Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round10.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round12.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round15.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round17.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round2.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round20.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round5.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round7.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_trainer_state.json +378 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round10.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round12.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round15.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round17.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round2.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round20.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round5.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round7.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_trainer_state.json +378 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round10.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round12.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round15.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round17.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round2.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round20.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round5.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round7.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_trainer_state.json +378 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round10.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round12.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round15.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round17.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round2.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round20.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round5.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round7.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_trainer_state.json +378 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round10.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round12.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round15.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round17.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round2.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round20.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round5.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round7.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_trainer_state.json +378 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round10.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round12.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round15.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round17.pth +3 -0
- client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round2.pth +3 -0
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round10.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28ef8239d82162ec0cd709c417e174890fa11c1a4d3bb24871eb9c35db756c95
|
| 3 |
+
size 368443438
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round12.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a91dbfbd03b0eda5eea78822e04cc2b1e3b5890093975eea94cc34f926140df
|
| 3 |
+
size 368443438
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round15.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f2a82b2beb0af43f3895ebcf348c413e0ecd5ed806c5831de2d4b891e4f46e1
|
| 3 |
+
size 368443438
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round17.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19a985f1833e783d4938f0a9ce7528e24a1d41ae07972a5bf344eb39d15eb2a0
|
| 3 |
+
size 368443438
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f7af35ffe4dc0fba5de51bfd5ad8b4fb9fe896b1b6c84c1e61b575b0c704b1a
|
| 3 |
+
size 368442474
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round20.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73b3d0a1d38c13e810a623753d551af1a679f168805592f2407c3b4171bee819
|
| 3 |
+
size 368443438
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73bff6b90d411e41eb15f92cbe7e80d64cc0c652de9e204bbdcab4facc759ccb
|
| 3 |
+
size 368442474
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_client_model_round7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a78ae0540308ef5b38f30f678f176dc07b31ac92282a3f1061b27960904527b6
|
| 3 |
+
size 368442474
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/0_trainer_state.json
ADDED
|
@@ -0,0 +1,378 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 97,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.020618556701030927,
|
| 13 |
+
"grad_norm": 1.1101641654968262,
|
| 14 |
+
"learning_rate": 2e-05,
|
| 15 |
+
"loss": 0.8945,
|
| 16 |
+
"step": 2
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.041237113402061855,
|
| 20 |
+
"grad_norm": 1.6314992904663086,
|
| 21 |
+
"learning_rate": 2e-05,
|
| 22 |
+
"loss": 1.4103,
|
| 23 |
+
"step": 4
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.061855670103092786,
|
| 27 |
+
"grad_norm": 2.4410200119018555,
|
| 28 |
+
"learning_rate": 2e-05,
|
| 29 |
+
"loss": 1.3191,
|
| 30 |
+
"step": 6
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08247422680412371,
|
| 34 |
+
"grad_norm": 3.6628243923187256,
|
| 35 |
+
"learning_rate": 2e-05,
|
| 36 |
+
"loss": 1.968,
|
| 37 |
+
"step": 8
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10309278350515463,
|
| 41 |
+
"grad_norm": 6.844225883483887,
|
| 42 |
+
"learning_rate": 2e-05,
|
| 43 |
+
"loss": 1.5729,
|
| 44 |
+
"step": 10
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12371134020618557,
|
| 48 |
+
"grad_norm": 0.7698001265525818,
|
| 49 |
+
"learning_rate": 2e-05,
|
| 50 |
+
"loss": 0.5719,
|
| 51 |
+
"step": 12
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14432989690721648,
|
| 55 |
+
"grad_norm": 0.5793606042861938,
|
| 56 |
+
"learning_rate": 2e-05,
|
| 57 |
+
"loss": 1.5151,
|
| 58 |
+
"step": 14
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.16494845360824742,
|
| 62 |
+
"grad_norm": 1.8425798416137695,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 1.0999,
|
| 65 |
+
"step": 16
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.18556701030927836,
|
| 69 |
+
"grad_norm": 1.134320855140686,
|
| 70 |
+
"learning_rate": 2e-05,
|
| 71 |
+
"loss": 0.9422,
|
| 72 |
+
"step": 18
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.20618556701030927,
|
| 76 |
+
"grad_norm": 1.168424367904663,
|
| 77 |
+
"learning_rate": 2e-05,
|
| 78 |
+
"loss": 0.7192,
|
| 79 |
+
"step": 20
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2268041237113402,
|
| 83 |
+
"grad_norm": 3.056905746459961,
|
| 84 |
+
"learning_rate": 2e-05,
|
| 85 |
+
"loss": 0.6074,
|
| 86 |
+
"step": 22
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.24742268041237114,
|
| 90 |
+
"grad_norm": 0.6598642468452454,
|
| 91 |
+
"learning_rate": 2e-05,
|
| 92 |
+
"loss": 0.346,
|
| 93 |
+
"step": 24
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.26804123711340205,
|
| 97 |
+
"grad_norm": 1.0676809549331665,
|
| 98 |
+
"learning_rate": 2e-05,
|
| 99 |
+
"loss": 1.1766,
|
| 100 |
+
"step": 26
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.28865979381443296,
|
| 104 |
+
"grad_norm": 1.2854700088500977,
|
| 105 |
+
"learning_rate": 2e-05,
|
| 106 |
+
"loss": 1.3399,
|
| 107 |
+
"step": 28
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.30927835051546393,
|
| 111 |
+
"grad_norm": 2.3572678565979004,
|
| 112 |
+
"learning_rate": 2e-05,
|
| 113 |
+
"loss": 1.1957,
|
| 114 |
+
"step": 30
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32989690721649484,
|
| 118 |
+
"grad_norm": 1.2973973751068115,
|
| 119 |
+
"learning_rate": 2e-05,
|
| 120 |
+
"loss": 1.8289,
|
| 121 |
+
"step": 32
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.35051546391752575,
|
| 125 |
+
"grad_norm": 0.6981891989707947,
|
| 126 |
+
"learning_rate": 2e-05,
|
| 127 |
+
"loss": 0.7445,
|
| 128 |
+
"step": 34
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3711340206185567,
|
| 132 |
+
"grad_norm": 0.41950735449790955,
|
| 133 |
+
"learning_rate": 2e-05,
|
| 134 |
+
"loss": 0.8945,
|
| 135 |
+
"step": 36
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3917525773195876,
|
| 139 |
+
"grad_norm": 2.4804954528808594,
|
| 140 |
+
"learning_rate": 2e-05,
|
| 141 |
+
"loss": 0.8976,
|
| 142 |
+
"step": 38
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.41237113402061853,
|
| 146 |
+
"grad_norm": 0.989696741104126,
|
| 147 |
+
"learning_rate": 2e-05,
|
| 148 |
+
"loss": 0.8965,
|
| 149 |
+
"step": 40
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4329896907216495,
|
| 153 |
+
"grad_norm": 0.7618101239204407,
|
| 154 |
+
"learning_rate": 2e-05,
|
| 155 |
+
"loss": 1.5496,
|
| 156 |
+
"step": 42
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.4536082474226804,
|
| 160 |
+
"grad_norm": 3.34183406829834,
|
| 161 |
+
"learning_rate": 2e-05,
|
| 162 |
+
"loss": 0.5556,
|
| 163 |
+
"step": 44
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.4742268041237113,
|
| 167 |
+
"grad_norm": 1.5414451360702515,
|
| 168 |
+
"learning_rate": 2e-05,
|
| 169 |
+
"loss": 0.9727,
|
| 170 |
+
"step": 46
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.4948453608247423,
|
| 174 |
+
"grad_norm": 3.042322874069214,
|
| 175 |
+
"learning_rate": 2e-05,
|
| 176 |
+
"loss": 0.8753,
|
| 177 |
+
"step": 48
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5154639175257731,
|
| 181 |
+
"grad_norm": 2.1184558868408203,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 1.381,
|
| 184 |
+
"step": 50
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5360824742268041,
|
| 188 |
+
"grad_norm": 0.3003075420856476,
|
| 189 |
+
"learning_rate": 2e-05,
|
| 190 |
+
"loss": 0.2195,
|
| 191 |
+
"step": 52
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5567010309278351,
|
| 195 |
+
"grad_norm": 0.7136024236679077,
|
| 196 |
+
"learning_rate": 2e-05,
|
| 197 |
+
"loss": 1.4137,
|
| 198 |
+
"step": 54
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5773195876288659,
|
| 202 |
+
"grad_norm": 2.5711076259613037,
|
| 203 |
+
"learning_rate": 2e-05,
|
| 204 |
+
"loss": 1.3358,
|
| 205 |
+
"step": 56
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5979381443298969,
|
| 209 |
+
"grad_norm": 3.2242660522460938,
|
| 210 |
+
"learning_rate": 2e-05,
|
| 211 |
+
"loss": 0.6124,
|
| 212 |
+
"step": 58
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6185567010309279,
|
| 216 |
+
"grad_norm": 1.4743932485580444,
|
| 217 |
+
"learning_rate": 2e-05,
|
| 218 |
+
"loss": 0.8501,
|
| 219 |
+
"step": 60
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6391752577319587,
|
| 223 |
+
"grad_norm": 1.8894503116607666,
|
| 224 |
+
"learning_rate": 2e-05,
|
| 225 |
+
"loss": 0.7892,
|
| 226 |
+
"step": 62
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6597938144329897,
|
| 230 |
+
"grad_norm": 4.588108539581299,
|
| 231 |
+
"learning_rate": 2e-05,
|
| 232 |
+
"loss": 1.0005,
|
| 233 |
+
"step": 64
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6804123711340206,
|
| 237 |
+
"grad_norm": 0.5758951306343079,
|
| 238 |
+
"learning_rate": 2e-05,
|
| 239 |
+
"loss": 0.1539,
|
| 240 |
+
"step": 66
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.7010309278350515,
|
| 244 |
+
"grad_norm": 3.7251107692718506,
|
| 245 |
+
"learning_rate": 2e-05,
|
| 246 |
+
"loss": 0.3073,
|
| 247 |
+
"step": 68
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7216494845360825,
|
| 251 |
+
"grad_norm": 2.5955007076263428,
|
| 252 |
+
"learning_rate": 2e-05,
|
| 253 |
+
"loss": 0.7984,
|
| 254 |
+
"step": 70
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7422680412371134,
|
| 258 |
+
"grad_norm": 2.8188343048095703,
|
| 259 |
+
"learning_rate": 2e-05,
|
| 260 |
+
"loss": 0.8585,
|
| 261 |
+
"step": 72
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7628865979381443,
|
| 265 |
+
"grad_norm": 4.2932634353637695,
|
| 266 |
+
"learning_rate": 2e-05,
|
| 267 |
+
"loss": 0.7,
|
| 268 |
+
"step": 74
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7835051546391752,
|
| 272 |
+
"grad_norm": 2.4785430431365967,
|
| 273 |
+
"learning_rate": 2e-05,
|
| 274 |
+
"loss": 1.4911,
|
| 275 |
+
"step": 76
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8041237113402062,
|
| 279 |
+
"grad_norm": 3.244961977005005,
|
| 280 |
+
"learning_rate": 2e-05,
|
| 281 |
+
"loss": 0.9058,
|
| 282 |
+
"step": 78
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8247422680412371,
|
| 286 |
+
"grad_norm": 3.3959240913391113,
|
| 287 |
+
"learning_rate": 2e-05,
|
| 288 |
+
"loss": 1.928,
|
| 289 |
+
"step": 80
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.845360824742268,
|
| 293 |
+
"grad_norm": 0.36406227946281433,
|
| 294 |
+
"learning_rate": 2e-05,
|
| 295 |
+
"loss": 1.2079,
|
| 296 |
+
"step": 82
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.865979381443299,
|
| 300 |
+
"grad_norm": 1.649201512336731,
|
| 301 |
+
"learning_rate": 2e-05,
|
| 302 |
+
"loss": 1.3682,
|
| 303 |
+
"step": 84
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8865979381443299,
|
| 307 |
+
"grad_norm": 1.951743721961975,
|
| 308 |
+
"learning_rate": 2e-05,
|
| 309 |
+
"loss": 0.9706,
|
| 310 |
+
"step": 86
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9072164948453608,
|
| 314 |
+
"grad_norm": 1.6012216806411743,
|
| 315 |
+
"learning_rate": 2e-05,
|
| 316 |
+
"loss": 0.956,
|
| 317 |
+
"step": 88
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9278350515463918,
|
| 321 |
+
"grad_norm": 1.059224247932434,
|
| 322 |
+
"learning_rate": 2e-05,
|
| 323 |
+
"loss": 0.7971,
|
| 324 |
+
"step": 90
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.9484536082474226,
|
| 328 |
+
"grad_norm": 1.2695350646972656,
|
| 329 |
+
"learning_rate": 2e-05,
|
| 330 |
+
"loss": 1.2903,
|
| 331 |
+
"step": 92
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.9690721649484536,
|
| 335 |
+
"grad_norm": 0.9126003980636597,
|
| 336 |
+
"learning_rate": 2e-05,
|
| 337 |
+
"loss": 0.486,
|
| 338 |
+
"step": 94
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9896907216494846,
|
| 342 |
+
"grad_norm": 0.9548375606536865,
|
| 343 |
+
"learning_rate": 2e-05,
|
| 344 |
+
"loss": 0.4924,
|
| 345 |
+
"step": 96
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.0,
|
| 349 |
+
"step": 97,
|
| 350 |
+
"total_flos": 5653713448337408.0,
|
| 351 |
+
"train_loss": 1.0004102569265463,
|
| 352 |
+
"train_runtime": 222.4287,
|
| 353 |
+
"train_samples_per_second": 1.744,
|
| 354 |
+
"train_steps_per_second": 0.436
|
| 355 |
+
}
|
| 356 |
+
],
|
| 357 |
+
"logging_steps": 2,
|
| 358 |
+
"max_steps": 97,
|
| 359 |
+
"num_input_tokens_seen": 0,
|
| 360 |
+
"num_train_epochs": 1,
|
| 361 |
+
"save_steps": 500,
|
| 362 |
+
"stateful_callbacks": {
|
| 363 |
+
"TrainerControl": {
|
| 364 |
+
"args": {
|
| 365 |
+
"should_epoch_stop": false,
|
| 366 |
+
"should_evaluate": false,
|
| 367 |
+
"should_log": false,
|
| 368 |
+
"should_save": false,
|
| 369 |
+
"should_training_stop": false
|
| 370 |
+
},
|
| 371 |
+
"attributes": {}
|
| 372 |
+
}
|
| 373 |
+
},
|
| 374 |
+
"total_flos": 5653713448337408.0,
|
| 375 |
+
"train_batch_size": 1,
|
| 376 |
+
"trial_name": null,
|
| 377 |
+
"trial_params": null
|
| 378 |
+
}
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round10.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac50ab133a7a5aa9dfea61e6d8ed7653bb6f285eb13975c45759804e1c5a0113
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round12.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66dd8680ac517ba6ca8d2ff3c8a3cbad6fe0cccec680ad711d030f9dd2c6554f
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round15.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:268c8d523348ff94235dca656d4225fe4d7311eda06f2dd672cb2995a74cf1ad
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round17.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5ee153c72366a156c658c58c08fb31ae4d89aaf7cf231b1a3fc2cf80098a182
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4a9d2ff90a509ed280525e7783832b83f0e780f8cc85032615ccff649e9190e
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round20.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c82a6b29ae63539e855ee08429abf75f2eb504c822f65b9cd7a456e87d8af94
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8ecbe4f845d20983d9fa3c345743a695c773b6f8352ba9e0b9e3fd4efcd378d
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_client_model_round7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21522413a5083211437c3a7d2e7d2f40176b80d21234179e98a978a40f816dac
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/1_trainer_state.json
ADDED
|
@@ -0,0 +1,378 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 97,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.020618556701030927,
|
| 13 |
+
"grad_norm": 5.094399929046631,
|
| 14 |
+
"learning_rate": 2e-05,
|
| 15 |
+
"loss": 1.5978,
|
| 16 |
+
"step": 2
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.041237113402061855,
|
| 20 |
+
"grad_norm": 2.2240917682647705,
|
| 21 |
+
"learning_rate": 2e-05,
|
| 22 |
+
"loss": 2.3088,
|
| 23 |
+
"step": 4
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.061855670103092786,
|
| 27 |
+
"grad_norm": 1.5078622102737427,
|
| 28 |
+
"learning_rate": 2e-05,
|
| 29 |
+
"loss": 0.9092,
|
| 30 |
+
"step": 6
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08247422680412371,
|
| 34 |
+
"grad_norm": 0.8912699818611145,
|
| 35 |
+
"learning_rate": 2e-05,
|
| 36 |
+
"loss": 0.6373,
|
| 37 |
+
"step": 8
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10309278350515463,
|
| 41 |
+
"grad_norm": 0.416252076625824,
|
| 42 |
+
"learning_rate": 2e-05,
|
| 43 |
+
"loss": 0.8186,
|
| 44 |
+
"step": 10
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12371134020618557,
|
| 48 |
+
"grad_norm": 0.8248510360717773,
|
| 49 |
+
"learning_rate": 2e-05,
|
| 50 |
+
"loss": 1.0876,
|
| 51 |
+
"step": 12
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14432989690721648,
|
| 55 |
+
"grad_norm": 0.9105907082557678,
|
| 56 |
+
"learning_rate": 2e-05,
|
| 57 |
+
"loss": 0.9326,
|
| 58 |
+
"step": 14
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.16494845360824742,
|
| 62 |
+
"grad_norm": 0.7425227761268616,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 1.1207,
|
| 65 |
+
"step": 16
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.18556701030927836,
|
| 69 |
+
"grad_norm": 1.2341773509979248,
|
| 70 |
+
"learning_rate": 2e-05,
|
| 71 |
+
"loss": 1.0813,
|
| 72 |
+
"step": 18
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.20618556701030927,
|
| 76 |
+
"grad_norm": 1.2988817691802979,
|
| 77 |
+
"learning_rate": 2e-05,
|
| 78 |
+
"loss": 1.5784,
|
| 79 |
+
"step": 20
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2268041237113402,
|
| 83 |
+
"grad_norm": 1.1802982091903687,
|
| 84 |
+
"learning_rate": 2e-05,
|
| 85 |
+
"loss": 1.1858,
|
| 86 |
+
"step": 22
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.24742268041237114,
|
| 90 |
+
"grad_norm": 0.40825000405311584,
|
| 91 |
+
"learning_rate": 2e-05,
|
| 92 |
+
"loss": 1.0679,
|
| 93 |
+
"step": 24
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.26804123711340205,
|
| 97 |
+
"grad_norm": 1.1907871961593628,
|
| 98 |
+
"learning_rate": 2e-05,
|
| 99 |
+
"loss": 0.9465,
|
| 100 |
+
"step": 26
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.28865979381443296,
|
| 104 |
+
"grad_norm": 2.7417049407958984,
|
| 105 |
+
"learning_rate": 2e-05,
|
| 106 |
+
"loss": 1.6913,
|
| 107 |
+
"step": 28
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.30927835051546393,
|
| 111 |
+
"grad_norm": 1.4188278913497925,
|
| 112 |
+
"learning_rate": 2e-05,
|
| 113 |
+
"loss": 1.126,
|
| 114 |
+
"step": 30
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32989690721649484,
|
| 118 |
+
"grad_norm": 1.1148083209991455,
|
| 119 |
+
"learning_rate": 2e-05,
|
| 120 |
+
"loss": 0.5853,
|
| 121 |
+
"step": 32
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.35051546391752575,
|
| 125 |
+
"grad_norm": 1.284436821937561,
|
| 126 |
+
"learning_rate": 2e-05,
|
| 127 |
+
"loss": 0.6709,
|
| 128 |
+
"step": 34
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3711340206185567,
|
| 132 |
+
"grad_norm": 1.7680528163909912,
|
| 133 |
+
"learning_rate": 2e-05,
|
| 134 |
+
"loss": 0.6444,
|
| 135 |
+
"step": 36
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3917525773195876,
|
| 139 |
+
"grad_norm": 1.0730924606323242,
|
| 140 |
+
"learning_rate": 2e-05,
|
| 141 |
+
"loss": 0.6197,
|
| 142 |
+
"step": 38
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.41237113402061853,
|
| 146 |
+
"grad_norm": 0.9840460419654846,
|
| 147 |
+
"learning_rate": 2e-05,
|
| 148 |
+
"loss": 0.9225,
|
| 149 |
+
"step": 40
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4329896907216495,
|
| 153 |
+
"grad_norm": 2.7215404510498047,
|
| 154 |
+
"learning_rate": 2e-05,
|
| 155 |
+
"loss": 1.1065,
|
| 156 |
+
"step": 42
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.4536082474226804,
|
| 160 |
+
"grad_norm": 1.0859086513519287,
|
| 161 |
+
"learning_rate": 2e-05,
|
| 162 |
+
"loss": 1.6448,
|
| 163 |
+
"step": 44
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.4742268041237113,
|
| 167 |
+
"grad_norm": 0.6916680335998535,
|
| 168 |
+
"learning_rate": 2e-05,
|
| 169 |
+
"loss": 0.5747,
|
| 170 |
+
"step": 46
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.4948453608247423,
|
| 174 |
+
"grad_norm": 0.5634881854057312,
|
| 175 |
+
"learning_rate": 2e-05,
|
| 176 |
+
"loss": 0.8072,
|
| 177 |
+
"step": 48
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5154639175257731,
|
| 181 |
+
"grad_norm": 1.403308629989624,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 0.757,
|
| 184 |
+
"step": 50
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5360824742268041,
|
| 188 |
+
"grad_norm": 1.0651711225509644,
|
| 189 |
+
"learning_rate": 2e-05,
|
| 190 |
+
"loss": 1.2329,
|
| 191 |
+
"step": 52
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5567010309278351,
|
| 195 |
+
"grad_norm": 3.2769243717193604,
|
| 196 |
+
"learning_rate": 2e-05,
|
| 197 |
+
"loss": 0.9653,
|
| 198 |
+
"step": 54
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5773195876288659,
|
| 202 |
+
"grad_norm": 1.2478959560394287,
|
| 203 |
+
"learning_rate": 2e-05,
|
| 204 |
+
"loss": 0.8591,
|
| 205 |
+
"step": 56
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5979381443298969,
|
| 209 |
+
"grad_norm": 3.3655834197998047,
|
| 210 |
+
"learning_rate": 2e-05,
|
| 211 |
+
"loss": 1.6742,
|
| 212 |
+
"step": 58
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6185567010309279,
|
| 216 |
+
"grad_norm": 1.0023908615112305,
|
| 217 |
+
"learning_rate": 2e-05,
|
| 218 |
+
"loss": 1.0587,
|
| 219 |
+
"step": 60
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6391752577319587,
|
| 223 |
+
"grad_norm": 4.638378620147705,
|
| 224 |
+
"learning_rate": 2e-05,
|
| 225 |
+
"loss": 1.052,
|
| 226 |
+
"step": 62
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6597938144329897,
|
| 230 |
+
"grad_norm": 1.3997457027435303,
|
| 231 |
+
"learning_rate": 2e-05,
|
| 232 |
+
"loss": 0.7772,
|
| 233 |
+
"step": 64
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6804123711340206,
|
| 237 |
+
"grad_norm": 2.229414224624634,
|
| 238 |
+
"learning_rate": 2e-05,
|
| 239 |
+
"loss": 1.2991,
|
| 240 |
+
"step": 66
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.7010309278350515,
|
| 244 |
+
"grad_norm": 0.4760843515396118,
|
| 245 |
+
"learning_rate": 2e-05,
|
| 246 |
+
"loss": 0.9409,
|
| 247 |
+
"step": 68
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7216494845360825,
|
| 251 |
+
"grad_norm": 1.4938956499099731,
|
| 252 |
+
"learning_rate": 2e-05,
|
| 253 |
+
"loss": 1.0356,
|
| 254 |
+
"step": 70
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7422680412371134,
|
| 258 |
+
"grad_norm": 1.7016663551330566,
|
| 259 |
+
"learning_rate": 2e-05,
|
| 260 |
+
"loss": 0.9062,
|
| 261 |
+
"step": 72
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7628865979381443,
|
| 265 |
+
"grad_norm": 1.5440024137496948,
|
| 266 |
+
"learning_rate": 2e-05,
|
| 267 |
+
"loss": 1.5393,
|
| 268 |
+
"step": 74
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7835051546391752,
|
| 272 |
+
"grad_norm": 0.718760073184967,
|
| 273 |
+
"learning_rate": 2e-05,
|
| 274 |
+
"loss": 0.5889,
|
| 275 |
+
"step": 76
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8041237113402062,
|
| 279 |
+
"grad_norm": 1.5525449514389038,
|
| 280 |
+
"learning_rate": 2e-05,
|
| 281 |
+
"loss": 1.275,
|
| 282 |
+
"step": 78
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8247422680412371,
|
| 286 |
+
"grad_norm": 0.4799572229385376,
|
| 287 |
+
"learning_rate": 2e-05,
|
| 288 |
+
"loss": 0.7098,
|
| 289 |
+
"step": 80
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.845360824742268,
|
| 293 |
+
"grad_norm": 1.6241955757141113,
|
| 294 |
+
"learning_rate": 2e-05,
|
| 295 |
+
"loss": 1.1278,
|
| 296 |
+
"step": 82
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.865979381443299,
|
| 300 |
+
"grad_norm": 1.5434255599975586,
|
| 301 |
+
"learning_rate": 2e-05,
|
| 302 |
+
"loss": 0.8865,
|
| 303 |
+
"step": 84
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8865979381443299,
|
| 307 |
+
"grad_norm": 4.1449079513549805,
|
| 308 |
+
"learning_rate": 2e-05,
|
| 309 |
+
"loss": 1.3477,
|
| 310 |
+
"step": 86
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9072164948453608,
|
| 314 |
+
"grad_norm": 1.316735029220581,
|
| 315 |
+
"learning_rate": 2e-05,
|
| 316 |
+
"loss": 0.5452,
|
| 317 |
+
"step": 88
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9278350515463918,
|
| 321 |
+
"grad_norm": 1.4440531730651855,
|
| 322 |
+
"learning_rate": 2e-05,
|
| 323 |
+
"loss": 0.7568,
|
| 324 |
+
"step": 90
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.9484536082474226,
|
| 328 |
+
"grad_norm": 0.6465669870376587,
|
| 329 |
+
"learning_rate": 2e-05,
|
| 330 |
+
"loss": 0.5148,
|
| 331 |
+
"step": 92
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.9690721649484536,
|
| 335 |
+
"grad_norm": 3.1840262413024902,
|
| 336 |
+
"learning_rate": 2e-05,
|
| 337 |
+
"loss": 0.8276,
|
| 338 |
+
"step": 94
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9896907216494846,
|
| 342 |
+
"grad_norm": 2.191006898880005,
|
| 343 |
+
"learning_rate": 2e-05,
|
| 344 |
+
"loss": 1.0671,
|
| 345 |
+
"step": 96
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.0,
|
| 349 |
+
"step": 97,
|
| 350 |
+
"total_flos": 1.3522822525616128e+16,
|
| 351 |
+
"train_loss": 1.024123240991966,
|
| 352 |
+
"train_runtime": 329.2802,
|
| 353 |
+
"train_samples_per_second": 1.178,
|
| 354 |
+
"train_steps_per_second": 0.295
|
| 355 |
+
}
|
| 356 |
+
],
|
| 357 |
+
"logging_steps": 2,
|
| 358 |
+
"max_steps": 97,
|
| 359 |
+
"num_input_tokens_seen": 0,
|
| 360 |
+
"num_train_epochs": 1,
|
| 361 |
+
"save_steps": 500,
|
| 362 |
+
"stateful_callbacks": {
|
| 363 |
+
"TrainerControl": {
|
| 364 |
+
"args": {
|
| 365 |
+
"should_epoch_stop": false,
|
| 366 |
+
"should_evaluate": false,
|
| 367 |
+
"should_log": false,
|
| 368 |
+
"should_save": false,
|
| 369 |
+
"should_training_stop": false
|
| 370 |
+
},
|
| 371 |
+
"attributes": {}
|
| 372 |
+
}
|
| 373 |
+
},
|
| 374 |
+
"total_flos": 1.3522822525616128e+16,
|
| 375 |
+
"train_batch_size": 1,
|
| 376 |
+
"trial_name": null,
|
| 377 |
+
"trial_params": null
|
| 378 |
+
}
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round10.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1953a5d682996bf7f4974e09538f0db4357535a914fdb147d514c228417d8158
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round12.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ed6b8398417dec05d433da0925ff0676355f83e22912bb2f246305243882495
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round15.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c86152694df067a9e575791bae4a78d113d1c1f3913acddace1c81189a40bf60
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round17.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:007fd7465945f737972fd09d5a5bfac976291174c2cc1f3e6c0abd42fb1e15cd
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be898f74488af89b95d07167565247944b62909ca7fb60878c1b9188e44ae294
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round20.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:660fb4b5ab21dd5fa6cd373a6b44bf7bd6aa27bb38b6ba9ee00958c570110ae9
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:072d545c26b4cc2a1e2ac58d2d97e7ee52ea4b8b2e6598e811dc3ab58507587d
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_client_model_round7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:01c36cedd65d7a1994dcd41a9ed47d891dcca4dba251524fa7d3877c98a16075
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/2_trainer_state.json
ADDED
|
@@ -0,0 +1,378 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 97,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.020618556701030927,
|
| 13 |
+
"grad_norm": 2.715688943862915,
|
| 14 |
+
"learning_rate": 2e-05,
|
| 15 |
+
"loss": 0.7903,
|
| 16 |
+
"step": 2
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.041237113402061855,
|
| 20 |
+
"grad_norm": 1.7270240783691406,
|
| 21 |
+
"learning_rate": 2e-05,
|
| 22 |
+
"loss": 0.8497,
|
| 23 |
+
"step": 4
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.061855670103092786,
|
| 27 |
+
"grad_norm": 2.6262009143829346,
|
| 28 |
+
"learning_rate": 2e-05,
|
| 29 |
+
"loss": 2.5311,
|
| 30 |
+
"step": 6
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08247422680412371,
|
| 34 |
+
"grad_norm": 2.0781748294830322,
|
| 35 |
+
"learning_rate": 2e-05,
|
| 36 |
+
"loss": 0.6511,
|
| 37 |
+
"step": 8
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10309278350515463,
|
| 41 |
+
"grad_norm": 3.878676176071167,
|
| 42 |
+
"learning_rate": 2e-05,
|
| 43 |
+
"loss": 1.6893,
|
| 44 |
+
"step": 10
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12371134020618557,
|
| 48 |
+
"grad_norm": 0.30152788758277893,
|
| 49 |
+
"learning_rate": 2e-05,
|
| 50 |
+
"loss": 0.2982,
|
| 51 |
+
"step": 12
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14432989690721648,
|
| 55 |
+
"grad_norm": 2.4543538093566895,
|
| 56 |
+
"learning_rate": 2e-05,
|
| 57 |
+
"loss": 1.4963,
|
| 58 |
+
"step": 14
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.16494845360824742,
|
| 62 |
+
"grad_norm": 1.1550655364990234,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 0.5657,
|
| 65 |
+
"step": 16
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.18556701030927836,
|
| 69 |
+
"grad_norm": 1.4158210754394531,
|
| 70 |
+
"learning_rate": 2e-05,
|
| 71 |
+
"loss": 0.6207,
|
| 72 |
+
"step": 18
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.20618556701030927,
|
| 76 |
+
"grad_norm": 1.8720401525497437,
|
| 77 |
+
"learning_rate": 2e-05,
|
| 78 |
+
"loss": 0.9528,
|
| 79 |
+
"step": 20
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2268041237113402,
|
| 83 |
+
"grad_norm": 1.3980697393417358,
|
| 84 |
+
"learning_rate": 2e-05,
|
| 85 |
+
"loss": 0.7742,
|
| 86 |
+
"step": 22
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.24742268041237114,
|
| 90 |
+
"grad_norm": 0.7552713751792908,
|
| 91 |
+
"learning_rate": 2e-05,
|
| 92 |
+
"loss": 0.56,
|
| 93 |
+
"step": 24
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.26804123711340205,
|
| 97 |
+
"grad_norm": 2.9224400520324707,
|
| 98 |
+
"learning_rate": 2e-05,
|
| 99 |
+
"loss": 0.761,
|
| 100 |
+
"step": 26
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.28865979381443296,
|
| 104 |
+
"grad_norm": 2.1200363636016846,
|
| 105 |
+
"learning_rate": 2e-05,
|
| 106 |
+
"loss": 1.8194,
|
| 107 |
+
"step": 28
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.30927835051546393,
|
| 111 |
+
"grad_norm": 1.0131165981292725,
|
| 112 |
+
"learning_rate": 2e-05,
|
| 113 |
+
"loss": 0.5858,
|
| 114 |
+
"step": 30
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32989690721649484,
|
| 118 |
+
"grad_norm": 1.1930532455444336,
|
| 119 |
+
"learning_rate": 2e-05,
|
| 120 |
+
"loss": 0.3527,
|
| 121 |
+
"step": 32
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.35051546391752575,
|
| 125 |
+
"grad_norm": 0.8298959136009216,
|
| 126 |
+
"learning_rate": 2e-05,
|
| 127 |
+
"loss": 0.7283,
|
| 128 |
+
"step": 34
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3711340206185567,
|
| 132 |
+
"grad_norm": 0.21372075378894806,
|
| 133 |
+
"learning_rate": 2e-05,
|
| 134 |
+
"loss": 0.679,
|
| 135 |
+
"step": 36
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3917525773195876,
|
| 139 |
+
"grad_norm": 0.6620970368385315,
|
| 140 |
+
"learning_rate": 2e-05,
|
| 141 |
+
"loss": 0.308,
|
| 142 |
+
"step": 38
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.41237113402061853,
|
| 146 |
+
"grad_norm": 1.3406527042388916,
|
| 147 |
+
"learning_rate": 2e-05,
|
| 148 |
+
"loss": 1.0992,
|
| 149 |
+
"step": 40
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4329896907216495,
|
| 153 |
+
"grad_norm": 8.218306541442871,
|
| 154 |
+
"learning_rate": 2e-05,
|
| 155 |
+
"loss": 2.0959,
|
| 156 |
+
"step": 42
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.4536082474226804,
|
| 160 |
+
"grad_norm": 1.171020269393921,
|
| 161 |
+
"learning_rate": 2e-05,
|
| 162 |
+
"loss": 0.6386,
|
| 163 |
+
"step": 44
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.4742268041237113,
|
| 167 |
+
"grad_norm": 2.1302266120910645,
|
| 168 |
+
"learning_rate": 2e-05,
|
| 169 |
+
"loss": 1.7207,
|
| 170 |
+
"step": 46
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.4948453608247423,
|
| 174 |
+
"grad_norm": 1.39640474319458,
|
| 175 |
+
"learning_rate": 2e-05,
|
| 176 |
+
"loss": 1.7815,
|
| 177 |
+
"step": 48
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5154639175257731,
|
| 181 |
+
"grad_norm": 1.7075337171554565,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 1.1565,
|
| 184 |
+
"step": 50
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5360824742268041,
|
| 188 |
+
"grad_norm": 0.19492167234420776,
|
| 189 |
+
"learning_rate": 2e-05,
|
| 190 |
+
"loss": 0.3214,
|
| 191 |
+
"step": 52
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5567010309278351,
|
| 195 |
+
"grad_norm": 2.8099043369293213,
|
| 196 |
+
"learning_rate": 2e-05,
|
| 197 |
+
"loss": 1.2814,
|
| 198 |
+
"step": 54
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5773195876288659,
|
| 202 |
+
"grad_norm": 3.1069672107696533,
|
| 203 |
+
"learning_rate": 2e-05,
|
| 204 |
+
"loss": 1.6989,
|
| 205 |
+
"step": 56
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5979381443298969,
|
| 209 |
+
"grad_norm": 1.5938727855682373,
|
| 210 |
+
"learning_rate": 2e-05,
|
| 211 |
+
"loss": 1.0741,
|
| 212 |
+
"step": 58
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6185567010309279,
|
| 216 |
+
"grad_norm": 2.409372329711914,
|
| 217 |
+
"learning_rate": 2e-05,
|
| 218 |
+
"loss": 0.712,
|
| 219 |
+
"step": 60
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6391752577319587,
|
| 223 |
+
"grad_norm": 1.280025839805603,
|
| 224 |
+
"learning_rate": 2e-05,
|
| 225 |
+
"loss": 0.6074,
|
| 226 |
+
"step": 62
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6597938144329897,
|
| 230 |
+
"grad_norm": 1.893931269645691,
|
| 231 |
+
"learning_rate": 2e-05,
|
| 232 |
+
"loss": 0.9652,
|
| 233 |
+
"step": 64
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6804123711340206,
|
| 237 |
+
"grad_norm": 1.0092408657073975,
|
| 238 |
+
"learning_rate": 2e-05,
|
| 239 |
+
"loss": 0.5104,
|
| 240 |
+
"step": 66
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.7010309278350515,
|
| 244 |
+
"grad_norm": 1.1251095533370972,
|
| 245 |
+
"learning_rate": 2e-05,
|
| 246 |
+
"loss": 0.5629,
|
| 247 |
+
"step": 68
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7216494845360825,
|
| 251 |
+
"grad_norm": 3.200808048248291,
|
| 252 |
+
"learning_rate": 2e-05,
|
| 253 |
+
"loss": 1.9959,
|
| 254 |
+
"step": 70
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7422680412371134,
|
| 258 |
+
"grad_norm": 2.1445953845977783,
|
| 259 |
+
"learning_rate": 2e-05,
|
| 260 |
+
"loss": 0.8926,
|
| 261 |
+
"step": 72
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7628865979381443,
|
| 265 |
+
"grad_norm": 0.7390415072441101,
|
| 266 |
+
"learning_rate": 2e-05,
|
| 267 |
+
"loss": 0.3945,
|
| 268 |
+
"step": 74
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7835051546391752,
|
| 272 |
+
"grad_norm": 3.7797181606292725,
|
| 273 |
+
"learning_rate": 2e-05,
|
| 274 |
+
"loss": 1.2296,
|
| 275 |
+
"step": 76
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8041237113402062,
|
| 279 |
+
"grad_norm": 1.9498169422149658,
|
| 280 |
+
"learning_rate": 2e-05,
|
| 281 |
+
"loss": 0.9621,
|
| 282 |
+
"step": 78
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8247422680412371,
|
| 286 |
+
"grad_norm": 2.3580148220062256,
|
| 287 |
+
"learning_rate": 2e-05,
|
| 288 |
+
"loss": 1.074,
|
| 289 |
+
"step": 80
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.845360824742268,
|
| 293 |
+
"grad_norm": 3.7689895629882812,
|
| 294 |
+
"learning_rate": 2e-05,
|
| 295 |
+
"loss": 0.8532,
|
| 296 |
+
"step": 82
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.865979381443299,
|
| 300 |
+
"grad_norm": 1.0212076902389526,
|
| 301 |
+
"learning_rate": 2e-05,
|
| 302 |
+
"loss": 0.3944,
|
| 303 |
+
"step": 84
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8865979381443299,
|
| 307 |
+
"grad_norm": 1.0816051959991455,
|
| 308 |
+
"learning_rate": 2e-05,
|
| 309 |
+
"loss": 0.7249,
|
| 310 |
+
"step": 86
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9072164948453608,
|
| 314 |
+
"grad_norm": 0.7434933185577393,
|
| 315 |
+
"learning_rate": 2e-05,
|
| 316 |
+
"loss": 0.4865,
|
| 317 |
+
"step": 88
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9278350515463918,
|
| 321 |
+
"grad_norm": 1.409759521484375,
|
| 322 |
+
"learning_rate": 2e-05,
|
| 323 |
+
"loss": 0.5403,
|
| 324 |
+
"step": 90
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.9484536082474226,
|
| 328 |
+
"grad_norm": 2.005401849746704,
|
| 329 |
+
"learning_rate": 2e-05,
|
| 330 |
+
"loss": 1.0927,
|
| 331 |
+
"step": 92
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.9690721649484536,
|
| 335 |
+
"grad_norm": 3.6402158737182617,
|
| 336 |
+
"learning_rate": 2e-05,
|
| 337 |
+
"loss": 0.7249,
|
| 338 |
+
"step": 94
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9896907216494846,
|
| 342 |
+
"grad_norm": 2.5542266368865967,
|
| 343 |
+
"learning_rate": 2e-05,
|
| 344 |
+
"loss": 1.3884,
|
| 345 |
+
"step": 96
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.0,
|
| 349 |
+
"step": 97,
|
| 350 |
+
"total_flos": 1.5329114357497856e+16,
|
| 351 |
+
"train_loss": 0.9519824686738634,
|
| 352 |
+
"train_runtime": 357.0654,
|
| 353 |
+
"train_samples_per_second": 1.087,
|
| 354 |
+
"train_steps_per_second": 0.272
|
| 355 |
+
}
|
| 356 |
+
],
|
| 357 |
+
"logging_steps": 2,
|
| 358 |
+
"max_steps": 97,
|
| 359 |
+
"num_input_tokens_seen": 0,
|
| 360 |
+
"num_train_epochs": 1,
|
| 361 |
+
"save_steps": 500,
|
| 362 |
+
"stateful_callbacks": {
|
| 363 |
+
"TrainerControl": {
|
| 364 |
+
"args": {
|
| 365 |
+
"should_epoch_stop": false,
|
| 366 |
+
"should_evaluate": false,
|
| 367 |
+
"should_log": false,
|
| 368 |
+
"should_save": false,
|
| 369 |
+
"should_training_stop": false
|
| 370 |
+
},
|
| 371 |
+
"attributes": {}
|
| 372 |
+
}
|
| 373 |
+
},
|
| 374 |
+
"total_flos": 1.5329114357497856e+16,
|
| 375 |
+
"train_batch_size": 1,
|
| 376 |
+
"trial_name": null,
|
| 377 |
+
"trial_params": null
|
| 378 |
+
}
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round10.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4803816dd8fe4467fa5f43227533a56f083b8ad0f6ddd457d12f761c554043f4
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round12.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17a6865b4d478bbb296c1066419695af40fc07ad1147c3fbff13658f9808bc29
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round15.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b66cc40edb9397433df9178db225f73eafa6ea65334a7650e8930a9fbbb8b4a4
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round17.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68112bcb7c765592e71c2a4308eb7fc02b16edd1004be3c05f8ea3fb426f24c7
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:821b9803cea659a4bee96d4875ddcf797492fe28233d4eb5c9f67c7c429d5d60
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round20.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2bf168da76f1338f80db83394c6a07d9359185723fa8e27c2aee29c881c64f1
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ceb89ff4f0927002a37fbf5b1b3961045b663c7ab2467c070fde86fc9f14a14
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_client_model_round7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fffbadb068adde75a38be22546d8dfa6c08ffaa70b7f23aa68aeb6eff62809f
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/3_trainer_state.json
ADDED
|
@@ -0,0 +1,378 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 97,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.020618556701030927,
|
| 13 |
+
"grad_norm": 0.5930002927780151,
|
| 14 |
+
"learning_rate": 2e-05,
|
| 15 |
+
"loss": 1.7236,
|
| 16 |
+
"step": 2
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.041237113402061855,
|
| 20 |
+
"grad_norm": 0.7709837555885315,
|
| 21 |
+
"learning_rate": 2e-05,
|
| 22 |
+
"loss": 2.0088,
|
| 23 |
+
"step": 4
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.061855670103092786,
|
| 27 |
+
"grad_norm": 0.4815305769443512,
|
| 28 |
+
"learning_rate": 2e-05,
|
| 29 |
+
"loss": 1.4131,
|
| 30 |
+
"step": 6
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08247422680412371,
|
| 34 |
+
"grad_norm": 0.48609688878059387,
|
| 35 |
+
"learning_rate": 2e-05,
|
| 36 |
+
"loss": 1.1724,
|
| 37 |
+
"step": 8
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10309278350515463,
|
| 41 |
+
"grad_norm": 0.4869016706943512,
|
| 42 |
+
"learning_rate": 2e-05,
|
| 43 |
+
"loss": 1.7295,
|
| 44 |
+
"step": 10
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12371134020618557,
|
| 48 |
+
"grad_norm": 0.37476807832717896,
|
| 49 |
+
"learning_rate": 2e-05,
|
| 50 |
+
"loss": 1.6475,
|
| 51 |
+
"step": 12
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14432989690721648,
|
| 55 |
+
"grad_norm": 0.6886877417564392,
|
| 56 |
+
"learning_rate": 2e-05,
|
| 57 |
+
"loss": 1.4917,
|
| 58 |
+
"step": 14
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.16494845360824742,
|
| 62 |
+
"grad_norm": 0.5024330615997314,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 1.4277,
|
| 65 |
+
"step": 16
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.18556701030927836,
|
| 69 |
+
"grad_norm": 0.6298432946205139,
|
| 70 |
+
"learning_rate": 2e-05,
|
| 71 |
+
"loss": 1.5801,
|
| 72 |
+
"step": 18
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.20618556701030927,
|
| 76 |
+
"grad_norm": 0.46709224581718445,
|
| 77 |
+
"learning_rate": 2e-05,
|
| 78 |
+
"loss": 1.625,
|
| 79 |
+
"step": 20
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2268041237113402,
|
| 83 |
+
"grad_norm": 0.4031917154788971,
|
| 84 |
+
"learning_rate": 2e-05,
|
| 85 |
+
"loss": 1.5425,
|
| 86 |
+
"step": 22
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.24742268041237114,
|
| 90 |
+
"grad_norm": 0.5809447169303894,
|
| 91 |
+
"learning_rate": 2e-05,
|
| 92 |
+
"loss": 1.8184,
|
| 93 |
+
"step": 24
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.26804123711340205,
|
| 97 |
+
"grad_norm": 0.43344253301620483,
|
| 98 |
+
"learning_rate": 2e-05,
|
| 99 |
+
"loss": 1.6924,
|
| 100 |
+
"step": 26
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.28865979381443296,
|
| 104 |
+
"grad_norm": 0.5011717081069946,
|
| 105 |
+
"learning_rate": 2e-05,
|
| 106 |
+
"loss": 1.3584,
|
| 107 |
+
"step": 28
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.30927835051546393,
|
| 111 |
+
"grad_norm": 0.4586883783340454,
|
| 112 |
+
"learning_rate": 2e-05,
|
| 113 |
+
"loss": 1.501,
|
| 114 |
+
"step": 30
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32989690721649484,
|
| 118 |
+
"grad_norm": 0.6749467253684998,
|
| 119 |
+
"learning_rate": 2e-05,
|
| 120 |
+
"loss": 1.6797,
|
| 121 |
+
"step": 32
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.35051546391752575,
|
| 125 |
+
"grad_norm": 0.38988545536994934,
|
| 126 |
+
"learning_rate": 2e-05,
|
| 127 |
+
"loss": 1.3579,
|
| 128 |
+
"step": 34
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3711340206185567,
|
| 132 |
+
"grad_norm": 0.48226863145828247,
|
| 133 |
+
"learning_rate": 2e-05,
|
| 134 |
+
"loss": 1.707,
|
| 135 |
+
"step": 36
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3917525773195876,
|
| 139 |
+
"grad_norm": 0.48712313175201416,
|
| 140 |
+
"learning_rate": 2e-05,
|
| 141 |
+
"loss": 1.4121,
|
| 142 |
+
"step": 38
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.41237113402061853,
|
| 146 |
+
"grad_norm": 0.44966915249824524,
|
| 147 |
+
"learning_rate": 2e-05,
|
| 148 |
+
"loss": 1.6276,
|
| 149 |
+
"step": 40
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4329896907216495,
|
| 153 |
+
"grad_norm": 0.5168526768684387,
|
| 154 |
+
"learning_rate": 2e-05,
|
| 155 |
+
"loss": 1.6426,
|
| 156 |
+
"step": 42
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.4536082474226804,
|
| 160 |
+
"grad_norm": 0.5030601620674133,
|
| 161 |
+
"learning_rate": 2e-05,
|
| 162 |
+
"loss": 1.5693,
|
| 163 |
+
"step": 44
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.4742268041237113,
|
| 167 |
+
"grad_norm": 0.7527281045913696,
|
| 168 |
+
"learning_rate": 2e-05,
|
| 169 |
+
"loss": 1.5518,
|
| 170 |
+
"step": 46
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.4948453608247423,
|
| 174 |
+
"grad_norm": 1.2194421291351318,
|
| 175 |
+
"learning_rate": 2e-05,
|
| 176 |
+
"loss": 1.5342,
|
| 177 |
+
"step": 48
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5154639175257731,
|
| 181 |
+
"grad_norm": 0.5955514907836914,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 1.5137,
|
| 184 |
+
"step": 50
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5360824742268041,
|
| 188 |
+
"grad_norm": 0.7728529572486877,
|
| 189 |
+
"learning_rate": 2e-05,
|
| 190 |
+
"loss": 1.6963,
|
| 191 |
+
"step": 52
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5567010309278351,
|
| 195 |
+
"grad_norm": 0.73228919506073,
|
| 196 |
+
"learning_rate": 2e-05,
|
| 197 |
+
"loss": 1.4502,
|
| 198 |
+
"step": 54
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5773195876288659,
|
| 202 |
+
"grad_norm": 0.4163931906223297,
|
| 203 |
+
"learning_rate": 2e-05,
|
| 204 |
+
"loss": 1.4219,
|
| 205 |
+
"step": 56
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5979381443298969,
|
| 209 |
+
"grad_norm": 0.5588869452476501,
|
| 210 |
+
"learning_rate": 2e-05,
|
| 211 |
+
"loss": 1.1425,
|
| 212 |
+
"step": 58
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6185567010309279,
|
| 216 |
+
"grad_norm": 0.6062840223312378,
|
| 217 |
+
"learning_rate": 2e-05,
|
| 218 |
+
"loss": 1.4331,
|
| 219 |
+
"step": 60
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6391752577319587,
|
| 223 |
+
"grad_norm": 1.0351043939590454,
|
| 224 |
+
"learning_rate": 2e-05,
|
| 225 |
+
"loss": 1.4609,
|
| 226 |
+
"step": 62
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6597938144329897,
|
| 230 |
+
"grad_norm": 0.6619802713394165,
|
| 231 |
+
"learning_rate": 2e-05,
|
| 232 |
+
"loss": 1.4189,
|
| 233 |
+
"step": 64
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6804123711340206,
|
| 237 |
+
"grad_norm": 0.37535732984542847,
|
| 238 |
+
"learning_rate": 2e-05,
|
| 239 |
+
"loss": 1.3813,
|
| 240 |
+
"step": 66
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.7010309278350515,
|
| 244 |
+
"grad_norm": 0.9288793802261353,
|
| 245 |
+
"learning_rate": 2e-05,
|
| 246 |
+
"loss": 1.397,
|
| 247 |
+
"step": 68
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7216494845360825,
|
| 251 |
+
"grad_norm": 1.1058944463729858,
|
| 252 |
+
"learning_rate": 2e-05,
|
| 253 |
+
"loss": 1.4009,
|
| 254 |
+
"step": 70
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7422680412371134,
|
| 258 |
+
"grad_norm": 0.7440882921218872,
|
| 259 |
+
"learning_rate": 2e-05,
|
| 260 |
+
"loss": 1.007,
|
| 261 |
+
"step": 72
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7628865979381443,
|
| 265 |
+
"grad_norm": 1.2134323120117188,
|
| 266 |
+
"learning_rate": 2e-05,
|
| 267 |
+
"loss": 1.3503,
|
| 268 |
+
"step": 74
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7835051546391752,
|
| 272 |
+
"grad_norm": 0.4168848693370819,
|
| 273 |
+
"learning_rate": 2e-05,
|
| 274 |
+
"loss": 1.0313,
|
| 275 |
+
"step": 76
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8041237113402062,
|
| 279 |
+
"grad_norm": 1.7301552295684814,
|
| 280 |
+
"learning_rate": 2e-05,
|
| 281 |
+
"loss": 1.4034,
|
| 282 |
+
"step": 78
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8247422680412371,
|
| 286 |
+
"grad_norm": 0.9088520407676697,
|
| 287 |
+
"learning_rate": 2e-05,
|
| 288 |
+
"loss": 1.8931,
|
| 289 |
+
"step": 80
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.845360824742268,
|
| 293 |
+
"grad_norm": 1.6952238082885742,
|
| 294 |
+
"learning_rate": 2e-05,
|
| 295 |
+
"loss": 1.0514,
|
| 296 |
+
"step": 82
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.865979381443299,
|
| 300 |
+
"grad_norm": 1.3661904335021973,
|
| 301 |
+
"learning_rate": 2e-05,
|
| 302 |
+
"loss": 1.5254,
|
| 303 |
+
"step": 84
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8865979381443299,
|
| 307 |
+
"grad_norm": 1.5365053415298462,
|
| 308 |
+
"learning_rate": 2e-05,
|
| 309 |
+
"loss": 1.5001,
|
| 310 |
+
"step": 86
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9072164948453608,
|
| 314 |
+
"grad_norm": 0.4049004018306732,
|
| 315 |
+
"learning_rate": 2e-05,
|
| 316 |
+
"loss": 1.4141,
|
| 317 |
+
"step": 88
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9278350515463918,
|
| 321 |
+
"grad_norm": 2.052964448928833,
|
| 322 |
+
"learning_rate": 2e-05,
|
| 323 |
+
"loss": 1.0455,
|
| 324 |
+
"step": 90
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.9484536082474226,
|
| 328 |
+
"grad_norm": 0.5465191602706909,
|
| 329 |
+
"learning_rate": 2e-05,
|
| 330 |
+
"loss": 0.741,
|
| 331 |
+
"step": 92
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.9690721649484536,
|
| 335 |
+
"grad_norm": 1.1672446727752686,
|
| 336 |
+
"learning_rate": 2e-05,
|
| 337 |
+
"loss": 1.199,
|
| 338 |
+
"step": 94
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9896907216494846,
|
| 342 |
+
"grad_norm": 5.172623634338379,
|
| 343 |
+
"learning_rate": 2e-05,
|
| 344 |
+
"loss": 1.6101,
|
| 345 |
+
"step": 96
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.0,
|
| 349 |
+
"step": 97,
|
| 350 |
+
"total_flos": 2.080121124238131e+16,
|
| 351 |
+
"train_loss": 1.4608057553006202,
|
| 352 |
+
"train_runtime": 363.6952,
|
| 353 |
+
"train_samples_per_second": 1.067,
|
| 354 |
+
"train_steps_per_second": 0.267
|
| 355 |
+
}
|
| 356 |
+
],
|
| 357 |
+
"logging_steps": 2,
|
| 358 |
+
"max_steps": 97,
|
| 359 |
+
"num_input_tokens_seen": 0,
|
| 360 |
+
"num_train_epochs": 1,
|
| 361 |
+
"save_steps": 500,
|
| 362 |
+
"stateful_callbacks": {
|
| 363 |
+
"TrainerControl": {
|
| 364 |
+
"args": {
|
| 365 |
+
"should_epoch_stop": false,
|
| 366 |
+
"should_evaluate": false,
|
| 367 |
+
"should_log": false,
|
| 368 |
+
"should_save": false,
|
| 369 |
+
"should_training_stop": false
|
| 370 |
+
},
|
| 371 |
+
"attributes": {}
|
| 372 |
+
}
|
| 373 |
+
},
|
| 374 |
+
"total_flos": 2.080121124238131e+16,
|
| 375 |
+
"train_batch_size": 1,
|
| 376 |
+
"trial_name": null,
|
| 377 |
+
"trial_params": null
|
| 378 |
+
}
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round10.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:628f3ab6268a995e5a6e695182629e912860d336c361110825fb7b14d330bdfc
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round12.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f1eee78bdc9cbaaa5ace24962d539c619df8286a9a8c422289db2d7ab9cf5b0
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round15.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e796e13255d7cee709ff69e3726f6fce731a851ab6157426a862a721e87d3cb
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round17.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cdf354f93337dea3a94a04a869e06347f6ddcdb402a620069706c8a3c277b8e4
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:54df4904f4d80d2309f8e2f522b18507c5d402c766e60b504ee27c785f98c7f6
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round20.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6cc7b2d7613df2e86e28c49ee82f08d88d80e81163985534d20ec6f6315e4088
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68b90ac9fa152488ded4afc57fe26add9bcab9bea5454ffa4ab484f45ff280c6
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_client_model_round7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:517901987a25dbccbe2e3c55929dd9f1d58b55e2c4ba4e450b7b6a6115e9ba8f
|
| 3 |
+
size 791576546
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/4_trainer_state.json
ADDED
|
@@ -0,0 +1,378 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 97,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.020618556701030927,
|
| 13 |
+
"grad_norm": 1.857858657836914,
|
| 14 |
+
"learning_rate": 2e-05,
|
| 15 |
+
"loss": 1.0093,
|
| 16 |
+
"step": 2
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.041237113402061855,
|
| 20 |
+
"grad_norm": 1.128074049949646,
|
| 21 |
+
"learning_rate": 2e-05,
|
| 22 |
+
"loss": 0.9811,
|
| 23 |
+
"step": 4
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.061855670103092786,
|
| 27 |
+
"grad_norm": 1.1853967905044556,
|
| 28 |
+
"learning_rate": 2e-05,
|
| 29 |
+
"loss": 0.6298,
|
| 30 |
+
"step": 6
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08247422680412371,
|
| 34 |
+
"grad_norm": 2.01977276802063,
|
| 35 |
+
"learning_rate": 2e-05,
|
| 36 |
+
"loss": 0.7973,
|
| 37 |
+
"step": 8
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.10309278350515463,
|
| 41 |
+
"grad_norm": 3.552171468734741,
|
| 42 |
+
"learning_rate": 2e-05,
|
| 43 |
+
"loss": 3.2413,
|
| 44 |
+
"step": 10
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12371134020618557,
|
| 48 |
+
"grad_norm": 2.4959380626678467,
|
| 49 |
+
"learning_rate": 2e-05,
|
| 50 |
+
"loss": 0.8036,
|
| 51 |
+
"step": 12
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.14432989690721648,
|
| 55 |
+
"grad_norm": 0.18528757989406586,
|
| 56 |
+
"learning_rate": 2e-05,
|
| 57 |
+
"loss": 0.8049,
|
| 58 |
+
"step": 14
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.16494845360824742,
|
| 62 |
+
"grad_norm": 0.3887205421924591,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 1.034,
|
| 65 |
+
"step": 16
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.18556701030927836,
|
| 69 |
+
"grad_norm": 1.167763352394104,
|
| 70 |
+
"learning_rate": 2e-05,
|
| 71 |
+
"loss": 1.0904,
|
| 72 |
+
"step": 18
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.20618556701030927,
|
| 76 |
+
"grad_norm": 1.1180120706558228,
|
| 77 |
+
"learning_rate": 2e-05,
|
| 78 |
+
"loss": 1.1927,
|
| 79 |
+
"step": 20
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2268041237113402,
|
| 83 |
+
"grad_norm": 0.34743738174438477,
|
| 84 |
+
"learning_rate": 2e-05,
|
| 85 |
+
"loss": 0.9405,
|
| 86 |
+
"step": 22
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.24742268041237114,
|
| 90 |
+
"grad_norm": 0.5589756965637207,
|
| 91 |
+
"learning_rate": 2e-05,
|
| 92 |
+
"loss": 1.2665,
|
| 93 |
+
"step": 24
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.26804123711340205,
|
| 97 |
+
"grad_norm": 1.364461898803711,
|
| 98 |
+
"learning_rate": 2e-05,
|
| 99 |
+
"loss": 0.9861,
|
| 100 |
+
"step": 26
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.28865979381443296,
|
| 104 |
+
"grad_norm": 1.4796721935272217,
|
| 105 |
+
"learning_rate": 2e-05,
|
| 106 |
+
"loss": 1.282,
|
| 107 |
+
"step": 28
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.30927835051546393,
|
| 111 |
+
"grad_norm": 1.5695762634277344,
|
| 112 |
+
"learning_rate": 2e-05,
|
| 113 |
+
"loss": 2.2431,
|
| 114 |
+
"step": 30
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.32989690721649484,
|
| 118 |
+
"grad_norm": 1.308424949645996,
|
| 119 |
+
"learning_rate": 2e-05,
|
| 120 |
+
"loss": 1.0269,
|
| 121 |
+
"step": 32
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.35051546391752575,
|
| 125 |
+
"grad_norm": 0.9424493908882141,
|
| 126 |
+
"learning_rate": 2e-05,
|
| 127 |
+
"loss": 0.8531,
|
| 128 |
+
"step": 34
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3711340206185567,
|
| 132 |
+
"grad_norm": 1.148472547531128,
|
| 133 |
+
"learning_rate": 2e-05,
|
| 134 |
+
"loss": 0.954,
|
| 135 |
+
"step": 36
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3917525773195876,
|
| 139 |
+
"grad_norm": 2.9015464782714844,
|
| 140 |
+
"learning_rate": 2e-05,
|
| 141 |
+
"loss": 1.3365,
|
| 142 |
+
"step": 38
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.41237113402061853,
|
| 146 |
+
"grad_norm": 2.0090081691741943,
|
| 147 |
+
"learning_rate": 2e-05,
|
| 148 |
+
"loss": 1.3861,
|
| 149 |
+
"step": 40
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4329896907216495,
|
| 153 |
+
"grad_norm": 1.2990866899490356,
|
| 154 |
+
"learning_rate": 2e-05,
|
| 155 |
+
"loss": 0.435,
|
| 156 |
+
"step": 42
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.4536082474226804,
|
| 160 |
+
"grad_norm": 1.2898091077804565,
|
| 161 |
+
"learning_rate": 2e-05,
|
| 162 |
+
"loss": 1.7354,
|
| 163 |
+
"step": 44
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.4742268041237113,
|
| 167 |
+
"grad_norm": 1.301903486251831,
|
| 168 |
+
"learning_rate": 2e-05,
|
| 169 |
+
"loss": 0.923,
|
| 170 |
+
"step": 46
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.4948453608247423,
|
| 174 |
+
"grad_norm": 1.1289761066436768,
|
| 175 |
+
"learning_rate": 2e-05,
|
| 176 |
+
"loss": 0.4642,
|
| 177 |
+
"step": 48
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5154639175257731,
|
| 181 |
+
"grad_norm": 3.627347707748413,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 1.187,
|
| 184 |
+
"step": 50
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5360824742268041,
|
| 188 |
+
"grad_norm": 0.9005461931228638,
|
| 189 |
+
"learning_rate": 2e-05,
|
| 190 |
+
"loss": 0.7929,
|
| 191 |
+
"step": 52
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5567010309278351,
|
| 195 |
+
"grad_norm": 1.4061144590377808,
|
| 196 |
+
"learning_rate": 2e-05,
|
| 197 |
+
"loss": 1.3737,
|
| 198 |
+
"step": 54
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5773195876288659,
|
| 202 |
+
"grad_norm": 1.6655174493789673,
|
| 203 |
+
"learning_rate": 2e-05,
|
| 204 |
+
"loss": 1.4762,
|
| 205 |
+
"step": 56
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5979381443298969,
|
| 209 |
+
"grad_norm": 2.249619483947754,
|
| 210 |
+
"learning_rate": 2e-05,
|
| 211 |
+
"loss": 1.3758,
|
| 212 |
+
"step": 58
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6185567010309279,
|
| 216 |
+
"grad_norm": 3.719571113586426,
|
| 217 |
+
"learning_rate": 2e-05,
|
| 218 |
+
"loss": 1.0628,
|
| 219 |
+
"step": 60
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6391752577319587,
|
| 223 |
+
"grad_norm": 2.0385303497314453,
|
| 224 |
+
"learning_rate": 2e-05,
|
| 225 |
+
"loss": 0.8212,
|
| 226 |
+
"step": 62
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6597938144329897,
|
| 230 |
+
"grad_norm": 1.5352033376693726,
|
| 231 |
+
"learning_rate": 2e-05,
|
| 232 |
+
"loss": 0.5521,
|
| 233 |
+
"step": 64
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6804123711340206,
|
| 237 |
+
"grad_norm": 1.1133532524108887,
|
| 238 |
+
"learning_rate": 2e-05,
|
| 239 |
+
"loss": 1.0727,
|
| 240 |
+
"step": 66
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.7010309278350515,
|
| 244 |
+
"grad_norm": 0.3139188289642334,
|
| 245 |
+
"learning_rate": 2e-05,
|
| 246 |
+
"loss": 0.6619,
|
| 247 |
+
"step": 68
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7216494845360825,
|
| 251 |
+
"grad_norm": 0.13998755812644958,
|
| 252 |
+
"learning_rate": 2e-05,
|
| 253 |
+
"loss": 1.2334,
|
| 254 |
+
"step": 70
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7422680412371134,
|
| 258 |
+
"grad_norm": 0.7787773013114929,
|
| 259 |
+
"learning_rate": 2e-05,
|
| 260 |
+
"loss": 1.425,
|
| 261 |
+
"step": 72
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7628865979381443,
|
| 265 |
+
"grad_norm": 2.650651693344116,
|
| 266 |
+
"learning_rate": 2e-05,
|
| 267 |
+
"loss": 1.5593,
|
| 268 |
+
"step": 74
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7835051546391752,
|
| 272 |
+
"grad_norm": 2.5681259632110596,
|
| 273 |
+
"learning_rate": 2e-05,
|
| 274 |
+
"loss": 0.8564,
|
| 275 |
+
"step": 76
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8041237113402062,
|
| 279 |
+
"grad_norm": 1.672986626625061,
|
| 280 |
+
"learning_rate": 2e-05,
|
| 281 |
+
"loss": 0.3764,
|
| 282 |
+
"step": 78
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8247422680412371,
|
| 286 |
+
"grad_norm": 1.4562022686004639,
|
| 287 |
+
"learning_rate": 2e-05,
|
| 288 |
+
"loss": 1.0597,
|
| 289 |
+
"step": 80
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.845360824742268,
|
| 293 |
+
"grad_norm": 1.849575161933899,
|
| 294 |
+
"learning_rate": 2e-05,
|
| 295 |
+
"loss": 0.6571,
|
| 296 |
+
"step": 82
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.865979381443299,
|
| 300 |
+
"grad_norm": 1.7122550010681152,
|
| 301 |
+
"learning_rate": 2e-05,
|
| 302 |
+
"loss": 0.7316,
|
| 303 |
+
"step": 84
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8865979381443299,
|
| 307 |
+
"grad_norm": 0.9788981080055237,
|
| 308 |
+
"learning_rate": 2e-05,
|
| 309 |
+
"loss": 1.8302,
|
| 310 |
+
"step": 86
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9072164948453608,
|
| 314 |
+
"grad_norm": 4.261756896972656,
|
| 315 |
+
"learning_rate": 2e-05,
|
| 316 |
+
"loss": 1.748,
|
| 317 |
+
"step": 88
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9278350515463918,
|
| 321 |
+
"grad_norm": 0.5180452466011047,
|
| 322 |
+
"learning_rate": 2e-05,
|
| 323 |
+
"loss": 0.6574,
|
| 324 |
+
"step": 90
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.9484536082474226,
|
| 328 |
+
"grad_norm": 3.8009090423583984,
|
| 329 |
+
"learning_rate": 2e-05,
|
| 330 |
+
"loss": 1.222,
|
| 331 |
+
"step": 92
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.9690721649484536,
|
| 335 |
+
"grad_norm": 2.206448793411255,
|
| 336 |
+
"learning_rate": 2e-05,
|
| 337 |
+
"loss": 2.1999,
|
| 338 |
+
"step": 94
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9896907216494846,
|
| 342 |
+
"grad_norm": 1.8258347511291504,
|
| 343 |
+
"learning_rate": 2e-05,
|
| 344 |
+
"loss": 1.6342,
|
| 345 |
+
"step": 96
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.0,
|
| 349 |
+
"step": 97,
|
| 350 |
+
"total_flos": 1.975033480085504e+16,
|
| 351 |
+
"train_loss": 1.1381244143259894,
|
| 352 |
+
"train_runtime": 347.3529,
|
| 353 |
+
"train_samples_per_second": 1.117,
|
| 354 |
+
"train_steps_per_second": 0.279
|
| 355 |
+
}
|
| 356 |
+
],
|
| 357 |
+
"logging_steps": 2,
|
| 358 |
+
"max_steps": 97,
|
| 359 |
+
"num_input_tokens_seen": 0,
|
| 360 |
+
"num_train_epochs": 1,
|
| 361 |
+
"save_steps": 500,
|
| 362 |
+
"stateful_callbacks": {
|
| 363 |
+
"TrainerControl": {
|
| 364 |
+
"args": {
|
| 365 |
+
"should_epoch_stop": false,
|
| 366 |
+
"should_evaluate": false,
|
| 367 |
+
"should_log": false,
|
| 368 |
+
"should_save": false,
|
| 369 |
+
"should_training_stop": false
|
| 370 |
+
},
|
| 371 |
+
"attributes": {}
|
| 372 |
+
}
|
| 373 |
+
},
|
| 374 |
+
"total_flos": 1.975033480085504e+16,
|
| 375 |
+
"train_batch_size": 1,
|
| 376 |
+
"trial_name": null,
|
| 377 |
+
"trial_params": null
|
| 378 |
+
}
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round10.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d75b434f434ac58ee7ff79a1192526027663a462f31003c61af666ca2c8197b
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round12.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e40c641acfb84bd93aec680098654d6533f26cea5a9317dba50857689945e350
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round15.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bea02c21b18787ce374ede742d2698c2221d15bd03cac84c80e8f25d54d2f3a3
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round17.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7534dde78b8b023855befc1292d21473ab56079172d82bf4f570b1cf7959e012
|
| 3 |
+
size 791578182
|
client_states_feddualMultipqfull2_NEWAensureOrth_Bcca_T05_freq10_bs4_saveoptim_lr2e-5_5e-5_sc132_4tasks_5rounds_fixitr97_T0125_decay099/5_client_model_round2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bccb9abfe6b71dff526a503d33d640fb761439901f9e58e425b441518d29bb8d
|
| 3 |
+
size 791576546
|