317880630493, + "learning_rate": 2e-05, + "loss": 0.5718, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 4.678717136383057, + "learning_rate": 2e-05, + "loss": 0.7658, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 4.618237018585205, + "learning_rate": 2e-05, + "loss": 0.5913, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 3.5979042053222656, + "learning_rate": 2e-05, + "loss": 0.6327, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 4.881328105926514, + "learning_rate": 2e-05, + "loss": 0.7574, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 6.3498945236206055, + "learning_rate": 2e-05, + "loss": 1.3032, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 3.847893476486206, + "learning_rate": 2e-05, + "loss": 0.6458, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 4.588125228881836, + "learning_rate": 2e-05, + "loss": 0.8257, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 4.323940277099609, + "learning_rate": 2e-05, + "loss": 1.1001, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.361527442932129, + "learning_rate": 2e-05, + "loss": 0.4357, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 2.4622833728790283, + "learning_rate": 2e-05, + "loss": 0.7082, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 3.3581650257110596, + "learning_rate": 2e-05, + "loss": 1.1638, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 1.7480125427246094, + "learning_rate": 2e-05, + "loss": 0.348, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 5.21283483505249, + "learning_rate": 2e-05, + "loss": 1.0939, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 2.6577634811401367, + "learning_rate": 2e-05, + "loss": 0.9062, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 3.4171979427337646, + "learning_rate": 2e-05, + "loss": 0.651, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 3.9935219287872314, + "learning_rate": 2e-05, + "loss": 0.7884, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 5.5232834815979, + "learning_rate": 2e-05, + "loss": 0.677, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 426.50921630859375, + "learning_rate": 2e-05, + "loss": 1.5006, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 3.0177435874938965, + "learning_rate": 2e-05, + "loss": 0.939, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 1.6758413314819336, + "learning_rate": 2e-05, + "loss": 0.1663, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 6.9303741455078125, + "learning_rate": 2e-05, + "loss": 0.553, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 8.198774337768555, + "learning_rate": 2e-05, + "loss": 0.8015, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 7.569382190704346, + "learning_rate": 2e-05, + "loss": 1.1962, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 8.9950590133667, + "learning_rate": 2e-05, + "loss": 1.2721, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 6.853026390075684, + "learning_rate": 2e-05, + "loss": 0.7473, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 2.8391356468200684, + "learning_rate": 2e-05, + "loss": 0.5548, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 5.503705978393555, + "learning_rate": 2e-05, + "loss": 1.0647, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 8.949329376220703, + "learning_rate": 2e-05, + "loss": 1.5009, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 3.5043232440948486, + "learning_rate": 2e-05, + "loss": 0.297, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 1.6734578609466553, + "learning_rate": 2e-05, + "loss": 0.1819, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 2.0592703819274902, + "learning_rate": 2e-05, + "loss": 0.6246, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 1.7888612747192383, + "learning_rate": 2e-05, + "loss": 1.1001, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 2.127927541732788, + "learning_rate": 2e-05, + "loss": 0.6432, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 3.362846612930298, + "learning_rate": 2e-05, + "loss": 0.655, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 13.346572875976562, + "learning_rate": 2e-05, + "loss": 1.0053, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 3.1919820308685303, + "learning_rate": 2e-05, + "loss": 1.7779, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 6.950253963470459, + "learning_rate": 2e-05, + "loss": 0.7888, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 3.242849349975586, + "learning_rate": 2e-05, + "loss": 0.3484, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 7.720677852630615, + "learning_rate": 2e-05, + "loss": 0.7812, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 13.759918212890625, + "learning_rate": 2e-05, + "loss": 1.8311, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 9986200676859904.0, + "train_loss": 0.8821463584899902, + "train_runtime": 280.8196, + "train_samples_per_second": 1.424, + "train_steps_per_second": 0.356 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 9986200676859904.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..a0ab26f35693e317579657a7b5a56b660d6e47e5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2eab3f23b1fe9ec14a9ec6c0650c25c20af1a8eb8dc6794cf3386d7322b1500d +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..f0cee4d8c1cf073ee5323d0665d9e09a1c110497 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db06b9509a446f6b5e3bc17f982a48838d935fe5628055a6d9c22433ca9143bd +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..1afa4eeba8abf35b94876afb3f665d672cf459d1 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6df0448032ae8d5149aa980e8057f238c22186ea72c43c88e26591bfc43f28e5 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..fbbca50bc3a2db4e6cfe5813ad3997385c237db8 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c25247fc4b7dcbb9ce23d8c08dda5b3e8f9ab124c49d1ea011434579a888d449 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..cbeecb67aa37e4a77736421c71e0af92d81e72ea --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b56d70943f61917c04e582d065c822d06c8c6b7fda99824d7d05e2c220cd465a +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..518d7449bedd6a2b805e380e3b08bed403638548 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc802842ae12d0d9c18517cf45dece590ebe50170bb96852aa4b0c24264e6712 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..4459e6cc4020401041517cb41981ddf2ef91fc80 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d023eb46c278fbd0b8489015075246ff52656a53ab5076e171c9daf4725aaa10 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..757cf14ee7c3fcb17b84503c41e39716471bb362 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:619de35084cbc13cd906a3b9534585a7cc4f356d446d3760cbcb1e43682a666f +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..9a69187d210d15ea5ce21e3ce6327228d4f08679 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/7_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 0.33381178975105286, + "learning_rate": 2e-05, + "loss": 0.0758, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 6.295736312866211, + "learning_rate": 2e-05, + "loss": 1.231, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 4.599400997161865, + "learning_rate": 2e-05, + "loss": 0.6982, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 0.38898980617523193, + "learning_rate": 2e-05, + "loss": 0.1531, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 5.583261013031006, + "learning_rate": 2e-05, + "loss": 1.2493, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 0.5936524868011475, + "learning_rate": 2e-05, + "loss": 0.133, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 2.9277100563049316, + "learning_rate": 2e-05, + "loss": 0.2602, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 6.60705041885376, + "learning_rate": 2e-05, + "loss": 0.9219, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 3.5278310775756836, + "learning_rate": 2e-05, + "loss": 0.2173, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 5.0895538330078125, + "learning_rate": 2e-05, + "loss": 0.7371, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 0.7779061794281006, + "learning_rate": 2e-05, + "loss": 0.557, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 1.9061380624771118, + "learning_rate": 2e-05, + "loss": 0.568, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 2.0734801292419434, + "learning_rate": 2e-05, + "loss": 0.2013, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 2.991359233856201, + "learning_rate": 2e-05, + "loss": 1.042, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 1.2337809801101685, + "learning_rate": 2e-05, + "loss": 0.5924, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 1.0585709810256958, + "learning_rate": 2e-05, + "loss": 0.5987, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 4.746422290802002, + "learning_rate": 2e-05, + "loss": 0.2653, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 6.219946384429932, + "learning_rate": 2e-05, + "loss": 0.9489, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 1.8796648979187012, + "learning_rate": 2e-05, + "loss": 0.2567, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 3.969242811203003, + "learning_rate": 2e-05, + "loss": 1.94, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 0.6053058505058289, + "learning_rate": 2e-05, + "loss": 0.7209, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 2.871701955795288, + "learning_rate": 2e-05, + "loss": 0.3984, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 4.533693313598633, + "learning_rate": 2e-05, + "loss": 0.7876, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 1.72139310836792, + "learning_rate": 2e-05, + "loss": 0.1649, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 0.3156372904777527, + "learning_rate": 2e-05, + "loss": 0.423, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 3.9974586963653564, + "learning_rate": 2e-05, + "loss": 1.0815, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 0.06653320044279099, + "learning_rate": 2e-05, + "loss": 0.63, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 2.5413870811462402, + "learning_rate": 2e-05, + "loss": 0.2142, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 0.27153152227401733, + "learning_rate": 2e-05, + "loss": 0.1234, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 9.618230819702148, + "learning_rate": 2e-05, + "loss": 1.7351, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 1.2576873302459717, + "learning_rate": 2e-05, + "loss": 0.2394, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 2.8308658599853516, + "learning_rate": 2e-05, + "loss": 0.8855, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 1.586479663848877, + "learning_rate": 2e-05, + "loss": 0.1994, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 4.185072422027588, + "learning_rate": 2e-05, + "loss": 0.4604, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 7.365324974060059, + "learning_rate": 2e-05, + "loss": 1.2657, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 1.6845974922180176, + "learning_rate": 2e-05, + "loss": 0.5656, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 7.1558308601379395, + "learning_rate": 2e-05, + "loss": 1.16, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 0.5682336091995239, + "learning_rate": 2e-05, + "loss": 0.7689, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 2.0558090209960938, + "learning_rate": 2e-05, + "loss": 0.962, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 7.233322620391846, + "learning_rate": 2e-05, + "loss": 1.5902, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 1.3614559173583984, + "learning_rate": 2e-05, + "loss": 0.2794, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 1.474043369293213, + "learning_rate": 2e-05, + "loss": 0.1001, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 2.1503713130950928, + "learning_rate": 2e-05, + "loss": 0.9227, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 6.022714614868164, + "learning_rate": 2e-05, + "loss": 0.9589, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 1.8667964935302734, + "learning_rate": 2e-05, + "loss": 0.4856, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 1.9206364154815674, + "learning_rate": 2e-05, + "loss": 0.2275, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 1.4191900491714478, + "learning_rate": 2e-05, + "loss": 0.2251, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 0.743366539478302, + "learning_rate": 2e-05, + "loss": 0.2968, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 2.904237985610962, + "learning_rate": 2e-05, + "loss": 0.5076, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 1.3773069381713867, + "learning_rate": 2e-05, + "loss": 0.2071, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5314727714488320.0, + "train_loss": 0.6246864032745362, + "train_runtime": 277.6091, + "train_samples_per_second": 1.441, + "train_steps_per_second": 0.36 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5314727714488320.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..8fe66d2781811c5645c5e2b195fa219bf9a7a582 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27ccc62289dc82a8760514b0cbc6568bc372e4c6329aac7c8e17b3676014d4d5 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..f0ee7555d38f6288346a21fb510eb800788b377a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e7aaff7368adefd724256217f9b20873a1a5a22fc9d06cc5025c1d1da634044 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..606036380dd7e3c2dcf134adc26756071d594460 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bff6ed70d894fd4bde31dd3d9eb480d54504384bd732a0c5612394623ca149dd +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..33234b586731d91758a27ac3a8cc6cb5a8cbea4b --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7511990365cb70a5ba9753a21eddc44f2a093aaede706b51086667a9ba101764 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..ffd2c061831df57e092e42a29bb6a5923025dfd0 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcaa6db0c3012281365bd2cbcbccfb6c02707e449adf0965817ed9f9ce866cc6 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..25d619db12183da35088db21f57a9e6c67e13f76 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:781c052467fe745189c8aa6f95773aec18ef3cd7d66783a37de1434bd4e627a5 +size 487636262 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..28afa3a6e59b5ef757f5b02c2596fa75df0a8a69 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d43ee120728e58fe2030cc8a923911cd54a153c54566cac3427f924ff0e2d81 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..4222f323a1f258e7f42a16e55a4971bf828f8cee --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:464a0b73d5beabb8a6145a93412e529e002ba2a79cad88e2efa06e703da590c2 +size 487635186 diff --git a/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..4a5c9581694c57aa8ebd0ddb2c5e9ccfc47939f1 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_pca_bs4_saveoptim_lr2e-5_sc1316_4tasks_5rounds_fixitr100_T0125_decay099/8_trainer_state.json @@ -0,0 +1,392 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 100, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02, + "grad_norm": 0.6196413636207581, + "learning_rate": 2e-05, + "loss": 0.1277, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 0.06034259498119354, + "learning_rate": 2e-05, + "loss": 0.1256, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 2.5118486881256104, + "learning_rate": 2e-05, + "loss": 0.2856, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 2.870713233947754, + "learning_rate": 2e-05, + "loss": 0.4674, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 0.022933989763259888, + "learning_rate": 2e-05, + "loss": 0.0032, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 1.3224023580551147, + "learning_rate": 2e-05, + "loss": 0.1759, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 0.08177559077739716, + "learning_rate": 2e-05, + "loss": 0.7011, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 5.685200214385986, + "learning_rate": 2e-05, + "loss": 0.3491, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 4.177793502807617, + "learning_rate": 2e-05, + "loss": 0.2841, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 5.158867835998535, + "learning_rate": 2e-05, + "loss": 3.5935, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 6.017444610595703, + "learning_rate": 2e-05, + "loss": 1.0031, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 1.6180634498596191, + "learning_rate": 2e-05, + "loss": 0.9357, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 1.280160903930664, + "learning_rate": 2e-05, + "loss": 0.2589, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 5.083629608154297, + "learning_rate": 2e-05, + "loss": 0.5791, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 1.6738539934158325, + "learning_rate": 2e-05, + "loss": 0.0866, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 0.025184577330946922, + "learning_rate": 2e-05, + "loss": 0.0094, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 4.564715385437012, + "learning_rate": 2e-05, + "loss": 1.7079, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 0.9750367403030396, + "learning_rate": 2e-05, + "loss": 0.5448, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.924103021621704, + "learning_rate": 2e-05, + "loss": 0.7789, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 2.195317029953003, + "learning_rate": 2e-05, + "loss": 0.9589, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 6.329569339752197, + "learning_rate": 2e-05, + "loss": 0.7875, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 0.007572048343718052, + "learning_rate": 2e-05, + "loss": 0.2973, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 2.1171188354492188, + "learning_rate": 2e-05, + "loss": 0.6948, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 4.389097690582275, + "learning_rate": 2e-05, + "loss": 0.6833, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 0.4250480830669403, + "learning_rate": 2e-05, + "loss": 0.0865, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 0.05439327284693718, + "learning_rate": 2e-05, + "loss": 0.6216, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 0.7046238780021667, + "learning_rate": 2e-05, + "loss": 0.6003, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 0.03884807601571083, + "learning_rate": 2e-05, + "loss": 0.3056, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 1.82545006275177, + "learning_rate": 2e-05, + "loss": 0.1344, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 0.11227758228778839, + "learning_rate": 2e-05, + "loss": 0.0125, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 1.3333393335342407, + "learning_rate": 2e-05, + "loss": 0.1389, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 8.498497009277344, + "learning_rate": 2e-05, + "loss": 2.123, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 3.8351352214813232, + "learning_rate": 2e-05, + "loss": 0.2997, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 3.265960931777954, + "learning_rate": 2e-05, + "loss": 0.157, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 2.2560501098632812, + "learning_rate": 2e-05, + "loss": 0.1224, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 2.8806986808776855, + "learning_rate": 2e-05, + "loss": 0.3414, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 1.6127833127975464, + "learning_rate": 2e-05, + "loss": 0.078, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 1.7134073972702026, + "learning_rate": 2e-05, + "loss": 0.1404, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 0.5160657167434692, + "learning_rate": 2e-05, + "loss": 0.4594, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 0.3315785825252533, + "learning_rate": 2e-05, + "loss": 0.0844, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 5.678584098815918, + "learning_rate": 2e-05, + "loss": 1.1721, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 0.2827218174934387, + "learning_rate": 2e-05, + "loss": 0.0254, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 0.30198338627815247, + "learning_rate": 2e-05, + "loss": 0.0221, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 6.353397369384766, + "learning_rate": 2e-05, + "loss": 0.5411, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 2.530363082885742, + "learning_rate": 2e-05, + "loss": 0.3266, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 3.9056644439697266, + "learning_rate": 2e-05, + "loss": 0.7858, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 0.11908189207315445, + "learning_rate": 2e-05, + "loss": 0.0128, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 2.016356945037842, + "learning_rate": 2e-05, + "loss": 0.202, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 1.4409465789794922, + "learning_rate": 2e-05, + "loss": 0.0936, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 0.9090744256973267, + "learning_rate": 2e-05, + "loss": 0.0919, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5173270752002048.0, + "train_loss": 0.4883705484867096, + "train_runtime": 281.7992, + "train_samples_per_second": 1.419, + "train_steps_per_second": 0.355 + } + ], + "logging_steps": 2, + "max_steps": 100, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5173270752002048.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}