diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..3cf76a7747f079aceb429b87fca436f8e7d9295c --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4be795e425a5d7075f02bcfe4fdba702b3f4a3e2e6a3d24568b5ecb7742b9be4 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..d41cb79d3ade3cea75a57004ca4249bff2e7bee4 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:887e60b457a7171a821e7871a0ac1e3d1f6b636237460b3a009dca4517d16002 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..75081de596f7eb57a62048fb2482e8904264ab5d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78830ea00219f3ab8d7def684eaf76650219e7485e437546e1063cb641384b16 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..bb21dd4b7a6b2f86b446624f8f48e7fdad8e5b74 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd7c02543def5cda3ee658b2596dfe8f19dad3f233f2edc88a7ad60838375445 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..4af42eb17c17807225cac78fb4ded237e5409eb7 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42460728f6212b8e84faeae4fb69f1132392c13071247a6520899c51c0794157 +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..b6a2cd2a082901f92020f93a6995ebfeafa36edb --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b0566cecd95b8cfd97288a78a54d0276b978b8e14bc10a02dc4e3acde077ca9 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..a889639b39e04dc0dfcb8ca348522c97514b5317 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22a5bda513396bb15e9746b4c13c779f4a11be27581445b4712425b479007d3a +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..700d0c955f49185c06913abba6536ddae3b15c9c --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bef6a46f1defc563a93d9ae0e0aa84ce4dd35f623f57c366a21b7e660cbe9c90 +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/0_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..5b9f237912083be53a70706847f453cb6b067503 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 4.8568243980407715, + "learning_rate": 2e-05, + "loss": 0.9372, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 4.345067977905273, + "learning_rate": 2e-05, + "loss": 0.9044, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 12.120655059814453, + "learning_rate": 2e-05, + "loss": 1.2838, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 5.687979698181152, + "learning_rate": 2e-05, + "loss": 2.4326, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 1.3347970247268677, + "learning_rate": 2e-05, + "loss": 0.3054, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 2.94572114944458, + "learning_rate": 2e-05, + "loss": 0.476, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 8.041915893554688, + "learning_rate": 2e-05, + "loss": 2.3675, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 2.2988779544830322, + "learning_rate": 2e-05, + "loss": 0.8773, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 10.082386016845703, + "learning_rate": 2e-05, + "loss": 1.2354, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 4.044953346252441, + "learning_rate": 2e-05, + "loss": 1.0228, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 5.216516494750977, + "learning_rate": 2e-05, + "loss": 0.8611, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 4.001774311065674, + "learning_rate": 2e-05, + "loss": 0.7416, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 5.1101155281066895, + "learning_rate": 2e-05, + "loss": 1.0476, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 6.11592960357666, + "learning_rate": 2e-05, + "loss": 0.9232, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 6.683459281921387, + "learning_rate": 2e-05, + "loss": 1.203, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 4.920234203338623, + "learning_rate": 2e-05, + "loss": 1.3295, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.9107036590576172, + "learning_rate": 2e-05, + "loss": 0.492, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 7.40504264831543, + "learning_rate": 2e-05, + "loss": 1.6457, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 8.718831062316895, + "learning_rate": 2e-05, + "loss": 2.8188, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 4.62222146987915, + "learning_rate": 2e-05, + "loss": 1.4451, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 7.5993452072143555, + "learning_rate": 2e-05, + "loss": 1.1404, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 11.086434364318848, + "learning_rate": 2e-05, + "loss": 0.7555, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 4.652540683746338, + "learning_rate": 2e-05, + "loss": 1.1318, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 3.010275363922119, + "learning_rate": 2e-05, + "loss": 0.5329, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 7.997224807739258, + "learning_rate": 2e-05, + "loss": 1.2541, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 3.179985284805298, + "learning_rate": 2e-05, + "loss": 0.4904, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 10.767970085144043, + "learning_rate": 2e-05, + "loss": 2.2404, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 5.549088001251221, + "learning_rate": 2e-05, + "loss": 1.4976, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 11.436861991882324, + "learning_rate": 2e-05, + "loss": 1.1036, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 7.861836910247803, + "learning_rate": 2e-05, + "loss": 0.8495, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 8.143918991088867, + "learning_rate": 2e-05, + "loss": 1.3565, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 4.367578029632568, + "learning_rate": 2e-05, + "loss": 1.9117, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 3.4422531127929688, + "learning_rate": 2e-05, + "loss": 0.2458, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 0.6072515845298767, + "learning_rate": 2e-05, + "loss": 0.4135, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 4.17832612991333, + "learning_rate": 2e-05, + "loss": 0.5555, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 4.957637786865234, + "learning_rate": 2e-05, + "loss": 0.7625, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 5.490946292877197, + "learning_rate": 2e-05, + "loss": 1.3255, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 2.61025071144104, + "learning_rate": 2e-05, + "loss": 0.7288, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 2.5497350692749023, + "learning_rate": 2e-05, + "loss": 1.277, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 5.070952892303467, + "learning_rate": 2e-05, + "loss": 1.5686, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 4.195366382598877, + "learning_rate": 2e-05, + "loss": 1.1058, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 4.0397443771362305, + "learning_rate": 2e-05, + "loss": 1.0513, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 2.9195706844329834, + "learning_rate": 2e-05, + "loss": 1.0052, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 6.697042465209961, + "learning_rate": 2e-05, + "loss": 1.3321, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 2.519200325012207, + "learning_rate": 2e-05, + "loss": 0.4151, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 3.2042155265808105, + "learning_rate": 2e-05, + "loss": 0.4957, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 3.8642756938934326, + "learning_rate": 2e-05, + "loss": 0.8047, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 4.016802787780762, + "learning_rate": 2e-05, + "loss": 0.7323, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 5.384059429168701, + "learning_rate": 2e-05, + "loss": 1.413, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 6.8328022956848145, + "learning_rate": 2e-05, + "loss": 1.0836, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5462826382524416.0, + "train_loss": 1.0986122703552246, + "train_runtime": 130.3604, + "train_samples_per_second": 3.068, + "train_steps_per_second": 0.767 + } + ], + "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": 5462826382524416.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..f7eeb127a1455527e6e14d9dd63a61933e765a06 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4edbe0cbb17d2990d159c0027f168bc857cbfb1a6bb449af22e6962cd89679f +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..d93746f1e102f765a2567b90fb9273da24163742 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45d6877435499f1e92528d26dbe10cb4c00a6fdc4fbf7631f55f64f0aaff855c +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..7205aedede54b6a7127a82ceac8bf99058e5a04b --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90db85b5c464796d6a8bf43529e6b677026cc1a1c7b6ff1728a14708ac21f19a +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..1baec8220b8dca79bebda98a20b9ad9169f51238 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecfa816282c26cd8f9f71d7a2b6905d6343a7a07b564523e6ecc2a0da324ee0f +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..1eb66b3b1c15bb118ab0600599d9723377308e9a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b63e55b69b38ac055d7be4a13a4cdcb92abb460717b45b7b24d7405f9ab8c21 +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..689c02afbde3f9521edcd1efed29c59842b493eb --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36d83d0d78759cce8c50907c09cc9983be8eb443ec79b7712eb373aba07ca8a0 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..47311b01fa870a2d4ec0af63e589eecc9a548fcb --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bd6627e16a20f099893f0d485c73a444e4df4ef3ca9cad9a95eadcfb6d943dd +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..57bcb623ac7b671da41d017c4ab32aadc26e344e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d13ccc291e872fe938a37933a665286eface4e5c9a0b4d14f9dfd930dccac439 +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/1_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..4c34f61e4074368e3aae77686ce093d7de714d03 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 11.147574424743652, + "learning_rate": 2e-05, + "loss": 1.3265, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 6.276861190795898, + "learning_rate": 2e-05, + "loss": 1.1753, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 7.431264400482178, + "learning_rate": 2e-05, + "loss": 1.4501, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 2.3549540042877197, + "learning_rate": 2e-05, + "loss": 0.7166, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 5.00397253036499, + "learning_rate": 2e-05, + "loss": 1.1938, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 4.596293926239014, + "learning_rate": 2e-05, + "loss": 0.983, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 8.429619789123535, + "learning_rate": 2e-05, + "loss": 1.1644, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 4.940812110900879, + "learning_rate": 2e-05, + "loss": 1.3457, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 3.7962863445281982, + "learning_rate": 2e-05, + "loss": 1.1819, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 8.068099975585938, + "learning_rate": 2e-05, + "loss": 1.009, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 2.047276258468628, + "learning_rate": 2e-05, + "loss": 0.9596, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 7.590549945831299, + "learning_rate": 2e-05, + "loss": 1.3425, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 4.109409809112549, + "learning_rate": 2e-05, + "loss": 0.9021, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 6.835687637329102, + "learning_rate": 2e-05, + "loss": 0.9056, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 3.146049976348877, + "learning_rate": 2e-05, + "loss": 1.2231, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 4.957376956939697, + "learning_rate": 2e-05, + "loss": 1.4825, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 3.7589638233184814, + "learning_rate": 2e-05, + "loss": 1.3005, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 12.209796905517578, + "learning_rate": 2e-05, + "loss": 1.1377, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 6.60794734954834, + "learning_rate": 2e-05, + "loss": 1.1643, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 4.360702991485596, + "learning_rate": 2e-05, + "loss": 1.1243, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 7.058619022369385, + "learning_rate": 2e-05, + "loss": 1.0932, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 2.354834794998169, + "learning_rate": 2e-05, + "loss": 1.4331, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 3.1146738529205322, + "learning_rate": 2e-05, + "loss": 1.3581, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 4.396064758300781, + "learning_rate": 2e-05, + "loss": 0.9706, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 4.462273597717285, + "learning_rate": 2e-05, + "loss": 1.08, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 7.432363510131836, + "learning_rate": 2e-05, + "loss": 1.8945, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 4.282905578613281, + "learning_rate": 2e-05, + "loss": 1.153, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 2.7390224933624268, + "learning_rate": 2e-05, + "loss": 0.6751, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 8.883049011230469, + "learning_rate": 2e-05, + "loss": 0.9915, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 4.453659534454346, + "learning_rate": 2e-05, + "loss": 0.9824, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 4.341336250305176, + "learning_rate": 2e-05, + "loss": 1.8174, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 2.2057862281799316, + "learning_rate": 2e-05, + "loss": 0.6967, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 5.3489861488342285, + "learning_rate": 2e-05, + "loss": 1.48, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 12.532249450683594, + "learning_rate": 2e-05, + "loss": 1.4761, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 3.928640365600586, + "learning_rate": 2e-05, + "loss": 1.1903, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 4.719779968261719, + "learning_rate": 2e-05, + "loss": 1.3064, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 3.3434858322143555, + "learning_rate": 2e-05, + "loss": 0.9087, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 8.262925148010254, + "learning_rate": 2e-05, + "loss": 1.1738, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 2.6751790046691895, + "learning_rate": 2e-05, + "loss": 0.9931, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 9.4998197555542, + "learning_rate": 2e-05, + "loss": 1.3631, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 1.2977250814437866, + "learning_rate": 2e-05, + "loss": 0.9746, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 7.485049724578857, + "learning_rate": 2e-05, + "loss": 1.0304, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 8.568105697631836, + "learning_rate": 2e-05, + "loss": 1.1331, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 4.983133316040039, + "learning_rate": 2e-05, + "loss": 1.3687, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 10.139394760131836, + "learning_rate": 2e-05, + "loss": 1.2744, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 8.025995254516602, + "learning_rate": 2e-05, + "loss": 1.285, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 14.446391105651855, + "learning_rate": 2e-05, + "loss": 1.0137, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 4.424881935119629, + "learning_rate": 2e-05, + "loss": 1.3005, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 3.0400640964508057, + "learning_rate": 2e-05, + "loss": 0.9536, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 4.7732696533203125, + "learning_rate": 2e-05, + "loss": 1.5835, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 5436726839345152.0, + "train_loss": 1.1808644104003907, + "train_runtime": 130.6818, + "train_samples_per_second": 3.061, + "train_steps_per_second": 0.765 + } + ], + "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": 5436726839345152.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..32f00f5509cffd9374f90948ef0dfcf54040b1d2 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1dc4aafd5bfdf57ef9aad42de2e74a7ac673e61cf2c722f0c95701bfa756cfa8 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..af4ee4a29d20b39e54ff8fc7f5fca8b5c72aacf5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46edfaafcc538df7834bdf30131ba6a4bb123364e6dde600b7ef807a23ed65af +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..f23e3652fe6575c4941c3113ee333d60de41a342 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe659fe724a0abd4f054f1ac7d2f5e441f32774d6c9232934e3c1d25d94edc8f +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..084121572d27139cbd43ddccd33439da7eccf509 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c74d2e3920345856a13f9560ef7c49a1f452c76a757e6d685064411180785d4e +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..edc0539052fbe523c48991318a3e33aa20e0f077 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:193ef6ece51f6f5e122228060eb01453e17b76c19aec098941c5a36ab4cf05e2 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..8c3ae3076a6701e44e9c30539b430446b7d4af0d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3dc33b8640d27b19a6e368461705556e6fc7a941e186befb2177025b283d6b87 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..65093240278dcb9f1e44ba04cfa48ae8c64b621c --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f44c40d282adf3fa4c7e8ab2c068b365fead07fb46193b2f764fb1d09b5c64a9 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..2267e72d6d9c2bafe313d9ff298228c62f095b9b --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef20b42802c8eb5161ff1f26d33dcc91a0b3f544c3183f0505a9ff3ab6b03953 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/2_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..b539369701371eb124cefbea64df17163b72bd74 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 7.4485626220703125, + "learning_rate": 2e-05, + "loss": 0.9813, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 1.9862580299377441, + "learning_rate": 2e-05, + "loss": 0.3618, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 5.480435848236084, + "learning_rate": 2e-05, + "loss": 0.7232, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 6.3666605949401855, + "learning_rate": 2e-05, + "loss": 0.5558, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 8.656007766723633, + "learning_rate": 2e-05, + "loss": 1.5084, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 4.521378040313721, + "learning_rate": 2e-05, + "loss": 1.0171, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 7.8770294189453125, + "learning_rate": 2e-05, + "loss": 1.1928, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 2.125856876373291, + "learning_rate": 2e-05, + "loss": 0.9306, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 5.606194496154785, + "learning_rate": 2e-05, + "loss": 1.0293, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 6.488819599151611, + "learning_rate": 2e-05, + "loss": 0.7309, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 5.991289138793945, + "learning_rate": 2e-05, + "loss": 0.8452, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 3.130293607711792, + "learning_rate": 2e-05, + "loss": 1.3333, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 2.767559766769409, + "learning_rate": 2e-05, + "loss": 0.6951, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 7.329335689544678, + "learning_rate": 2e-05, + "loss": 1.1264, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 2.395749092102051, + "learning_rate": 2e-05, + "loss": 0.8407, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 2.593240737915039, + "learning_rate": 2e-05, + "loss": 0.6638, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 5.472085475921631, + "learning_rate": 2e-05, + "loss": 0.7324, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 7.004269123077393, + "learning_rate": 2e-05, + "loss": 1.0726, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.786731243133545, + "learning_rate": 2e-05, + "loss": 0.6328, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 6.682612419128418, + "learning_rate": 2e-05, + "loss": 0.6289, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 7.152703762054443, + "learning_rate": 2e-05, + "loss": 1.2498, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 3.1759555339813232, + "learning_rate": 2e-05, + "loss": 0.3251, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 2.7986321449279785, + "learning_rate": 2e-05, + "loss": 0.9933, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 1.8032745122909546, + "learning_rate": 2e-05, + "loss": 0.5448, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 3.621896982192993, + "learning_rate": 2e-05, + "loss": 1.1534, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 5.912569046020508, + "learning_rate": 2e-05, + "loss": 0.8669, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 1.9287352561950684, + "learning_rate": 2e-05, + "loss": 0.7631, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 1.8568964004516602, + "learning_rate": 2e-05, + "loss": 0.4661, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 2.823870897293091, + "learning_rate": 2e-05, + "loss": 0.6656, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 3.9624710083007812, + "learning_rate": 2e-05, + "loss": 1.1263, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 18.075254440307617, + "learning_rate": 2e-05, + "loss": 2.0835, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 1.5604605674743652, + "learning_rate": 2e-05, + "loss": 0.6604, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 5.896483898162842, + "learning_rate": 2e-05, + "loss": 1.2012, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 7.719961166381836, + "learning_rate": 2e-05, + "loss": 0.7057, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 7.1494951248168945, + "learning_rate": 2e-05, + "loss": 0.8958, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 2.3935749530792236, + "learning_rate": 2e-05, + "loss": 0.5964, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 3.0517525672912598, + "learning_rate": 2e-05, + "loss": 1.0687, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 4.149694919586182, + "learning_rate": 2e-05, + "loss": 1.4968, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 1.5019639730453491, + "learning_rate": 2e-05, + "loss": 0.1589, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 2.0508153438568115, + "learning_rate": 2e-05, + "loss": 0.7247, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 7.231861114501953, + "learning_rate": 2e-05, + "loss": 0.732, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 3.0040225982666016, + "learning_rate": 2e-05, + "loss": 0.4023, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 4.380476951599121, + "learning_rate": 2e-05, + "loss": 0.5608, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 9.996171951293945, + "learning_rate": 2e-05, + "loss": 1.1683, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 2.28837251663208, + "learning_rate": 2e-05, + "loss": 0.4167, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 3.0701327323913574, + "learning_rate": 2e-05, + "loss": 0.7562, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 6.403076648712158, + "learning_rate": 2e-05, + "loss": 2.0225, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 7.473080635070801, + "learning_rate": 2e-05, + "loss": 1.862, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 7.931491851806641, + "learning_rate": 2e-05, + "loss": 1.4611, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 11.673038482666016, + "learning_rate": 2e-05, + "loss": 1.4778, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 1.4596800069500928e+16, + "train_loss": 0.9241710472106933, + "train_runtime": 285.2412, + "train_samples_per_second": 1.402, + "train_steps_per_second": 0.351 + } + ], + "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": 1.4596800069500928e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..98e5b02a99a415e0ba14266a34ecea7aac5f1cac --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88e8987f26740c7009f1f5f56057fbc4b5f96a85ae20b53dfa1e18194705f022 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..111974ecc371852fbb10815ef7581ff4974ae4d6 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba53466426b547d1aec2ccf59cc6308fee74e0f8fb734a2954d3034d3ce2a281 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..6f9034557a13ba4f401c15d268f82e986990e533 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:439bd1eb80adb68fbcc840c96bcacf823288cbaf730476808fb3394ff6ad1df3 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..cfa33b8d88f2f363ba0167dec216af4b0ef2af9e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c86ab622fe1feb23b59fb83cf5e472ee71c79dfd4f20eb96701f7393919679e6 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..fa512049c589bebe58fdc0559ca5954d0f75ca18 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2f5547d82e18f42b23227f391bcddd915c066de4230694981cada509211d187 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..40f62bfe17ce58ccf1840136f3a9ba8b26fc41d8 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:879aa734b796b610505a95fc544fd8d2e58f5306fd4ff0b9e059b542dca1fcc8 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..2d64e6b337afe322f8d2d3b55fa6031aecec3724 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36db059cda41bca570abcc3af1cb8b8a3a686963511afb00352aa80ecfa4d6cd +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..59dda69b400137702dbddcb0b314067bc7f0025e --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca8e5d8dd79a4cab3e6cc3b4f9b2ad4b29b3e9129607c8bbb7680422d6cce54f +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/3_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..007953ce1fa0dd07863c406f0e6c0329cf06dd4a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 5.035049915313721, + "learning_rate": 2e-05, + "loss": 1.1652, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 3.3497910499572754, + "learning_rate": 2e-05, + "loss": 1.6484, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 2.469871997833252, + "learning_rate": 2e-05, + "loss": 0.737, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 6.721245288848877, + "learning_rate": 2e-05, + "loss": 1.0691, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 4.207581996917725, + "learning_rate": 2e-05, + "loss": 1.1694, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 4.003703594207764, + "learning_rate": 2e-05, + "loss": 0.6077, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 3.2081942558288574, + "learning_rate": 2e-05, + "loss": 0.6724, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 2.170851230621338, + "learning_rate": 2e-05, + "loss": 1.0347, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 2.2809324264526367, + "learning_rate": 2e-05, + "loss": 1.4614, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 1.2174735069274902, + "learning_rate": 2e-05, + "loss": 0.9074, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 6.532240867614746, + "learning_rate": 2e-05, + "loss": 0.824, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 9.217572212219238, + "learning_rate": 2e-05, + "loss": 2.1938, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 1.5741959810256958, + "learning_rate": 2e-05, + "loss": 0.8184, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 4.815344333648682, + "learning_rate": 2e-05, + "loss": 1.5176, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 1.7724013328552246, + "learning_rate": 2e-05, + "loss": 1.5767, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 3.960803747177124, + "learning_rate": 2e-05, + "loss": 1.0888, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 3.7917325496673584, + "learning_rate": 2e-05, + "loss": 0.983, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 7.79496431350708, + "learning_rate": 2e-05, + "loss": 1.0458, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 8.211302757263184, + "learning_rate": 2e-05, + "loss": 1.1427, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 3.298877239227295, + "learning_rate": 2e-05, + "loss": 1.0652, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 6.475459098815918, + "learning_rate": 2e-05, + "loss": 0.7741, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 5.8340888023376465, + "learning_rate": 2e-05, + "loss": 1.1124, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 4.417236804962158, + "learning_rate": 2e-05, + "loss": 1.0009, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 3.54665470123291, + "learning_rate": 2e-05, + "loss": 0.6321, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 1.6683158874511719, + "learning_rate": 2e-05, + "loss": 1.3171, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 0.7134782671928406, + "learning_rate": 2e-05, + "loss": 0.3386, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 2.1631276607513428, + "learning_rate": 2e-05, + "loss": 0.9637, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 2.5291998386383057, + "learning_rate": 2e-05, + "loss": 1.1431, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 3.054178237915039, + "learning_rate": 2e-05, + "loss": 0.7363, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 2.6544301509857178, + "learning_rate": 2e-05, + "loss": 0.7261, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 2.754007339477539, + "learning_rate": 2e-05, + "loss": 0.8795, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 2.406623601913452, + "learning_rate": 2e-05, + "loss": 1.0624, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 3.3705132007598877, + "learning_rate": 2e-05, + "loss": 1.2757, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 15.125097274780273, + "learning_rate": 2e-05, + "loss": 1.0644, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 2.651677131652832, + "learning_rate": 2e-05, + "loss": 0.6554, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 3.574554920196533, + "learning_rate": 2e-05, + "loss": 0.7245, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 6.320649147033691, + "learning_rate": 2e-05, + "loss": 0.7212, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 7.0648016929626465, + "learning_rate": 2e-05, + "loss": 1.2978, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 2.5221383571624756, + "learning_rate": 2e-05, + "loss": 1.1599, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 1.474943995475769, + "learning_rate": 2e-05, + "loss": 0.5594, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 2.3496599197387695, + "learning_rate": 2e-05, + "loss": 0.9978, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 4.546678066253662, + "learning_rate": 2e-05, + "loss": 1.222, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 4.939904689788818, + "learning_rate": 2e-05, + "loss": 1.1428, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 6.095272064208984, + "learning_rate": 2e-05, + "loss": 1.1364, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 4.423612117767334, + "learning_rate": 2e-05, + "loss": 1.5325, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 5.676146984100342, + "learning_rate": 2e-05, + "loss": 0.8407, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 1.8238693475723267, + "learning_rate": 2e-05, + "loss": 0.8097, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 3.0705935955047607, + "learning_rate": 2e-05, + "loss": 0.8379, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 3.3262546062469482, + "learning_rate": 2e-05, + "loss": 0.5019, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 2.406374454498291, + "learning_rate": 2e-05, + "loss": 0.4392, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 2.037956633375539e+16, + "train_loss": 1.0066848659515382, + "train_runtime": 374.3247, + "train_samples_per_second": 1.069, + "train_steps_per_second": 0.267 + } + ], + "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": 2.037956633375539e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..d74c0b19bf6b4af03ba58f73081876547516fe10 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b77fab24eb8b332cd18fd8c613d655a23e63cbef32b593244d37b471987aac31 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..94fbea5ccb5a9ccae70aa7edd231495b49fbacc6 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9e5894f1f2391d28f0a2518e63ae6d7ae31cf51cdc672da5551adbb73951be3 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..0bad58696b75702ad64d86e2cdb2071659bbbeef --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6d983c539ab6a47bd8c88d749b581aea3e7b30e4e7a24eeaa1beac037fc01c2 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..72e90de2ced45b956bc24e28f3f9ccc4426c80bf --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa14bb8ccd78b4e38f462af106f27d59d2d84d7a6fe27f505e77f91ee4cc3a64 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..99cd9a5545217f8abe06f9dda1b51d4496066953 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b607ca226260ff2c5836d568daf269f77c26d3177e0d05b353360338b82fb2b +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..1bcbe5d08e3e2d6941200228dde71dc698eacac9 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84cf7a57d19cfa7b4eea900c6474a8680b4807fdda2008ecdf5df80fcb3ee27c +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..37eda7932cd040a0752878c70055630e9b093f24 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d518f368b997bbe01a591d170363aba7dd9aa337cb40fea2050fbc3a8f66c08a +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..e397846cf433589b5de90706c572edf366c02803 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0bcb452a6784fbb73d3479cde29cf4c15f889dc79f087ad04d73d0b2bfc6633 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/4_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..d62326bd8b671ac8fcea4acbdc827dee3bc279e8 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 1.189469814300537, + "learning_rate": 2e-05, + "loss": 0.31, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 2.8435537815093994, + "learning_rate": 2e-05, + "loss": 0.5298, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 5.566661834716797, + "learning_rate": 2e-05, + "loss": 1.2779, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 1.0178768634796143, + "learning_rate": 2e-05, + "loss": 0.4592, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 9.928922653198242, + "learning_rate": 2e-05, + "loss": 1.2289, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 1.7801988124847412, + "learning_rate": 2e-05, + "loss": 1.1655, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 0.27690789103507996, + "learning_rate": 2e-05, + "loss": 0.2859, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 4.786653995513916, + "learning_rate": 2e-05, + "loss": 1.0218, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 1.004693627357483, + "learning_rate": 2e-05, + "loss": 1.154, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 4.915676593780518, + "learning_rate": 2e-05, + "loss": 0.2657, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 6.210068225860596, + "learning_rate": 2e-05, + "loss": 0.7747, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 4.257296085357666, + "learning_rate": 2e-05, + "loss": 1.6313, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 7.215267181396484, + "learning_rate": 2e-05, + "loss": 3.7582, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 7.644487380981445, + "learning_rate": 2e-05, + "loss": 2.2401, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 5.116546630859375, + "learning_rate": 2e-05, + "loss": 1.995, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 3.723872184753418, + "learning_rate": 2e-05, + "loss": 0.8439, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 4.7276716232299805, + "learning_rate": 2e-05, + "loss": 1.0697, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 3.0036442279815674, + "learning_rate": 2e-05, + "loss": 1.7633, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.7193286418914795, + "learning_rate": 2e-05, + "loss": 0.5447, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 6.585648059844971, + "learning_rate": 2e-05, + "loss": 1.3138, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 0.06341306120157242, + "learning_rate": 2e-05, + "loss": 0.5346, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 3.21247935295105, + "learning_rate": 2e-05, + "loss": 1.5504, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 2.877152442932129, + "learning_rate": 2e-05, + "loss": 1.1479, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 2.434115171432495, + "learning_rate": 2e-05, + "loss": 0.7492, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 4.6888427734375, + "learning_rate": 2e-05, + "loss": 1.0197, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 4.457335948944092, + "learning_rate": 2e-05, + "loss": 1.9108, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 0.416705459356308, + "learning_rate": 2e-05, + "loss": 0.7948, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 1.0772849321365356, + "learning_rate": 2e-05, + "loss": 0.4287, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 3.1008191108703613, + "learning_rate": 2e-05, + "loss": 1.094, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 2.63857102394104, + "learning_rate": 2e-05, + "loss": 0.7174, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 2.9035606384277344, + "learning_rate": 2e-05, + "loss": 0.814, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 6.539764881134033, + "learning_rate": 2e-05, + "loss": 1.699, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 3.706289529800415, + "learning_rate": 2e-05, + "loss": 1.649, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 0.7016932368278503, + "learning_rate": 2e-05, + "loss": 0.2642, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 3.280506134033203, + "learning_rate": 2e-05, + "loss": 1.2083, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 7.748496055603027, + "learning_rate": 2e-05, + "loss": 1.104, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 2.6094746589660645, + "learning_rate": 2e-05, + "loss": 1.2405, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 3.584652900695801, + "learning_rate": 2e-05, + "loss": 0.9338, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 6.3299102783203125, + "learning_rate": 2e-05, + "loss": 1.2104, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 6.807102203369141, + "learning_rate": 2e-05, + "loss": 1.6096, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 4.525906562805176, + "learning_rate": 2e-05, + "loss": 1.0336, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 8.219979286193848, + "learning_rate": 2e-05, + "loss": 2.3766, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 3.3661632537841797, + "learning_rate": 2e-05, + "loss": 0.4931, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 8.628079414367676, + "learning_rate": 2e-05, + "loss": 1.5695, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 4.711212635040283, + "learning_rate": 2e-05, + "loss": 0.7535, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 8.135887145996094, + "learning_rate": 2e-05, + "loss": 1.7705, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 3.585470676422119, + "learning_rate": 2e-05, + "loss": 0.6988, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 3.6448802947998047, + "learning_rate": 2e-05, + "loss": 1.502, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 4.504702568054199, + "learning_rate": 2e-05, + "loss": 0.5163, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 3.5633938312530518, + "learning_rate": 2e-05, + "loss": 0.9534, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 1.920155780330291e+16, + "train_loss": 1.1396221828460693, + "train_runtime": 369.8727, + "train_samples_per_second": 1.081, + "train_steps_per_second": 0.27 + } + ], + "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": 1.920155780330291e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..7a3204fd08a386e9168625df5a1a4b9d071e5554 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f65d8e8208f1f43a118be2af13f6c0657768c6cf2f6a6c8821826f6c08f45ac9 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..53108700f80d5fa08821183c6a532eb870cbc9ba --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e793cff570be69867002683a9ef9c4425eda39e30bec1773d52551340007a66f +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..aa53b3bdee58421841cbbd115a75cdb6cf976c3d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8857a69c6f759d8234f7d396fc37cb7e39d7bde1d4083fbff4994ba63e14f244 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..33385020f1db980e3528ffd1615d878ca19c512f --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:363dce9b6bbff794e49f733aaea7a3d6287dd7fb30dea5cec0bb128abe809134 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..280d656c0b39355a9345f21b9582e6921a1f9768 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d5c7a96bf536e66bd688162ae2f559b164db041a290865bc5a303e90040fb9 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..2b726fd2497b7816b3ceaeffa5d8ed1cc66b4dea --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c91d18dafcf349c39c97f0f1315ab6636b0baacdffacae1259a34ddba3c602b +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..e948bb6c90b84e85cbf3774ede9cc224cd37ca79 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb783bbcda53b936f2f168edc48e9c83c25d0d29d6ad5772d4ca84ccb68e4cfc +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..45252a2ec38779c8f77a94d3748ad63152bdbf6a --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0cd344a13f45368d059e957be322d03d69481003adbb7daeb7e95aa5a5810b0 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/5_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..26491cb64ad32dd01afee06c3853bc5a186cd91b --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 3.7614798545837402, + "learning_rate": 2e-05, + "loss": 1.7094, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 0.14592814445495605, + "learning_rate": 2e-05, + "loss": 0.5169, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 0.07696948945522308, + "learning_rate": 2e-05, + "loss": 0.6507, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 1.6947181224822998, + "learning_rate": 2e-05, + "loss": 0.5144, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 0.4415757656097412, + "learning_rate": 2e-05, + "loss": 0.4157, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 4.0385355949401855, + "learning_rate": 2e-05, + "loss": 0.995, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 1.51202392578125, + "learning_rate": 2e-05, + "loss": 0.1536, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 4.5873870849609375, + "learning_rate": 2e-05, + "loss": 0.7368, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 2.2251596450805664, + "learning_rate": 2e-05, + "loss": 0.6487, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 2.2275595664978027, + "learning_rate": 2e-05, + "loss": 0.5625, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 3.4227261543273926, + "learning_rate": 2e-05, + "loss": 1.2273, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 3.2022793292999268, + "learning_rate": 2e-05, + "loss": 0.9876, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 1.3675581216812134, + "learning_rate": 2e-05, + "loss": 1.9748, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 0.44088882207870483, + "learning_rate": 2e-05, + "loss": 0.0797, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 0.12196239829063416, + "learning_rate": 2e-05, + "loss": 0.2387, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 2.4486653804779053, + "learning_rate": 2e-05, + "loss": 0.8451, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 2.7781643867492676, + "learning_rate": 2e-05, + "loss": 0.7008, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 3.694227695465088, + "learning_rate": 2e-05, + "loss": 1.2565, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.878048896789551, + "learning_rate": 2e-05, + "loss": 0.5124, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 1.6161788702011108, + "learning_rate": 2e-05, + "loss": 0.3762, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 2.1570777893066406, + "learning_rate": 2e-05, + "loss": 0.3512, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 2.194190740585327, + "learning_rate": 2e-05, + "loss": 1.3605, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 3.8701064586639404, + "learning_rate": 2e-05, + "loss": 0.7972, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 2.7052366733551025, + "learning_rate": 2e-05, + "loss": 0.3978, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 0.9993058443069458, + "learning_rate": 2e-05, + "loss": 0.0748, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 5.839305877685547, + "learning_rate": 2e-05, + "loss": 0.5265, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 1.3489971160888672, + "learning_rate": 2e-05, + "loss": 0.2132, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 3.394754409790039, + "learning_rate": 2e-05, + "loss": 0.398, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 2.830280065536499, + "learning_rate": 2e-05, + "loss": 0.5341, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 10.30225658416748, + "learning_rate": 2e-05, + "loss": 1.8645, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 2.828972578048706, + "learning_rate": 2e-05, + "loss": 0.5225, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 1.2658295631408691, + "learning_rate": 2e-05, + "loss": 0.8316, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 8.20359992980957, + "learning_rate": 2e-05, + "loss": 0.6559, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 4.623135566711426, + "learning_rate": 2e-05, + "loss": 1.0693, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 0.7521807551383972, + "learning_rate": 2e-05, + "loss": 0.5186, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 1.7425742149353027, + "learning_rate": 2e-05, + "loss": 0.166, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 2.5554254055023193, + "learning_rate": 2e-05, + "loss": 0.8117, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 1.9501384496688843, + "learning_rate": 2e-05, + "loss": 0.7333, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 9.361940383911133, + "learning_rate": 2e-05, + "loss": 1.1752, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 1.1605331897735596, + "learning_rate": 2e-05, + "loss": 0.3622, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 4.790265083312988, + "learning_rate": 2e-05, + "loss": 0.6556, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 5.6687822341918945, + "learning_rate": 2e-05, + "loss": 0.8543, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 0.021462885662913322, + "learning_rate": 2e-05, + "loss": 0.3529, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 0.41082873940467834, + "learning_rate": 2e-05, + "loss": 0.0443, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 1.9942395687103271, + "learning_rate": 2e-05, + "loss": 0.5919, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 2.959017038345337, + "learning_rate": 2e-05, + "loss": 0.7349, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 1.7007932662963867, + "learning_rate": 2e-05, + "loss": 0.4445, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 3.450399160385132, + "learning_rate": 2e-05, + "loss": 0.3417, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 4.364017009735107, + "learning_rate": 2e-05, + "loss": 1.0934, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 2.784651279449463, + "learning_rate": 2e-05, + "loss": 0.8364, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 2.475445626024755e+16, + "train_loss": 0.6883406066894531, + "train_runtime": 475.1804, + "train_samples_per_second": 0.842, + "train_steps_per_second": 0.21 + } + ], + "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": 2.475445626024755e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..f8e4c4e96a633f1d8d1b9a53a3c8aede28bada64 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c4448c457e0f58cfee3a723f0e312aa5db09b6b247b6a2e099eeb8b01ca52bb +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..2325ebea6d1e955483711e61ec4db75f635471d1 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:975653be82971a3352f1ad6756651edd4837c17b22e306b001f1e55c7dec5faf +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..5588bad1e3dd8c87734798345f940f10a7ee90cc --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a507e9693979e340dc03a738f9b35652b14fb0fb304349ff22bcb111c6c08abc +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..f25c0471afd99b60c7a6ee38f1f23abac03db563 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e5629e3daef9e03b01d9946d02a615b00cb8691a8d1ed39d37d442645d61809 +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..e78fa4954ac57890638d7b7b060de6123722c9a0 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d913e6bf799a4d203fd3283a94bbc8719f2dc76f6c2590f61f27930e05e647f6 +size 678820394 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..8eae505bad97cd6d8b5c7bb96af68ace36748321 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63db52658097de725c9fbd6151f6760c0f80197d455e7748d91aa602ae104493 +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..9291b52759fc7e21fc736d1b3bfb59f484a2c4a5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f3e2a962f3bff927d704c7f4458064f9843328b67a6d903ddb7fa9bdab7412e +size 678820394 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..dfa24abb48eb66204ad0af1989ed9ef9d5c80922 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c5fa8bcfdad615f4b992f0f78cbaaaed8d7fa7a74282da5f0fc5cb2e0de5cd4 +size 678820394 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/6_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..bcf896942b0c549770c78feea57b9f4136124aaa --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 0.46593788266181946, + "learning_rate": 2e-05, + "loss": 1.4212, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 2.7293293476104736, + "learning_rate": 2e-05, + "loss": 1.5405, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 9.2821683883667, + "learning_rate": 2e-05, + "loss": 1.1381, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 1.7584102153778076, + "learning_rate": 2e-05, + "loss": 0.9004, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 1.6535090208053589, + "learning_rate": 2e-05, + "loss": 0.9172, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 7.006019592285156, + "learning_rate": 2e-05, + "loss": 2.5627, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 2.5369937419891357, + "learning_rate": 2e-05, + "loss": 3.3069, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 1.6373969316482544, + "learning_rate": 2e-05, + "loss": 2.1227, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 16.877363204956055, + "learning_rate": 2e-05, + "loss": 1.6699, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 2.3537302017211914, + "learning_rate": 2e-05, + "loss": 0.7411, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 4.379072189331055, + "learning_rate": 2e-05, + "loss": 1.4951, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 4.6082611083984375, + "learning_rate": 2e-05, + "loss": 1.4893, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 2.6327953338623047, + "learning_rate": 2e-05, + "loss": 1.0142, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 17.227697372436523, + "learning_rate": 2e-05, + "loss": 1.7676, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 5.5618977546691895, + "learning_rate": 2e-05, + "loss": 1.0957, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 2.1117804050445557, + "learning_rate": 2e-05, + "loss": 1.6618, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.9826680421829224, + "learning_rate": 2e-05, + "loss": 0.7603, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 3.9996843338012695, + "learning_rate": 2e-05, + "loss": 0.623, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 5.085465908050537, + "learning_rate": 2e-05, + "loss": 0.6493, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 14.83130931854248, + "learning_rate": 2e-05, + "loss": 0.8613, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 2.8493003845214844, + "learning_rate": 2e-05, + "loss": 1.5862, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 2.908425807952881, + "learning_rate": 2e-05, + "loss": 1.3001, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 5.74837589263916, + "learning_rate": 2e-05, + "loss": 1.001, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 5.138143539428711, + "learning_rate": 2e-05, + "loss": 1.0682, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 1.2600206136703491, + "learning_rate": 2e-05, + "loss": 0.5255, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 5.478726863861084, + "learning_rate": 2e-05, + "loss": 2.6286, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 3.6804332733154297, + "learning_rate": 2e-05, + "loss": 1.5683, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 6.222417831420898, + "learning_rate": 2e-05, + "loss": 1.5791, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 1.1432355642318726, + "learning_rate": 2e-05, + "loss": 1.4672, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 3.978116512298584, + "learning_rate": 2e-05, + "loss": 1.513, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 9.291589736938477, + "learning_rate": 2e-05, + "loss": 2.2455, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 3.612612009048462, + "learning_rate": 2e-05, + "loss": 0.331, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 1.645302176475525, + "learning_rate": 2e-05, + "loss": 1.7037, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 3.346055746078491, + "learning_rate": 2e-05, + "loss": 2.1318, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 5.71278715133667, + "learning_rate": 2e-05, + "loss": 1.213, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 9.712203025817871, + "learning_rate": 2e-05, + "loss": 1.7376, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 3.5931832790374756, + "learning_rate": 2e-05, + "loss": 2.1017, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 6.286159515380859, + "learning_rate": 2e-05, + "loss": 1.6007, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 7.470596790313721, + "learning_rate": 2e-05, + "loss": 2.9712, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 2.175534725189209, + "learning_rate": 2e-05, + "loss": 0.4087, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 1.9663554430007935, + "learning_rate": 2e-05, + "loss": 1.4516, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 10.726587295532227, + "learning_rate": 2e-05, + "loss": 2.2267, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 6.728185653686523, + "learning_rate": 2e-05, + "loss": 3.2496, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 8.222474098205566, + "learning_rate": 2e-05, + "loss": 1.3167, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 2.9531803131103516, + "learning_rate": 2e-05, + "loss": 2.2639, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 1.9350966215133667, + "learning_rate": 2e-05, + "loss": 2.1261, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 8.211071014404297, + "learning_rate": 2e-05, + "loss": 2.5981, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 1.3526861667633057, + "learning_rate": 2e-05, + "loss": 0.4025, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 0.9626100063323975, + "learning_rate": 2e-05, + "loss": 0.4061, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 5.2118635177612305, + "learning_rate": 2e-05, + "loss": 1.0989, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 1.6219733070184448e+16, + "train_loss": 1.5112138652801514, + "train_runtime": 263.0293, + "train_samples_per_second": 1.521, + "train_steps_per_second": 0.38 + } + ], + "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": 1.6219733070184448e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..078404f3a81653bb130e2e3b57f47868086a67ff --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ea5e7c6f5d94a7f9dcb52295e11b0e403c9bc42a672caf82b0e228cf841ec3e +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..8e040a69bab5dc11813a6ebee93264d103fb792c --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fdc6b57cff923a83a2f9a88be75ce08d78ff197a7904cd7218e62c4ca0ecfe8 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..dfc20b8999fe879d8b9d42f6fe960e634cad5d80 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:616f080ab4de685366e913becb48115409af0aaa60558c7a0b3624b83c0f3327 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..a80dd067be4cc2da59d420a50a156c81ccd990e1 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4df316df49a033a5e096f04615264d2d14d0fbab959908bb21fb9d1e0f3d29d +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..9b25fd7a164a8bfe7f4b974d7d4b10bef3caed9c --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:468b3effaf1452f863566ae5579cceaefd26fc81ca29fe6a0d5e0c8caffbc442 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..d343bde2b624eb1c63749e5bea087a0eb8e0deb5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f4a4593e82c8a515fe80d9d011ae01ef6fa484cc556018d83b38084701593b3 +size 395787774 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..62b67caa663dcf9c08650ba06c5df5020ec880f9 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc6c221929bf5c6c274e0f464f6919eb47fa020a6c64e31a0481580cde06e3fb +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..6f69df6fe0058f0b7f91a3c0dc9075c9ba509fe5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93669b357803d75bd75a30bebc60e5dc118beea2b43259a52bb691e76d37dee6 +size 395786922 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/7_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..6429a11d6e8b9455ea0cfa96c241f7a503ec3acf --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 12.068602561950684, + "learning_rate": 2e-05, + "loss": 1.7776, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 6.531093120574951, + "learning_rate": 2e-05, + "loss": 1.8584, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 2.197633981704712, + "learning_rate": 2e-05, + "loss": 1.813, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 7.011760234832764, + "learning_rate": 2e-05, + "loss": 1.677, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 3.8723106384277344, + "learning_rate": 2e-05, + "loss": 1.1979, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 4.476450443267822, + "learning_rate": 2e-05, + "loss": 1.1753, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 6.934531211853027, + "learning_rate": 2e-05, + "loss": 1.3873, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 3.920132875442505, + "learning_rate": 2e-05, + "loss": 0.7201, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 4.918713092803955, + "learning_rate": 2e-05, + "loss": 1.3185, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 5.342037677764893, + "learning_rate": 2e-05, + "loss": 1.547, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 4.474369049072266, + "learning_rate": 2e-05, + "loss": 1.6316, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 2.9470338821411133, + "learning_rate": 2e-05, + "loss": 0.5776, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 5.321770668029785, + "learning_rate": 2e-05, + "loss": 0.9258, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 7.642664432525635, + "learning_rate": 2e-05, + "loss": 1.9107, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 5.791833400726318, + "learning_rate": 2e-05, + "loss": 1.1289, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 3.643873691558838, + "learning_rate": 2e-05, + "loss": 1.1691, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 6.564858913421631, + "learning_rate": 2e-05, + "loss": 1.112, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 6.323983669281006, + "learning_rate": 2e-05, + "loss": 1.3032, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 6.3836588859558105, + "learning_rate": 2e-05, + "loss": 0.8167, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 3.984720468521118, + "learning_rate": 2e-05, + "loss": 0.6488, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 3.169071674346924, + "learning_rate": 2e-05, + "loss": 0.6885, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 3.60630202293396, + "learning_rate": 2e-05, + "loss": 1.5268, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 6.369883060455322, + "learning_rate": 2e-05, + "loss": 0.9459, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 3.9025063514709473, + "learning_rate": 2e-05, + "loss": 1.2159, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 5.445680141448975, + "learning_rate": 2e-05, + "loss": 1.2092, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 7.2683587074279785, + "learning_rate": 2e-05, + "loss": 2.3108, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 5.6280741691589355, + "learning_rate": 2e-05, + "loss": 1.6024, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 10.328232765197754, + "learning_rate": 2e-05, + "loss": 1.7271, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 5.541163921356201, + "learning_rate": 2e-05, + "loss": 1.8269, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 4.968420028686523, + "learning_rate": 2e-05, + "loss": 0.8484, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 4.030519485473633, + "learning_rate": 2e-05, + "loss": 0.9026, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 7.375129699707031, + "learning_rate": 2e-05, + "loss": 1.0315, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 7.278647422790527, + "learning_rate": 2e-05, + "loss": 1.9075, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 8.468052864074707, + "learning_rate": 2e-05, + "loss": 1.9799, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 8.026609420776367, + "learning_rate": 2e-05, + "loss": 1.3111, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 7.3639116287231445, + "learning_rate": 2e-05, + "loss": 1.4198, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 8.574053764343262, + "learning_rate": 2e-05, + "loss": 2.4585, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 5.82910680770874, + "learning_rate": 2e-05, + "loss": 0.7334, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 8.020891189575195, + "learning_rate": 2e-05, + "loss": 2.5793, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 5.482921600341797, + "learning_rate": 2e-05, + "loss": 1.5728, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 2.912376642227173, + "learning_rate": 2e-05, + "loss": 1.0129, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 6.266790390014648, + "learning_rate": 2e-05, + "loss": 0.9668, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 9.593369483947754, + "learning_rate": 2e-05, + "loss": 1.4147, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 5.849009990692139, + "learning_rate": 2e-05, + "loss": 1.8889, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 3.4311914443969727, + "learning_rate": 2e-05, + "loss": 0.9566, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 3.691415786743164, + "learning_rate": 2e-05, + "loss": 2.0316, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 7.05644416809082, + "learning_rate": 2e-05, + "loss": 1.3933, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 5.451862812042236, + "learning_rate": 2e-05, + "loss": 0.8934, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 10.2772216796875, + "learning_rate": 2e-05, + "loss": 1.7324, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 5.86804723739624, + "learning_rate": 2e-05, + "loss": 0.9668, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 8549139625279488.0, + "train_loss": 1.3750475025177002, + "train_runtime": 222.9593, + "train_samples_per_second": 1.794, + "train_steps_per_second": 0.449 + } + ], + "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": 8549139625279488.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..894cb4842387fe5a082ec79bdaa5b5713e868a9b --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5da8dbd437d4e9ae452334e1d9e45a12c7b7b0fee3480cad7ad16eba496f2aca +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..d9273cf201eec1fd93b089960411ff88700d6f18 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9230388c13f0563d5cb20b8b75a064e1eb011249bc402b1228256288f2ad4988 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..229a6311a779ca2ba76cd954bf6cba802745727d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a68725dc197c37dda0e0bd6b8c96e36627834532722f8eb16e64868410f42925 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..41fe62d2aa9ac8524a2c89d39c1b78f1a6014496 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d257af8001a27d1025dc58b31dd5eac0a91d7f027d53aae0790552231a09017 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..15da1953fc048611e0c16cc91341e8356834cc29 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7aa11e4f5dd3383b4f1cd2932f284886f4465e864f4653e97aa46ef99fca054 +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..ed8991cd156bef82baa4fcf2228cc6c22196a061 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84d8e8470e731e6c362299326e7de72703bd3fb51470f77fb35ff1438029b6e3 +size 184221358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..b3dad94373898075308dfd28a999f5512dd3c753 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b7c20c34da48af9ad4eb4663d44f9d67a70d387d723efd0e76cf221e5d697dd +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e6376f58b2e478feb8fe8550c5bb57bee1a58a3 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8aca3fb31e32a0796b33e6a334433ac69bec0fe9bd34c1c5958d20f8f9b4b513 +size 184220842 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/8_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..ed4176675165980ceb3a9a003f172991333e9d94 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_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": 1.7061518430709839, + "learning_rate": 2e-05, + "loss": 1.1925, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 6.53925085067749, + "learning_rate": 2e-05, + "loss": 1.7138, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 6.937854766845703, + "learning_rate": 2e-05, + "loss": 1.4859, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 8.083235740661621, + "learning_rate": 2e-05, + "loss": 3.1386, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 9.314245223999023, + "learning_rate": 2e-05, + "loss": 3.4677, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 7.069417953491211, + "learning_rate": 2e-05, + "loss": 2.2931, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 4.106843948364258, + "learning_rate": 2e-05, + "loss": 1.1953, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 7.066917896270752, + "learning_rate": 2e-05, + "loss": 1.1264, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 7.475583076477051, + "learning_rate": 2e-05, + "loss": 1.6035, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 11.9594144821167, + "learning_rate": 2e-05, + "loss": 2.489, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 9.436518669128418, + "learning_rate": 2e-05, + "loss": 2.0764, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 9.553932189941406, + "learning_rate": 2e-05, + "loss": 2.0292, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 8.222797393798828, + "learning_rate": 2e-05, + "loss": 2.4272, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 6.939860820770264, + "learning_rate": 2e-05, + "loss": 2.6952, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 4.106882572174072, + "learning_rate": 2e-05, + "loss": 1.3535, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 9.197134017944336, + "learning_rate": 2e-05, + "loss": 2.3694, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 10.052316665649414, + "learning_rate": 2e-05, + "loss": 2.0721, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 12.314714431762695, + "learning_rate": 2e-05, + "loss": 2.4687, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 2.2270240783691406, + "learning_rate": 2e-05, + "loss": 0.6964, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 11.931086540222168, + "learning_rate": 2e-05, + "loss": 3.3661, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 6.998178005218506, + "learning_rate": 2e-05, + "loss": 1.3806, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 10.16988754272461, + "learning_rate": 2e-05, + "loss": 3.1914, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 7.041690349578857, + "learning_rate": 2e-05, + "loss": 0.6591, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 4.950750827789307, + "learning_rate": 2e-05, + "loss": 1.2721, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 5.443760871887207, + "learning_rate": 2e-05, + "loss": 2.8281, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 7.689634323120117, + "learning_rate": 2e-05, + "loss": 1.9726, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 3.3720381259918213, + "learning_rate": 2e-05, + "loss": 1.9971, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 9.029294967651367, + "learning_rate": 2e-05, + "loss": 2.8538, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 6.2818708419799805, + "learning_rate": 2e-05, + "loss": 3.6545, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 11.004154205322266, + "learning_rate": 2e-05, + "loss": 3.9829, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 18.696840286254883, + "learning_rate": 2e-05, + "loss": 2.3689, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 6.145025730133057, + "learning_rate": 2e-05, + "loss": 2.4314, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 11.131457328796387, + "learning_rate": 2e-05, + "loss": 2.1187, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 11.315707206726074, + "learning_rate": 2e-05, + "loss": 2.5487, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 16.595457077026367, + "learning_rate": 2e-05, + "loss": 2.5804, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 10.90677547454834, + "learning_rate": 2e-05, + "loss": 2.66, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 6.262564659118652, + "learning_rate": 2e-05, + "loss": 1.5396, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 24.616371154785156, + "learning_rate": 2e-05, + "loss": 5.588, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 13.945603370666504, + "learning_rate": 2e-05, + "loss": 1.9208, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 4.174859523773193, + "learning_rate": 2e-05, + "loss": 0.9455, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 7.081887245178223, + "learning_rate": 2e-05, + "loss": 1.3299, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 2.1284427642822266, + "learning_rate": 2e-05, + "loss": 1.8537, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 8.366132736206055, + "learning_rate": 2e-05, + "loss": 5.4844, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 8.645503044128418, + "learning_rate": 2e-05, + "loss": 2.5933, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 15.805315971374512, + "learning_rate": 2e-05, + "loss": 2.2525, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 8.33022403717041, + "learning_rate": 2e-05, + "loss": 1.72, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 2.566896915435791, + "learning_rate": 2e-05, + "loss": 0.2577, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 5.386092185974121, + "learning_rate": 2e-05, + "loss": 1.5405, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 5.493599891662598, + "learning_rate": 2e-05, + "loss": 0.8738, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 13.045679092407227, + "learning_rate": 2e-05, + "loss": 1.6061, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 2065907056640000.0, + "train_loss": 2.185313892364502, + "train_runtime": 104.3275, + "train_samples_per_second": 3.834, + "train_steps_per_second": 0.959 + } + ], + "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": 2065907056640000.0, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +} diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round10.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round10.pth new file mode 100644 index 0000000000000000000000000000000000000000..0794f04b55f0908703fe17b35a400b5189246a83 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round10.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eacaf7c48fd1283a0097723483980fabc47313a1f8df0a3dad936085b00f11b2 +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round12.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round12.pth new file mode 100644 index 0000000000000000000000000000000000000000..7fe614920b6093ef1e8834254a5e5c0bbd148f43 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round12.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb56e1c6563bc83459d67ee93f5bdba3acfc02b47c9a7923c4300b41e95724af +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round15.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round15.pth new file mode 100644 index 0000000000000000000000000000000000000000..6edb84b5198a216101c264b1c65f33f1a8fddeaf --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round15.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cd0c318c04003873e0140efcf789dbb33d37766fd995422d82e31cbef6ee13a +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round17.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round17.pth new file mode 100644 index 0000000000000000000000000000000000000000..02ffe9d48dbf903489aae7679dc33f969f058670 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round17.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df6dbb1d6b45fd53efb324685d92358dd673291fb89bd35578dfe7175523d550 +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round2.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round2.pth new file mode 100644 index 0000000000000000000000000000000000000000..f2808e2c29844db8d410b5fbd643b3a4f641864d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:270afc06a0eb8e6cc58247881d100335d2a966b9b866c40e2cae50edd2bbb8e2 +size 678820394 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round20.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round20.pth new file mode 100644 index 0000000000000000000000000000000000000000..4a24a61daeceade02c6cce1fab3831d7806c2369 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round20.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af9b707f9292c0dd997f7488bf98d0857881589e57fa84121abbc322a5baa86f +size 678821358 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round5.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round5.pth new file mode 100644 index 0000000000000000000000000000000000000000..7c8e65ac1367428ba5046ff0d0ebaf0e533eb02d --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b8d62dcff88965b1fe01562a687534e4c5828d9f40e7ce04e7be266df97031f +size 678820394 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round7.pth b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round7.pth new file mode 100644 index 0000000000000000000000000000000000000000..e66da9e683e6cb70ac1a342364e6134cf7e9a2a5 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_client_model_round7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:431e9197d18d21a7b2aeb3ce59075eda317899bc6a8dd8093dc10a72e2ca7a01 +size 678820394 diff --git a/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_trainer_state.json b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..18e2aaacc1cca4faf9fdf25867e9eee9f98503e4 --- /dev/null +++ b/client_states_fedMultipqfullfreeze_sft_bs4_saveoptim_lr2e-5_sc137_4tasks_5rounds_fixitr100_T0125_decay099/9_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": 2.6313834190368652, + "learning_rate": 2e-05, + "loss": 1.0492, + "step": 2 + }, + { + "epoch": 0.04, + "grad_norm": 4.533816337585449, + "learning_rate": 2e-05, + "loss": 1.9921, + "step": 4 + }, + { + "epoch": 0.06, + "grad_norm": 4.989090442657471, + "learning_rate": 2e-05, + "loss": 0.473, + "step": 6 + }, + { + "epoch": 0.08, + "grad_norm": 15.079896926879883, + "learning_rate": 2e-05, + "loss": 2.9076, + "step": 8 + }, + { + "epoch": 0.1, + "grad_norm": 4.404162883758545, + "learning_rate": 2e-05, + "loss": 1.7797, + "step": 10 + }, + { + "epoch": 0.12, + "grad_norm": 6.694067001342773, + "learning_rate": 2e-05, + "loss": 0.9664, + "step": 12 + }, + { + "epoch": 0.14, + "grad_norm": 6.5315842628479, + "learning_rate": 2e-05, + "loss": 2.0612, + "step": 14 + }, + { + "epoch": 0.16, + "grad_norm": 0.723440945148468, + "learning_rate": 2e-05, + "loss": 1.4464, + "step": 16 + }, + { + "epoch": 0.18, + "grad_norm": 3.0090744495391846, + "learning_rate": 2e-05, + "loss": 1.9595, + "step": 18 + }, + { + "epoch": 0.2, + "grad_norm": 9.052999496459961, + "learning_rate": 2e-05, + "loss": 1.4913, + "step": 20 + }, + { + "epoch": 0.22, + "grad_norm": 10.191319465637207, + "learning_rate": 2e-05, + "loss": 2.3721, + "step": 22 + }, + { + "epoch": 0.24, + "grad_norm": 2.751216173171997, + "learning_rate": 2e-05, + "loss": 1.0765, + "step": 24 + }, + { + "epoch": 0.26, + "grad_norm": 6.357051849365234, + "learning_rate": 2e-05, + "loss": 1.7153, + "step": 26 + }, + { + "epoch": 0.28, + "grad_norm": 1.1915141344070435, + "learning_rate": 2e-05, + "loss": 0.6944, + "step": 28 + }, + { + "epoch": 0.3, + "grad_norm": 5.416281700134277, + "learning_rate": 2e-05, + "loss": 0.9509, + "step": 30 + }, + { + "epoch": 0.32, + "grad_norm": 9.90926742553711, + "learning_rate": 2e-05, + "loss": 1.2579, + "step": 32 + }, + { + "epoch": 0.34, + "grad_norm": 1.401187777519226, + "learning_rate": 2e-05, + "loss": 0.5905, + "step": 34 + }, + { + "epoch": 0.36, + "grad_norm": 2.7381670475006104, + "learning_rate": 2e-05, + "loss": 1.3317, + "step": 36 + }, + { + "epoch": 0.38, + "grad_norm": 3.8321311473846436, + "learning_rate": 2e-05, + "loss": 0.7657, + "step": 38 + }, + { + "epoch": 0.4, + "grad_norm": 3.804307222366333, + "learning_rate": 2e-05, + "loss": 0.5641, + "step": 40 + }, + { + "epoch": 0.42, + "grad_norm": 2.954988718032837, + "learning_rate": 2e-05, + "loss": 1.1049, + "step": 42 + }, + { + "epoch": 0.44, + "grad_norm": 12.859837532043457, + "learning_rate": 2e-05, + "loss": 2.4993, + "step": 44 + }, + { + "epoch": 0.46, + "grad_norm": 1.3593528270721436, + "learning_rate": 2e-05, + "loss": 0.5819, + "step": 46 + }, + { + "epoch": 0.48, + "grad_norm": 9.18574333190918, + "learning_rate": 2e-05, + "loss": 1.477, + "step": 48 + }, + { + "epoch": 0.5, + "grad_norm": 7.759741306304932, + "learning_rate": 2e-05, + "loss": 1.751, + "step": 50 + }, + { + "epoch": 0.52, + "grad_norm": 2.3041605949401855, + "learning_rate": 2e-05, + "loss": 1.6484, + "step": 52 + }, + { + "epoch": 0.54, + "grad_norm": 2.937114953994751, + "learning_rate": 2e-05, + "loss": 1.3826, + "step": 54 + }, + { + "epoch": 0.56, + "grad_norm": 7.1777215003967285, + "learning_rate": 2e-05, + "loss": 0.7476, + "step": 56 + }, + { + "epoch": 0.58, + "grad_norm": 7.131524085998535, + "learning_rate": 2e-05, + "loss": 2.1071, + "step": 58 + }, + { + "epoch": 0.6, + "grad_norm": 2.245123863220215, + "learning_rate": 2e-05, + "loss": 1.225, + "step": 60 + }, + { + "epoch": 0.62, + "grad_norm": 2.359640121459961, + "learning_rate": 2e-05, + "loss": 1.9538, + "step": 62 + }, + { + "epoch": 0.64, + "grad_norm": 11.923921585083008, + "learning_rate": 2e-05, + "loss": 1.3044, + "step": 64 + }, + { + "epoch": 0.66, + "grad_norm": 11.412819862365723, + "learning_rate": 2e-05, + "loss": 1.2069, + "step": 66 + }, + { + "epoch": 0.68, + "grad_norm": 8.552916526794434, + "learning_rate": 2e-05, + "loss": 3.0746, + "step": 68 + }, + { + "epoch": 0.7, + "grad_norm": 6.0205230712890625, + "learning_rate": 2e-05, + "loss": 1.2432, + "step": 70 + }, + { + "epoch": 0.72, + "grad_norm": 4.929473400115967, + "learning_rate": 2e-05, + "loss": 2.3425, + "step": 72 + }, + { + "epoch": 0.74, + "grad_norm": 2.7426578998565674, + "learning_rate": 2e-05, + "loss": 0.534, + "step": 74 + }, + { + "epoch": 0.76, + "grad_norm": 4.7789692878723145, + "learning_rate": 2e-05, + "loss": 0.6954, + "step": 76 + }, + { + "epoch": 0.78, + "grad_norm": 4.892092704772949, + "learning_rate": 2e-05, + "loss": 0.9157, + "step": 78 + }, + { + "epoch": 0.8, + "grad_norm": 3.2699296474456787, + "learning_rate": 2e-05, + "loss": 0.7456, + "step": 80 + }, + { + "epoch": 0.82, + "grad_norm": 8.563464164733887, + "learning_rate": 2e-05, + "loss": 1.8777, + "step": 82 + }, + { + "epoch": 0.84, + "grad_norm": 2.663980007171631, + "learning_rate": 2e-05, + "loss": 0.8767, + "step": 84 + }, + { + "epoch": 0.86, + "grad_norm": 11.196078300476074, + "learning_rate": 2e-05, + "loss": 1.3835, + "step": 86 + }, + { + "epoch": 0.88, + "grad_norm": 4.945417404174805, + "learning_rate": 2e-05, + "loss": 1.3922, + "step": 88 + }, + { + "epoch": 0.9, + "grad_norm": 12.084806442260742, + "learning_rate": 2e-05, + "loss": 1.7619, + "step": 90 + }, + { + "epoch": 0.92, + "grad_norm": 8.386174201965332, + "learning_rate": 2e-05, + "loss": 1.9914, + "step": 92 + }, + { + "epoch": 0.94, + "grad_norm": 4.200559616088867, + "learning_rate": 2e-05, + "loss": 1.113, + "step": 94 + }, + { + "epoch": 0.96, + "grad_norm": 6.844775199890137, + "learning_rate": 2e-05, + "loss": 1.8205, + "step": 96 + }, + { + "epoch": 0.98, + "grad_norm": 7.386288642883301, + "learning_rate": 2e-05, + "loss": 0.6322, + "step": 98 + }, + { + "epoch": 1.0, + "grad_norm": 2.694425582885742, + "learning_rate": 2e-05, + "loss": 0.8945, + "step": 100 + }, + { + "epoch": 1.0, + "step": 100, + "total_flos": 1.5799240638857216e+16, + "train_loss": 1.3945933890342712, + "train_runtime": 258.724, + "train_samples_per_second": 1.546, + "train_steps_per_second": 0.387 + } + ], + "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": 1.5799240638857216e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}