Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round10.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round12.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round15.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round17.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round2.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round20.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round5.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round7.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_trainer_state.json +392 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round10.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round12.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round15.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round17.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round2.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round20.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round5.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round7.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_trainer_state.json +392 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round10.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round12.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round15.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round17.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round2.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round20.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round5.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round7.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_trainer_state.json +392 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round10.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round12.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round15.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round17.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round2.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round20.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round5.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round7.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_trainer_state.json +392 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round10.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round12.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round15.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round17.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round2.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round20.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round5.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round7.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_trainer_state.json +392 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round10.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round12.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round15.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round17.pth +3 -0
- client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round2.pth +3 -0
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round10.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d7a4d0d6615d2510711f99c6b666e858a2598446d3263d37eed691a52c48cafa
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round12.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8467983a8faf53ac385f579b43f293895165ec3553aa52dcb326de610583d6a0
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round15.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b5bf18123b293dbbba941833be776184ede92ebbb3304e9365d5bf968ecc50b6
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round17.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:35940a8c8eddd6b7454570d0e6c61016292d9d2e5197647f71eb2ddcdb093c33
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea4a6fb5a88281d0acef0944c1cc603e502dfe0190a9f8f36296a7c9476d566a
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round20.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:468aaa38b2dbdf861f538b48b6969482f6d8913aeac15e921782031762a86c89
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round5.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:eed2f8b54154e62827fc7ccb130532f5449940421549b41b08b74e89ff8a3b4c
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_client_model_round7.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:81889dbecd6a4ea0784812af1ca62a08caa07cf01b024f7335fd93dd392681c9
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/0_trainer_state.json
ADDED
@@ -0,0 +1,392 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 100,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.02,
|
13 |
+
"grad_norm": 1.4507830142974854,
|
14 |
+
"learning_rate": 2e-05,
|
15 |
+
"loss": 0.2925,
|
16 |
+
"step": 2
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.04,
|
20 |
+
"grad_norm": 3.2782809734344482,
|
21 |
+
"learning_rate": 2e-05,
|
22 |
+
"loss": 0.5226,
|
23 |
+
"step": 4
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.06,
|
27 |
+
"grad_norm": 3.6665804386138916,
|
28 |
+
"learning_rate": 2e-05,
|
29 |
+
"loss": 0.4469,
|
30 |
+
"step": 6
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.08,
|
34 |
+
"grad_norm": 6.7659735679626465,
|
35 |
+
"learning_rate": 2e-05,
|
36 |
+
"loss": 0.6084,
|
37 |
+
"step": 8
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.1,
|
41 |
+
"grad_norm": 2.461778163909912,
|
42 |
+
"learning_rate": 2e-05,
|
43 |
+
"loss": 0.6173,
|
44 |
+
"step": 10
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.12,
|
48 |
+
"grad_norm": 0.8903517723083496,
|
49 |
+
"learning_rate": 2e-05,
|
50 |
+
"loss": 0.095,
|
51 |
+
"step": 12
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.14,
|
55 |
+
"grad_norm": 4.321805000305176,
|
56 |
+
"learning_rate": 2e-05,
|
57 |
+
"loss": 0.4503,
|
58 |
+
"step": 14
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.16,
|
62 |
+
"grad_norm": 2.4119670391082764,
|
63 |
+
"learning_rate": 2e-05,
|
64 |
+
"loss": 0.3114,
|
65 |
+
"step": 16
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.18,
|
69 |
+
"grad_norm": 4.858500003814697,
|
70 |
+
"learning_rate": 2e-05,
|
71 |
+
"loss": 0.3571,
|
72 |
+
"step": 18
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.2,
|
76 |
+
"grad_norm": 3.7888195514678955,
|
77 |
+
"learning_rate": 2e-05,
|
78 |
+
"loss": 0.8946,
|
79 |
+
"step": 20
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.22,
|
83 |
+
"grad_norm": 0.9685637950897217,
|
84 |
+
"learning_rate": 2e-05,
|
85 |
+
"loss": 0.1218,
|
86 |
+
"step": 22
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.24,
|
90 |
+
"grad_norm": 6.1121015548706055,
|
91 |
+
"learning_rate": 2e-05,
|
92 |
+
"loss": 0.5919,
|
93 |
+
"step": 24
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.26,
|
97 |
+
"grad_norm": 5.337240219116211,
|
98 |
+
"learning_rate": 2e-05,
|
99 |
+
"loss": 0.366,
|
100 |
+
"step": 26
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.28,
|
104 |
+
"grad_norm": 4.253252983093262,
|
105 |
+
"learning_rate": 2e-05,
|
106 |
+
"loss": 0.4007,
|
107 |
+
"step": 28
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.3,
|
111 |
+
"grad_norm": 8.691818237304688,
|
112 |
+
"learning_rate": 2e-05,
|
113 |
+
"loss": 0.9219,
|
114 |
+
"step": 30
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.32,
|
118 |
+
"grad_norm": 4.221813201904297,
|
119 |
+
"learning_rate": 2e-05,
|
120 |
+
"loss": 0.3515,
|
121 |
+
"step": 32
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.34,
|
125 |
+
"grad_norm": 6.260042667388916,
|
126 |
+
"learning_rate": 2e-05,
|
127 |
+
"loss": 0.8002,
|
128 |
+
"step": 34
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.36,
|
132 |
+
"grad_norm": 11.410113334655762,
|
133 |
+
"learning_rate": 2e-05,
|
134 |
+
"loss": 1.451,
|
135 |
+
"step": 36
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.38,
|
139 |
+
"grad_norm": 3.6883702278137207,
|
140 |
+
"learning_rate": 2e-05,
|
141 |
+
"loss": 0.3644,
|
142 |
+
"step": 38
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.4,
|
146 |
+
"grad_norm": 2.033090353012085,
|
147 |
+
"learning_rate": 2e-05,
|
148 |
+
"loss": 0.2821,
|
149 |
+
"step": 40
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.42,
|
153 |
+
"grad_norm": 6.383613109588623,
|
154 |
+
"learning_rate": 2e-05,
|
155 |
+
"loss": 1.392,
|
156 |
+
"step": 42
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.44,
|
160 |
+
"grad_norm": 2.7196388244628906,
|
161 |
+
"learning_rate": 2e-05,
|
162 |
+
"loss": 0.1335,
|
163 |
+
"step": 44
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.46,
|
167 |
+
"grad_norm": 9.112273216247559,
|
168 |
+
"learning_rate": 2e-05,
|
169 |
+
"loss": 1.9183,
|
170 |
+
"step": 46
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.48,
|
174 |
+
"grad_norm": 4.017404556274414,
|
175 |
+
"learning_rate": 2e-05,
|
176 |
+
"loss": 0.5077,
|
177 |
+
"step": 48
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.5,
|
181 |
+
"grad_norm": 0.14292454719543457,
|
182 |
+
"learning_rate": 2e-05,
|
183 |
+
"loss": 0.323,
|
184 |
+
"step": 50
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.52,
|
188 |
+
"grad_norm": 2.870044469833374,
|
189 |
+
"learning_rate": 2e-05,
|
190 |
+
"loss": 0.679,
|
191 |
+
"step": 52
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.54,
|
195 |
+
"grad_norm": 7.838882923126221,
|
196 |
+
"learning_rate": 2e-05,
|
197 |
+
"loss": 0.5556,
|
198 |
+
"step": 54
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.56,
|
202 |
+
"grad_norm": 1.5267337560653687,
|
203 |
+
"learning_rate": 2e-05,
|
204 |
+
"loss": 0.6559,
|
205 |
+
"step": 56
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.58,
|
209 |
+
"grad_norm": 4.8004374504089355,
|
210 |
+
"learning_rate": 2e-05,
|
211 |
+
"loss": 1.1925,
|
212 |
+
"step": 58
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.6,
|
216 |
+
"grad_norm": 0.9164693355560303,
|
217 |
+
"learning_rate": 2e-05,
|
218 |
+
"loss": 0.3563,
|
219 |
+
"step": 60
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.62,
|
223 |
+
"grad_norm": 10.06078815460205,
|
224 |
+
"learning_rate": 2e-05,
|
225 |
+
"loss": 1.1676,
|
226 |
+
"step": 62
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.64,
|
230 |
+
"grad_norm": 4.685103416442871,
|
231 |
+
"learning_rate": 2e-05,
|
232 |
+
"loss": 0.4408,
|
233 |
+
"step": 64
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.66,
|
237 |
+
"grad_norm": 2.2955524921417236,
|
238 |
+
"learning_rate": 2e-05,
|
239 |
+
"loss": 0.1538,
|
240 |
+
"step": 66
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.68,
|
244 |
+
"grad_norm": 2.193854331970215,
|
245 |
+
"learning_rate": 2e-05,
|
246 |
+
"loss": 0.8281,
|
247 |
+
"step": 68
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.7,
|
251 |
+
"grad_norm": 0.9019300937652588,
|
252 |
+
"learning_rate": 2e-05,
|
253 |
+
"loss": 1.1189,
|
254 |
+
"step": 70
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.72,
|
258 |
+
"grad_norm": 10.440985679626465,
|
259 |
+
"learning_rate": 2e-05,
|
260 |
+
"loss": 2.1768,
|
261 |
+
"step": 72
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.74,
|
265 |
+
"grad_norm": 2.911909818649292,
|
266 |
+
"learning_rate": 2e-05,
|
267 |
+
"loss": 0.7279,
|
268 |
+
"step": 74
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.76,
|
272 |
+
"grad_norm": 0.03667999058961868,
|
273 |
+
"learning_rate": 2e-05,
|
274 |
+
"loss": 0.115,
|
275 |
+
"step": 76
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.78,
|
279 |
+
"grad_norm": 2.2721805572509766,
|
280 |
+
"learning_rate": 2e-05,
|
281 |
+
"loss": 0.1444,
|
282 |
+
"step": 78
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.8,
|
286 |
+
"grad_norm": 2.097175121307373,
|
287 |
+
"learning_rate": 2e-05,
|
288 |
+
"loss": 0.2108,
|
289 |
+
"step": 80
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.82,
|
293 |
+
"grad_norm": 0.7106596231460571,
|
294 |
+
"learning_rate": 2e-05,
|
295 |
+
"loss": 0.0699,
|
296 |
+
"step": 82
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.84,
|
300 |
+
"grad_norm": 2.563626766204834,
|
301 |
+
"learning_rate": 2e-05,
|
302 |
+
"loss": 0.1482,
|
303 |
+
"step": 84
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.86,
|
307 |
+
"grad_norm": 0.34837377071380615,
|
308 |
+
"learning_rate": 2e-05,
|
309 |
+
"loss": 0.0672,
|
310 |
+
"step": 86
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.88,
|
314 |
+
"grad_norm": 0.41047564148902893,
|
315 |
+
"learning_rate": 2e-05,
|
316 |
+
"loss": 0.3899,
|
317 |
+
"step": 88
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.9,
|
321 |
+
"grad_norm": 0.7772120833396912,
|
322 |
+
"learning_rate": 2e-05,
|
323 |
+
"loss": 0.0882,
|
324 |
+
"step": 90
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.92,
|
328 |
+
"grad_norm": 0.9385977387428284,
|
329 |
+
"learning_rate": 2e-05,
|
330 |
+
"loss": 1.0358,
|
331 |
+
"step": 92
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.94,
|
335 |
+
"grad_norm": 6.794859886169434,
|
336 |
+
"learning_rate": 2e-05,
|
337 |
+
"loss": 1.1388,
|
338 |
+
"step": 94
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.96,
|
342 |
+
"grad_norm": 8.23221492767334,
|
343 |
+
"learning_rate": 2e-05,
|
344 |
+
"loss": 0.3678,
|
345 |
+
"step": 96
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.98,
|
349 |
+
"grad_norm": 0.18338899314403534,
|
350 |
+
"learning_rate": 2e-05,
|
351 |
+
"loss": 0.2155,
|
352 |
+
"step": 98
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 1.0,
|
356 |
+
"grad_norm": 0.3215850591659546,
|
357 |
+
"learning_rate": 2e-05,
|
358 |
+
"loss": 0.0276,
|
359 |
+
"step": 100
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 1.0,
|
363 |
+
"step": 100,
|
364 |
+
"total_flos": 2053446886752256.0,
|
365 |
+
"train_loss": 0.5718875336647034,
|
366 |
+
"train_runtime": 100.3201,
|
367 |
+
"train_samples_per_second": 3.987,
|
368 |
+
"train_steps_per_second": 0.997
|
369 |
+
}
|
370 |
+
],
|
371 |
+
"logging_steps": 2,
|
372 |
+
"max_steps": 100,
|
373 |
+
"num_input_tokens_seen": 0,
|
374 |
+
"num_train_epochs": 1,
|
375 |
+
"save_steps": 500,
|
376 |
+
"stateful_callbacks": {
|
377 |
+
"TrainerControl": {
|
378 |
+
"args": {
|
379 |
+
"should_epoch_stop": false,
|
380 |
+
"should_evaluate": false,
|
381 |
+
"should_log": false,
|
382 |
+
"should_save": false,
|
383 |
+
"should_training_stop": false
|
384 |
+
},
|
385 |
+
"attributes": {}
|
386 |
+
}
|
387 |
+
},
|
388 |
+
"total_flos": 2053446886752256.0,
|
389 |
+
"train_batch_size": 1,
|
390 |
+
"trial_name": null,
|
391 |
+
"trial_params": null
|
392 |
+
}
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round10.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b9983a1b6c779a4d52cdc627f04f956d61efac456204bcc4dabada42d9a3dfdd
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round12.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f2de129e62beeb4971790b5e1f46c5a709dcf59503e9475b5787bf0cd581368e
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round15.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:24ed512b6dcd1a66d95d053edadd3e3fbdf255ba36752a7ade6a545f7a8c1654
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round17.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a4219da9ea85c1442677e975ed2d3f527a357245f0998ca76a59bd078777e3fd
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c1ceea20bd9bb93c4ec0fd7ee4511cfb48fbb6ab8dac82e3d4240b59dba7c865
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round20.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fac5c439ec99857d8e543ff416a2fd33e4648cf7a54164d4ff91a2487ca1202e
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round5.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6ca3cc56d5d8ef6bf8ceb6673227c86c6fdfa5c11342ee85b31de747c9ee1296
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_client_model_round7.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bf892a9fd738f4567b9929fca404c0a7a73c02558fbe6ed6dc4303c2e51f42b8
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/1_trainer_state.json
ADDED
@@ -0,0 +1,392 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 100,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.02,
|
13 |
+
"grad_norm": 4.540980339050293,
|
14 |
+
"learning_rate": 2e-05,
|
15 |
+
"loss": 0.422,
|
16 |
+
"step": 2
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.04,
|
20 |
+
"grad_norm": 0.08287880569696426,
|
21 |
+
"learning_rate": 2e-05,
|
22 |
+
"loss": 0.0048,
|
23 |
+
"step": 4
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.06,
|
27 |
+
"grad_norm": 3.329127788543701,
|
28 |
+
"learning_rate": 2e-05,
|
29 |
+
"loss": 0.2607,
|
30 |
+
"step": 6
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.08,
|
34 |
+
"grad_norm": 0.11668016016483307,
|
35 |
+
"learning_rate": 2e-05,
|
36 |
+
"loss": 0.0061,
|
37 |
+
"step": 8
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.1,
|
41 |
+
"grad_norm": 1.266418218612671,
|
42 |
+
"learning_rate": 2e-05,
|
43 |
+
"loss": 0.0361,
|
44 |
+
"step": 10
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.12,
|
48 |
+
"grad_norm": 1.0858792066574097,
|
49 |
+
"learning_rate": 2e-05,
|
50 |
+
"loss": 0.0498,
|
51 |
+
"step": 12
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.14,
|
55 |
+
"grad_norm": 1.4955309629440308,
|
56 |
+
"learning_rate": 2e-05,
|
57 |
+
"loss": 0.0903,
|
58 |
+
"step": 14
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.16,
|
62 |
+
"grad_norm": 3.658975839614868,
|
63 |
+
"learning_rate": 2e-05,
|
64 |
+
"loss": 0.4419,
|
65 |
+
"step": 16
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.18,
|
69 |
+
"grad_norm": 0.029739608988165855,
|
70 |
+
"learning_rate": 2e-05,
|
71 |
+
"loss": 0.0057,
|
72 |
+
"step": 18
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.2,
|
76 |
+
"grad_norm": 0.04279458895325661,
|
77 |
+
"learning_rate": 2e-05,
|
78 |
+
"loss": 0.0027,
|
79 |
+
"step": 20
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.22,
|
83 |
+
"grad_norm": 1.1662931442260742,
|
84 |
+
"learning_rate": 2e-05,
|
85 |
+
"loss": 0.0642,
|
86 |
+
"step": 22
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.24,
|
90 |
+
"grad_norm": 1.0573233366012573,
|
91 |
+
"learning_rate": 2e-05,
|
92 |
+
"loss": 0.0731,
|
93 |
+
"step": 24
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.26,
|
97 |
+
"grad_norm": 4.016772747039795,
|
98 |
+
"learning_rate": 2e-05,
|
99 |
+
"loss": 0.3845,
|
100 |
+
"step": 26
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.28,
|
104 |
+
"grad_norm": 0.2739083766937256,
|
105 |
+
"learning_rate": 2e-05,
|
106 |
+
"loss": 0.0173,
|
107 |
+
"step": 28
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.3,
|
111 |
+
"grad_norm": 0.004112154711037874,
|
112 |
+
"learning_rate": 2e-05,
|
113 |
+
"loss": 0.5049,
|
114 |
+
"step": 30
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.32,
|
118 |
+
"grad_norm": 1.817938208580017,
|
119 |
+
"learning_rate": 2e-05,
|
120 |
+
"loss": 0.1084,
|
121 |
+
"step": 32
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.34,
|
125 |
+
"grad_norm": 0.040277149528265,
|
126 |
+
"learning_rate": 2e-05,
|
127 |
+
"loss": 0.0015,
|
128 |
+
"step": 34
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.36,
|
132 |
+
"grad_norm": 0.03433386608958244,
|
133 |
+
"learning_rate": 2e-05,
|
134 |
+
"loss": 0.0028,
|
135 |
+
"step": 36
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.38,
|
139 |
+
"grad_norm": 3.3714048862457275,
|
140 |
+
"learning_rate": 2e-05,
|
141 |
+
"loss": 0.1396,
|
142 |
+
"step": 38
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.4,
|
146 |
+
"grad_norm": 0.03902699798345566,
|
147 |
+
"learning_rate": 2e-05,
|
148 |
+
"loss": 0.0027,
|
149 |
+
"step": 40
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.42,
|
153 |
+
"grad_norm": 0.17097237706184387,
|
154 |
+
"learning_rate": 2e-05,
|
155 |
+
"loss": 0.006,
|
156 |
+
"step": 42
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.44,
|
160 |
+
"grad_norm": 0.004333522170782089,
|
161 |
+
"learning_rate": 2e-05,
|
162 |
+
"loss": 0.0081,
|
163 |
+
"step": 44
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.46,
|
167 |
+
"grad_norm": 0.13571485877037048,
|
168 |
+
"learning_rate": 2e-05,
|
169 |
+
"loss": 0.006,
|
170 |
+
"step": 46
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.48,
|
174 |
+
"grad_norm": 0.31877508759498596,
|
175 |
+
"learning_rate": 2e-05,
|
176 |
+
"loss": 0.0122,
|
177 |
+
"step": 48
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.5,
|
181 |
+
"grad_norm": 0.09019943326711655,
|
182 |
+
"learning_rate": 2e-05,
|
183 |
+
"loss": 0.0051,
|
184 |
+
"step": 50
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.52,
|
188 |
+
"grad_norm": 10.819501876831055,
|
189 |
+
"learning_rate": 2e-05,
|
190 |
+
"loss": 1.0651,
|
191 |
+
"step": 52
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.54,
|
195 |
+
"grad_norm": 0.5622828602790833,
|
196 |
+
"learning_rate": 2e-05,
|
197 |
+
"loss": 0.0183,
|
198 |
+
"step": 54
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.56,
|
202 |
+
"grad_norm": 0.3623190224170685,
|
203 |
+
"learning_rate": 2e-05,
|
204 |
+
"loss": 0.0462,
|
205 |
+
"step": 56
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.58,
|
209 |
+
"grad_norm": 0.40045422315597534,
|
210 |
+
"learning_rate": 2e-05,
|
211 |
+
"loss": 0.1874,
|
212 |
+
"step": 58
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.6,
|
216 |
+
"grad_norm": 0.0034316980745643377,
|
217 |
+
"learning_rate": 2e-05,
|
218 |
+
"loss": 0.0095,
|
219 |
+
"step": 60
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.62,
|
223 |
+
"grad_norm": 0.01208802405744791,
|
224 |
+
"learning_rate": 2e-05,
|
225 |
+
"loss": 0.0055,
|
226 |
+
"step": 62
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.64,
|
230 |
+
"grad_norm": 0.01345849223434925,
|
231 |
+
"learning_rate": 2e-05,
|
232 |
+
"loss": 0.0033,
|
233 |
+
"step": 64
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.66,
|
237 |
+
"grad_norm": 0.6658453941345215,
|
238 |
+
"learning_rate": 2e-05,
|
239 |
+
"loss": 0.0433,
|
240 |
+
"step": 66
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.68,
|
244 |
+
"grad_norm": 2.384672164916992,
|
245 |
+
"learning_rate": 2e-05,
|
246 |
+
"loss": 0.2239,
|
247 |
+
"step": 68
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.7,
|
251 |
+
"grad_norm": 0.07835771143436432,
|
252 |
+
"learning_rate": 2e-05,
|
253 |
+
"loss": 0.0073,
|
254 |
+
"step": 70
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.72,
|
258 |
+
"grad_norm": 1.4132821559906006,
|
259 |
+
"learning_rate": 2e-05,
|
260 |
+
"loss": 0.0383,
|
261 |
+
"step": 72
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.74,
|
265 |
+
"grad_norm": 0.9961208701133728,
|
266 |
+
"learning_rate": 2e-05,
|
267 |
+
"loss": 0.0747,
|
268 |
+
"step": 74
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.76,
|
272 |
+
"grad_norm": 0.9463048577308655,
|
273 |
+
"learning_rate": 2e-05,
|
274 |
+
"loss": 0.0433,
|
275 |
+
"step": 76
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.78,
|
279 |
+
"grad_norm": 0.01034417562186718,
|
280 |
+
"learning_rate": 2e-05,
|
281 |
+
"loss": 0.0328,
|
282 |
+
"step": 78
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.8,
|
286 |
+
"grad_norm": 0.0005425452254712582,
|
287 |
+
"learning_rate": 2e-05,
|
288 |
+
"loss": 0.3519,
|
289 |
+
"step": 80
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.82,
|
293 |
+
"grad_norm": 0.04632725194096565,
|
294 |
+
"learning_rate": 2e-05,
|
295 |
+
"loss": 0.0369,
|
296 |
+
"step": 82
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.84,
|
300 |
+
"grad_norm": 1.2183465957641602,
|
301 |
+
"learning_rate": 2e-05,
|
302 |
+
"loss": 0.0654,
|
303 |
+
"step": 84
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.86,
|
307 |
+
"grad_norm": 0.009153845719993114,
|
308 |
+
"learning_rate": 2e-05,
|
309 |
+
"loss": 0.0014,
|
310 |
+
"step": 86
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.88,
|
314 |
+
"grad_norm": 0.8759830594062805,
|
315 |
+
"learning_rate": 2e-05,
|
316 |
+
"loss": 0.0918,
|
317 |
+
"step": 88
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.9,
|
321 |
+
"grad_norm": 0.031464505940675735,
|
322 |
+
"learning_rate": 2e-05,
|
323 |
+
"loss": 0.0014,
|
324 |
+
"step": 90
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.92,
|
328 |
+
"grad_norm": 0.038530562072992325,
|
329 |
+
"learning_rate": 2e-05,
|
330 |
+
"loss": 0.0023,
|
331 |
+
"step": 92
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.94,
|
335 |
+
"grad_norm": 0.04050149768590927,
|
336 |
+
"learning_rate": 2e-05,
|
337 |
+
"loss": 0.0215,
|
338 |
+
"step": 94
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.96,
|
342 |
+
"grad_norm": 0.009601338766515255,
|
343 |
+
"learning_rate": 2e-05,
|
344 |
+
"loss": 0.001,
|
345 |
+
"step": 96
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.98,
|
349 |
+
"grad_norm": 0.02644410915672779,
|
350 |
+
"learning_rate": 2e-05,
|
351 |
+
"loss": 0.0018,
|
352 |
+
"step": 98
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 1.0,
|
356 |
+
"grad_norm": 13.793920516967773,
|
357 |
+
"learning_rate": 2e-05,
|
358 |
+
"loss": 0.8695,
|
359 |
+
"step": 100
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 1.0,
|
363 |
+
"step": 100,
|
364 |
+
"total_flos": 2069576665792512.0,
|
365 |
+
"train_loss": 0.11801103919744492,
|
366 |
+
"train_runtime": 100.2591,
|
367 |
+
"train_samples_per_second": 3.99,
|
368 |
+
"train_steps_per_second": 0.997
|
369 |
+
}
|
370 |
+
],
|
371 |
+
"logging_steps": 2,
|
372 |
+
"max_steps": 100,
|
373 |
+
"num_input_tokens_seen": 0,
|
374 |
+
"num_train_epochs": 1,
|
375 |
+
"save_steps": 500,
|
376 |
+
"stateful_callbacks": {
|
377 |
+
"TrainerControl": {
|
378 |
+
"args": {
|
379 |
+
"should_epoch_stop": false,
|
380 |
+
"should_evaluate": false,
|
381 |
+
"should_log": false,
|
382 |
+
"should_save": false,
|
383 |
+
"should_training_stop": false
|
384 |
+
},
|
385 |
+
"attributes": {}
|
386 |
+
}
|
387 |
+
},
|
388 |
+
"total_flos": 2069576665792512.0,
|
389 |
+
"train_batch_size": 1,
|
390 |
+
"trial_name": null,
|
391 |
+
"trial_params": null
|
392 |
+
}
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round10.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:03582a65c5f942f662a8eafdf27eccb06167fe68a235376a577c83b61464520f
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round12.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7f25cc37e7b0ba30193079fe56dc4c3aa9b011c90f9b100c4edb8af121e24a47
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round15.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3e73eded9bd524f1dce2d82e0dcd45a995aec36460f360989c7f0a3633f9064b
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round17.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4f015a9439fe81810ee686e3def97f0a5628eb03eeb6647f2361e14acb54c6f1
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6f1d2a8a90b6e9f82b7cd6b9d41f2426100a7a948c8be73b9cd03822077c3984
|
3 |
+
size 395786922
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round20.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ff05ff0a2fd64497a4dc87555b334386e1340bf51e0cb26b9563e5605c8b120e
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round5.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:abb0d68b551454c4d09eebc6d9d9cedb828570c4d6224ed8d949afef34928fd5
|
3 |
+
size 395786922
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_client_model_round7.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2e31665b8c7d78dc560352ba149d087420a03bb9929818edc19a4016a5f1b8ba
|
3 |
+
size 395786922
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/2_trainer_state.json
ADDED
@@ -0,0 +1,392 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 100,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.02,
|
13 |
+
"grad_norm": 6.307283401489258,
|
14 |
+
"learning_rate": 2e-05,
|
15 |
+
"loss": 1.5954,
|
16 |
+
"step": 2
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.04,
|
20 |
+
"grad_norm": 1.8216760158538818,
|
21 |
+
"learning_rate": 2e-05,
|
22 |
+
"loss": 0.278,
|
23 |
+
"step": 4
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.06,
|
27 |
+
"grad_norm": 7.7118024826049805,
|
28 |
+
"learning_rate": 2e-05,
|
29 |
+
"loss": 1.2105,
|
30 |
+
"step": 6
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.08,
|
34 |
+
"grad_norm": 6.2828192710876465,
|
35 |
+
"learning_rate": 2e-05,
|
36 |
+
"loss": 1.0291,
|
37 |
+
"step": 8
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.1,
|
41 |
+
"grad_norm": 1.8280245065689087,
|
42 |
+
"learning_rate": 2e-05,
|
43 |
+
"loss": 0.3982,
|
44 |
+
"step": 10
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.12,
|
48 |
+
"grad_norm": 0.9868451356887817,
|
49 |
+
"learning_rate": 2e-05,
|
50 |
+
"loss": 1.2771,
|
51 |
+
"step": 12
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.14,
|
55 |
+
"grad_norm": 1.3482744693756104,
|
56 |
+
"learning_rate": 2e-05,
|
57 |
+
"loss": 0.6548,
|
58 |
+
"step": 14
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.16,
|
62 |
+
"grad_norm": 1.1375393867492676,
|
63 |
+
"learning_rate": 2e-05,
|
64 |
+
"loss": 0.6873,
|
65 |
+
"step": 16
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.18,
|
69 |
+
"grad_norm": 1.463499903678894,
|
70 |
+
"learning_rate": 2e-05,
|
71 |
+
"loss": 0.6526,
|
72 |
+
"step": 18
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.2,
|
76 |
+
"grad_norm": 5.554112911224365,
|
77 |
+
"learning_rate": 2e-05,
|
78 |
+
"loss": 1.2926,
|
79 |
+
"step": 20
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.22,
|
83 |
+
"grad_norm": 1.0732632875442505,
|
84 |
+
"learning_rate": 2e-05,
|
85 |
+
"loss": 0.461,
|
86 |
+
"step": 22
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.24,
|
90 |
+
"grad_norm": 2.5138278007507324,
|
91 |
+
"learning_rate": 2e-05,
|
92 |
+
"loss": 0.7348,
|
93 |
+
"step": 24
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.26,
|
97 |
+
"grad_norm": 2.534879446029663,
|
98 |
+
"learning_rate": 2e-05,
|
99 |
+
"loss": 0.8571,
|
100 |
+
"step": 26
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.28,
|
104 |
+
"grad_norm": 2.698505163192749,
|
105 |
+
"learning_rate": 2e-05,
|
106 |
+
"loss": 0.521,
|
107 |
+
"step": 28
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.3,
|
111 |
+
"grad_norm": 1.4479676485061646,
|
112 |
+
"learning_rate": 2e-05,
|
113 |
+
"loss": 0.3902,
|
114 |
+
"step": 30
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.32,
|
118 |
+
"grad_norm": 1.2816400527954102,
|
119 |
+
"learning_rate": 2e-05,
|
120 |
+
"loss": 0.4518,
|
121 |
+
"step": 32
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.34,
|
125 |
+
"grad_norm": 5.400059223175049,
|
126 |
+
"learning_rate": 2e-05,
|
127 |
+
"loss": 0.9999,
|
128 |
+
"step": 34
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.36,
|
132 |
+
"grad_norm": 2.5884220600128174,
|
133 |
+
"learning_rate": 2e-05,
|
134 |
+
"loss": 0.3547,
|
135 |
+
"step": 36
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.38,
|
139 |
+
"grad_norm": 2.3713040351867676,
|
140 |
+
"learning_rate": 2e-05,
|
141 |
+
"loss": 0.7269,
|
142 |
+
"step": 38
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.4,
|
146 |
+
"grad_norm": 0.25373563170433044,
|
147 |
+
"learning_rate": 2e-05,
|
148 |
+
"loss": 0.0746,
|
149 |
+
"step": 40
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.42,
|
153 |
+
"grad_norm": 4.246715068817139,
|
154 |
+
"learning_rate": 2e-05,
|
155 |
+
"loss": 0.6696,
|
156 |
+
"step": 42
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.44,
|
160 |
+
"grad_norm": 1.6761362552642822,
|
161 |
+
"learning_rate": 2e-05,
|
162 |
+
"loss": 0.6276,
|
163 |
+
"step": 44
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.46,
|
167 |
+
"grad_norm": 3.399017572402954,
|
168 |
+
"learning_rate": 2e-05,
|
169 |
+
"loss": 0.9067,
|
170 |
+
"step": 46
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.48,
|
174 |
+
"grad_norm": 0.692711591720581,
|
175 |
+
"learning_rate": 2e-05,
|
176 |
+
"loss": 0.1943,
|
177 |
+
"step": 48
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.5,
|
181 |
+
"grad_norm": 1.116909146308899,
|
182 |
+
"learning_rate": 2e-05,
|
183 |
+
"loss": 0.2336,
|
184 |
+
"step": 50
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.52,
|
188 |
+
"grad_norm": 1.456255555152893,
|
189 |
+
"learning_rate": 2e-05,
|
190 |
+
"loss": 0.1465,
|
191 |
+
"step": 52
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.54,
|
195 |
+
"grad_norm": 0.06800251454114914,
|
196 |
+
"learning_rate": 2e-05,
|
197 |
+
"loss": 0.2881,
|
198 |
+
"step": 54
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.56,
|
202 |
+
"grad_norm": 4.372195720672607,
|
203 |
+
"learning_rate": 2e-05,
|
204 |
+
"loss": 0.6529,
|
205 |
+
"step": 56
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.58,
|
209 |
+
"grad_norm": 4.281543254852295,
|
210 |
+
"learning_rate": 2e-05,
|
211 |
+
"loss": 1.313,
|
212 |
+
"step": 58
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.6,
|
216 |
+
"grad_norm": 3.5601418018341064,
|
217 |
+
"learning_rate": 2e-05,
|
218 |
+
"loss": 0.3216,
|
219 |
+
"step": 60
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.62,
|
223 |
+
"grad_norm": 1.4214577674865723,
|
224 |
+
"learning_rate": 2e-05,
|
225 |
+
"loss": 0.1358,
|
226 |
+
"step": 62
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.64,
|
230 |
+
"grad_norm": 2.7352545261383057,
|
231 |
+
"learning_rate": 2e-05,
|
232 |
+
"loss": 0.4594,
|
233 |
+
"step": 64
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.66,
|
237 |
+
"grad_norm": 3.6488208770751953,
|
238 |
+
"learning_rate": 2e-05,
|
239 |
+
"loss": 1.1297,
|
240 |
+
"step": 66
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.68,
|
244 |
+
"grad_norm": 0.18936391174793243,
|
245 |
+
"learning_rate": 2e-05,
|
246 |
+
"loss": 0.2297,
|
247 |
+
"step": 68
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.7,
|
251 |
+
"grad_norm": 0.24005119502544403,
|
252 |
+
"learning_rate": 2e-05,
|
253 |
+
"loss": 1.2341,
|
254 |
+
"step": 70
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.72,
|
258 |
+
"grad_norm": 0.6526658535003662,
|
259 |
+
"learning_rate": 2e-05,
|
260 |
+
"loss": 0.4384,
|
261 |
+
"step": 72
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.74,
|
265 |
+
"grad_norm": 1.2854243516921997,
|
266 |
+
"learning_rate": 2e-05,
|
267 |
+
"loss": 0.2658,
|
268 |
+
"step": 74
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.76,
|
272 |
+
"grad_norm": 1.8762297630310059,
|
273 |
+
"learning_rate": 2e-05,
|
274 |
+
"loss": 0.5192,
|
275 |
+
"step": 76
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.78,
|
279 |
+
"grad_norm": 8.563094139099121,
|
280 |
+
"learning_rate": 2e-05,
|
281 |
+
"loss": 0.3001,
|
282 |
+
"step": 78
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.8,
|
286 |
+
"grad_norm": 0.5608336329460144,
|
287 |
+
"learning_rate": 2e-05,
|
288 |
+
"loss": 0.2274,
|
289 |
+
"step": 80
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.82,
|
293 |
+
"grad_norm": 0.9091539978981018,
|
294 |
+
"learning_rate": 2e-05,
|
295 |
+
"loss": 0.2695,
|
296 |
+
"step": 82
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.84,
|
300 |
+
"grad_norm": 1.8625966310501099,
|
301 |
+
"learning_rate": 2e-05,
|
302 |
+
"loss": 0.2104,
|
303 |
+
"step": 84
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.86,
|
307 |
+
"grad_norm": 3.4447784423828125,
|
308 |
+
"learning_rate": 2e-05,
|
309 |
+
"loss": 0.5892,
|
310 |
+
"step": 86
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.88,
|
314 |
+
"grad_norm": 0.39084768295288086,
|
315 |
+
"learning_rate": 2e-05,
|
316 |
+
"loss": 0.2646,
|
317 |
+
"step": 88
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.9,
|
321 |
+
"grad_norm": 4.247977256774902,
|
322 |
+
"learning_rate": 2e-05,
|
323 |
+
"loss": 0.3688,
|
324 |
+
"step": 90
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.92,
|
328 |
+
"grad_norm": 0.8638454079627991,
|
329 |
+
"learning_rate": 2e-05,
|
330 |
+
"loss": 2.1514,
|
331 |
+
"step": 92
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.94,
|
335 |
+
"grad_norm": 0.26021599769592285,
|
336 |
+
"learning_rate": 2e-05,
|
337 |
+
"loss": 0.0341,
|
338 |
+
"step": 94
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.96,
|
342 |
+
"grad_norm": 2.017536163330078,
|
343 |
+
"learning_rate": 2e-05,
|
344 |
+
"loss": 0.2675,
|
345 |
+
"step": 96
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.98,
|
349 |
+
"grad_norm": 2.068211793899536,
|
350 |
+
"learning_rate": 2e-05,
|
351 |
+
"loss": 0.256,
|
352 |
+
"step": 98
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 1.0,
|
356 |
+
"grad_norm": 6.009552478790283,
|
357 |
+
"learning_rate": 2e-05,
|
358 |
+
"loss": 0.8954,
|
359 |
+
"step": 100
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 1.0,
|
363 |
+
"step": 100,
|
364 |
+
"total_flos": 4915713315700736.0,
|
365 |
+
"train_loss": 0.6049574661254883,
|
366 |
+
"train_runtime": 165.4178,
|
367 |
+
"train_samples_per_second": 2.418,
|
368 |
+
"train_steps_per_second": 0.605
|
369 |
+
}
|
370 |
+
],
|
371 |
+
"logging_steps": 2,
|
372 |
+
"max_steps": 100,
|
373 |
+
"num_input_tokens_seen": 0,
|
374 |
+
"num_train_epochs": 1,
|
375 |
+
"save_steps": 500,
|
376 |
+
"stateful_callbacks": {
|
377 |
+
"TrainerControl": {
|
378 |
+
"args": {
|
379 |
+
"should_epoch_stop": false,
|
380 |
+
"should_evaluate": false,
|
381 |
+
"should_log": false,
|
382 |
+
"should_save": false,
|
383 |
+
"should_training_stop": false
|
384 |
+
},
|
385 |
+
"attributes": {}
|
386 |
+
}
|
387 |
+
},
|
388 |
+
"total_flos": 4915713315700736.0,
|
389 |
+
"train_batch_size": 1,
|
390 |
+
"trial_name": null,
|
391 |
+
"trial_params": null
|
392 |
+
}
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round10.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e8ec9a76aa2a365afb2403325260e12dec27ad4b984f60c9e7f5a9f76ebbdad1
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round12.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0a23a2ed8e03fd3a4b21c1fea3e90f4eb1c0a3a99bedee884c388ca90b15211
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round15.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70680ac43b83326240e7835655831bc4ceaf9f0de955840f5c6d0c1f4afcc2f1
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round17.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8e4b81455da8eff19455dde4b74de4297a754f18d10b63e691a538c6b9946a3f
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:af9b733c0813536ef08508eb156501dcb5bd61a3091f43cb3f3f6010fe4b4917
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round20.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7e03eca3be62c25b2a2aa30e419a67ca025067a04b2c6b04bf385daaae97ddbd
|
3 |
+
size 184221358
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round5.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d3787e3f8abd12f49c6d80d3cdcdede3a2f53808976bc36599bbc4428eea38f7
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_client_model_round7.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a6adb60b55506bfdd98859d59018a85cefb09d6e024b8fa128b3beb888dc5315
|
3 |
+
size 184220842
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/3_trainer_state.json
ADDED
@@ -0,0 +1,392 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 100,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.02,
|
13 |
+
"grad_norm": 2.0921337604522705,
|
14 |
+
"learning_rate": 2e-05,
|
15 |
+
"loss": 0.6894,
|
16 |
+
"step": 2
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.04,
|
20 |
+
"grad_norm": 4.510809421539307,
|
21 |
+
"learning_rate": 2e-05,
|
22 |
+
"loss": 0.6862,
|
23 |
+
"step": 4
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.06,
|
27 |
+
"grad_norm": 4.531048774719238,
|
28 |
+
"learning_rate": 2e-05,
|
29 |
+
"loss": 1.2015,
|
30 |
+
"step": 6
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.08,
|
34 |
+
"grad_norm": 3.931663751602173,
|
35 |
+
"learning_rate": 2e-05,
|
36 |
+
"loss": 0.5979,
|
37 |
+
"step": 8
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.1,
|
41 |
+
"grad_norm": 6.292150020599365,
|
42 |
+
"learning_rate": 2e-05,
|
43 |
+
"loss": 1.058,
|
44 |
+
"step": 10
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.12,
|
48 |
+
"grad_norm": 4.606175422668457,
|
49 |
+
"learning_rate": 2e-05,
|
50 |
+
"loss": 0.8743,
|
51 |
+
"step": 12
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.14,
|
55 |
+
"grad_norm": 3.087507963180542,
|
56 |
+
"learning_rate": 2e-05,
|
57 |
+
"loss": 0.6243,
|
58 |
+
"step": 14
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.16,
|
62 |
+
"grad_norm": 8.593318939208984,
|
63 |
+
"learning_rate": 2e-05,
|
64 |
+
"loss": 0.4402,
|
65 |
+
"step": 16
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.18,
|
69 |
+
"grad_norm": 5.68522834777832,
|
70 |
+
"learning_rate": 2e-05,
|
71 |
+
"loss": 1.3279,
|
72 |
+
"step": 18
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.2,
|
76 |
+
"grad_norm": 1.9175809621810913,
|
77 |
+
"learning_rate": 2e-05,
|
78 |
+
"loss": 0.7962,
|
79 |
+
"step": 20
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.22,
|
83 |
+
"grad_norm": 8.656201362609863,
|
84 |
+
"learning_rate": 2e-05,
|
85 |
+
"loss": 2.0415,
|
86 |
+
"step": 22
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.24,
|
90 |
+
"grad_norm": 1.99509859085083,
|
91 |
+
"learning_rate": 2e-05,
|
92 |
+
"loss": 0.2472,
|
93 |
+
"step": 24
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.26,
|
97 |
+
"grad_norm": 7.396963119506836,
|
98 |
+
"learning_rate": 2e-05,
|
99 |
+
"loss": 0.8892,
|
100 |
+
"step": 26
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.28,
|
104 |
+
"grad_norm": 5.109688758850098,
|
105 |
+
"learning_rate": 2e-05,
|
106 |
+
"loss": 0.4696,
|
107 |
+
"step": 28
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.3,
|
111 |
+
"grad_norm": 3.3529930114746094,
|
112 |
+
"learning_rate": 2e-05,
|
113 |
+
"loss": 0.6545,
|
114 |
+
"step": 30
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.32,
|
118 |
+
"grad_norm": 1.4711360931396484,
|
119 |
+
"learning_rate": 2e-05,
|
120 |
+
"loss": 0.5218,
|
121 |
+
"step": 32
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.34,
|
125 |
+
"grad_norm": 4.621761322021484,
|
126 |
+
"learning_rate": 2e-05,
|
127 |
+
"loss": 0.9679,
|
128 |
+
"step": 34
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.36,
|
132 |
+
"grad_norm": 3.575629472732544,
|
133 |
+
"learning_rate": 2e-05,
|
134 |
+
"loss": 0.5973,
|
135 |
+
"step": 36
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.38,
|
139 |
+
"grad_norm": 9.701947212219238,
|
140 |
+
"learning_rate": 2e-05,
|
141 |
+
"loss": 1.4922,
|
142 |
+
"step": 38
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.4,
|
146 |
+
"grad_norm": 18.55413818359375,
|
147 |
+
"learning_rate": 2e-05,
|
148 |
+
"loss": 1.1465,
|
149 |
+
"step": 40
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.42,
|
153 |
+
"grad_norm": 7.169260025024414,
|
154 |
+
"learning_rate": 2e-05,
|
155 |
+
"loss": 0.9151,
|
156 |
+
"step": 42
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.44,
|
160 |
+
"grad_norm": 7.758701324462891,
|
161 |
+
"learning_rate": 2e-05,
|
162 |
+
"loss": 1.3213,
|
163 |
+
"step": 44
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.46,
|
167 |
+
"grad_norm": 7.772892475128174,
|
168 |
+
"learning_rate": 2e-05,
|
169 |
+
"loss": 0.4431,
|
170 |
+
"step": 46
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.48,
|
174 |
+
"grad_norm": 3.169400215148926,
|
175 |
+
"learning_rate": 2e-05,
|
176 |
+
"loss": 0.9307,
|
177 |
+
"step": 48
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.5,
|
181 |
+
"grad_norm": 6.446079730987549,
|
182 |
+
"learning_rate": 2e-05,
|
183 |
+
"loss": 1.4162,
|
184 |
+
"step": 50
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.52,
|
188 |
+
"grad_norm": 4.463912487030029,
|
189 |
+
"learning_rate": 2e-05,
|
190 |
+
"loss": 0.8901,
|
191 |
+
"step": 52
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.54,
|
195 |
+
"grad_norm": 7.208362102508545,
|
196 |
+
"learning_rate": 2e-05,
|
197 |
+
"loss": 1.4832,
|
198 |
+
"step": 54
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.56,
|
202 |
+
"grad_norm": 11.086746215820312,
|
203 |
+
"learning_rate": 2e-05,
|
204 |
+
"loss": 2.0512,
|
205 |
+
"step": 56
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.58,
|
209 |
+
"grad_norm": 6.340665340423584,
|
210 |
+
"learning_rate": 2e-05,
|
211 |
+
"loss": 1.0799,
|
212 |
+
"step": 58
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.6,
|
216 |
+
"grad_norm": 2.8643109798431396,
|
217 |
+
"learning_rate": 2e-05,
|
218 |
+
"loss": 0.5789,
|
219 |
+
"step": 60
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.62,
|
223 |
+
"grad_norm": 7.802653789520264,
|
224 |
+
"learning_rate": 2e-05,
|
225 |
+
"loss": 1.8293,
|
226 |
+
"step": 62
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.64,
|
230 |
+
"grad_norm": 4.388473033905029,
|
231 |
+
"learning_rate": 2e-05,
|
232 |
+
"loss": 1.0891,
|
233 |
+
"step": 64
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.66,
|
237 |
+
"grad_norm": 4.431455612182617,
|
238 |
+
"learning_rate": 2e-05,
|
239 |
+
"loss": 0.9788,
|
240 |
+
"step": 66
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.68,
|
244 |
+
"grad_norm": 0.8305065035820007,
|
245 |
+
"learning_rate": 2e-05,
|
246 |
+
"loss": 0.6982,
|
247 |
+
"step": 68
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.7,
|
251 |
+
"grad_norm": 14.678071975708008,
|
252 |
+
"learning_rate": 2e-05,
|
253 |
+
"loss": 1.9248,
|
254 |
+
"step": 70
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.72,
|
258 |
+
"grad_norm": 2.8914413452148438,
|
259 |
+
"learning_rate": 2e-05,
|
260 |
+
"loss": 0.5398,
|
261 |
+
"step": 72
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.74,
|
265 |
+
"grad_norm": 2.2980284690856934,
|
266 |
+
"learning_rate": 2e-05,
|
267 |
+
"loss": 1.6722,
|
268 |
+
"step": 74
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.76,
|
272 |
+
"grad_norm": 4.637665271759033,
|
273 |
+
"learning_rate": 2e-05,
|
274 |
+
"loss": 1.3349,
|
275 |
+
"step": 76
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.78,
|
279 |
+
"grad_norm": 2.5008809566497803,
|
280 |
+
"learning_rate": 2e-05,
|
281 |
+
"loss": 1.134,
|
282 |
+
"step": 78
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.8,
|
286 |
+
"grad_norm": 2.6805694103240967,
|
287 |
+
"learning_rate": 2e-05,
|
288 |
+
"loss": 0.935,
|
289 |
+
"step": 80
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.82,
|
293 |
+
"grad_norm": 7.456270694732666,
|
294 |
+
"learning_rate": 2e-05,
|
295 |
+
"loss": 1.5703,
|
296 |
+
"step": 82
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.84,
|
300 |
+
"grad_norm": 3.0763790607452393,
|
301 |
+
"learning_rate": 2e-05,
|
302 |
+
"loss": 0.3613,
|
303 |
+
"step": 84
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.86,
|
307 |
+
"grad_norm": 8.599203109741211,
|
308 |
+
"learning_rate": 2e-05,
|
309 |
+
"loss": 0.9858,
|
310 |
+
"step": 86
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.88,
|
314 |
+
"grad_norm": 7.865464687347412,
|
315 |
+
"learning_rate": 2e-05,
|
316 |
+
"loss": 1.5265,
|
317 |
+
"step": 88
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.9,
|
321 |
+
"grad_norm": 7.330792427062988,
|
322 |
+
"learning_rate": 2e-05,
|
323 |
+
"loss": 1.392,
|
324 |
+
"step": 90
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.92,
|
328 |
+
"grad_norm": 5.383084774017334,
|
329 |
+
"learning_rate": 2e-05,
|
330 |
+
"loss": 1.248,
|
331 |
+
"step": 92
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.94,
|
335 |
+
"grad_norm": 6.0959649085998535,
|
336 |
+
"learning_rate": 2e-05,
|
337 |
+
"loss": 1.9657,
|
338 |
+
"step": 94
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.96,
|
342 |
+
"grad_norm": 2.155146598815918,
|
343 |
+
"learning_rate": 2e-05,
|
344 |
+
"loss": 0.8844,
|
345 |
+
"step": 96
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.98,
|
349 |
+
"grad_norm": 5.614543437957764,
|
350 |
+
"learning_rate": 2e-05,
|
351 |
+
"loss": 1.4389,
|
352 |
+
"step": 98
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 1.0,
|
356 |
+
"grad_norm": 8.140727043151855,
|
357 |
+
"learning_rate": 2e-05,
|
358 |
+
"loss": 1.3601,
|
359 |
+
"step": 100
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 1.0,
|
363 |
+
"step": 100,
|
364 |
+
"total_flos": 2098644891205632.0,
|
365 |
+
"train_loss": 1.066049919128418,
|
366 |
+
"train_runtime": 99.4099,
|
367 |
+
"train_samples_per_second": 4.024,
|
368 |
+
"train_steps_per_second": 1.006
|
369 |
+
}
|
370 |
+
],
|
371 |
+
"logging_steps": 2,
|
372 |
+
"max_steps": 100,
|
373 |
+
"num_input_tokens_seen": 0,
|
374 |
+
"num_train_epochs": 1,
|
375 |
+
"save_steps": 500,
|
376 |
+
"stateful_callbacks": {
|
377 |
+
"TrainerControl": {
|
378 |
+
"args": {
|
379 |
+
"should_epoch_stop": false,
|
380 |
+
"should_evaluate": false,
|
381 |
+
"should_log": false,
|
382 |
+
"should_save": false,
|
383 |
+
"should_training_stop": false
|
384 |
+
},
|
385 |
+
"attributes": {}
|
386 |
+
}
|
387 |
+
},
|
388 |
+
"total_flos": 2098644891205632.0,
|
389 |
+
"train_batch_size": 1,
|
390 |
+
"trial_name": null,
|
391 |
+
"trial_params": null
|
392 |
+
}
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round10.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3506ffc3670759125f1d1c08b6420c1bf38b2f186b977f9cbb77ec8fcbeea88e
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round12.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7be4b6efdf8aeda1d55eab73770ec3ff6a9a14501a2aa60e833ec969ed086533
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round15.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:90585364faf7ecfbee025db3a036cc013caf849e1bed9b8c5d1c4570c6065777
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round17.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9236b8e15ebe274a58931ff42727de3f819bbb0f8493426399ff78877af9de2b
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a161db119e3dd4b0ad54e2dc5dae4212100d532c0ec87749b3eb4d37f395bf48
|
3 |
+
size 395786922
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round20.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:16b3859505775fae6ab29fd2053ab49c87ca8a84e54cda0d5304f7c72b066afd
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round5.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:05e363377b0c5548371d96debaaf6fff6fa705a7ca1321530510e181d3556b7e
|
3 |
+
size 395786922
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_client_model_round7.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9cd95080fae073060aa659888b415d2135edd62af172653cfed0f99ab94464f6
|
3 |
+
size 395786922
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/4_trainer_state.json
ADDED
@@ -0,0 +1,392 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 100,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.02,
|
13 |
+
"grad_norm": 3.6383607387542725,
|
14 |
+
"learning_rate": 2e-05,
|
15 |
+
"loss": 0.9771,
|
16 |
+
"step": 2
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.04,
|
20 |
+
"grad_norm": 1.4260591268539429,
|
21 |
+
"learning_rate": 2e-05,
|
22 |
+
"loss": 0.5006,
|
23 |
+
"step": 4
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.06,
|
27 |
+
"grad_norm": 1.6014289855957031,
|
28 |
+
"learning_rate": 2e-05,
|
29 |
+
"loss": 0.782,
|
30 |
+
"step": 6
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.08,
|
34 |
+
"grad_norm": 2.321310043334961,
|
35 |
+
"learning_rate": 2e-05,
|
36 |
+
"loss": 0.3546,
|
37 |
+
"step": 8
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.1,
|
41 |
+
"grad_norm": 1.937113881111145,
|
42 |
+
"learning_rate": 2e-05,
|
43 |
+
"loss": 1.2421,
|
44 |
+
"step": 10
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.12,
|
48 |
+
"grad_norm": 2.4271795749664307,
|
49 |
+
"learning_rate": 2e-05,
|
50 |
+
"loss": 1.005,
|
51 |
+
"step": 12
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.14,
|
55 |
+
"grad_norm": 0.9630257487297058,
|
56 |
+
"learning_rate": 2e-05,
|
57 |
+
"loss": 0.635,
|
58 |
+
"step": 14
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.16,
|
62 |
+
"grad_norm": 2.273303985595703,
|
63 |
+
"learning_rate": 2e-05,
|
64 |
+
"loss": 0.9471,
|
65 |
+
"step": 16
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.18,
|
69 |
+
"grad_norm": 1.193511724472046,
|
70 |
+
"learning_rate": 2e-05,
|
71 |
+
"loss": 0.6021,
|
72 |
+
"step": 18
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.2,
|
76 |
+
"grad_norm": 2.983036518096924,
|
77 |
+
"learning_rate": 2e-05,
|
78 |
+
"loss": 0.6765,
|
79 |
+
"step": 20
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.22,
|
83 |
+
"grad_norm": 1.626178503036499,
|
84 |
+
"learning_rate": 2e-05,
|
85 |
+
"loss": 1.0259,
|
86 |
+
"step": 22
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.24,
|
90 |
+
"grad_norm": 1.8325883150100708,
|
91 |
+
"learning_rate": 2e-05,
|
92 |
+
"loss": 0.5989,
|
93 |
+
"step": 24
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.26,
|
97 |
+
"grad_norm": 2.876636266708374,
|
98 |
+
"learning_rate": 2e-05,
|
99 |
+
"loss": 0.8092,
|
100 |
+
"step": 26
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.28,
|
104 |
+
"grad_norm": 2.8380074501037598,
|
105 |
+
"learning_rate": 2e-05,
|
106 |
+
"loss": 0.5194,
|
107 |
+
"step": 28
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.3,
|
111 |
+
"grad_norm": 2.227553129196167,
|
112 |
+
"learning_rate": 2e-05,
|
113 |
+
"loss": 0.6971,
|
114 |
+
"step": 30
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.32,
|
118 |
+
"grad_norm": 2.7015953063964844,
|
119 |
+
"learning_rate": 2e-05,
|
120 |
+
"loss": 0.6297,
|
121 |
+
"step": 32
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.34,
|
125 |
+
"grad_norm": 3.0312774181365967,
|
126 |
+
"learning_rate": 2e-05,
|
127 |
+
"loss": 1.3067,
|
128 |
+
"step": 34
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.36,
|
132 |
+
"grad_norm": 2.7130300998687744,
|
133 |
+
"learning_rate": 2e-05,
|
134 |
+
"loss": 0.702,
|
135 |
+
"step": 36
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.38,
|
139 |
+
"grad_norm": 0.3681621849536896,
|
140 |
+
"learning_rate": 2e-05,
|
141 |
+
"loss": 0.1352,
|
142 |
+
"step": 38
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.4,
|
146 |
+
"grad_norm": 0.8578549027442932,
|
147 |
+
"learning_rate": 2e-05,
|
148 |
+
"loss": 0.1746,
|
149 |
+
"step": 40
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.42,
|
153 |
+
"grad_norm": 1.2145777940750122,
|
154 |
+
"learning_rate": 2e-05,
|
155 |
+
"loss": 0.62,
|
156 |
+
"step": 42
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.44,
|
160 |
+
"grad_norm": 0.8366961479187012,
|
161 |
+
"learning_rate": 2e-05,
|
162 |
+
"loss": 0.237,
|
163 |
+
"step": 44
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.46,
|
167 |
+
"grad_norm": 3.8259220123291016,
|
168 |
+
"learning_rate": 2e-05,
|
169 |
+
"loss": 1.1652,
|
170 |
+
"step": 46
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.48,
|
174 |
+
"grad_norm": 2.451016426086426,
|
175 |
+
"learning_rate": 2e-05,
|
176 |
+
"loss": 0.2307,
|
177 |
+
"step": 48
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.5,
|
181 |
+
"grad_norm": 7.296032905578613,
|
182 |
+
"learning_rate": 2e-05,
|
183 |
+
"loss": 0.6684,
|
184 |
+
"step": 50
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.52,
|
188 |
+
"grad_norm": 2.7998406887054443,
|
189 |
+
"learning_rate": 2e-05,
|
190 |
+
"loss": 0.5629,
|
191 |
+
"step": 52
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.54,
|
195 |
+
"grad_norm": 1.9942561388015747,
|
196 |
+
"learning_rate": 2e-05,
|
197 |
+
"loss": 0.4002,
|
198 |
+
"step": 54
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.56,
|
202 |
+
"grad_norm": 3.880875587463379,
|
203 |
+
"learning_rate": 2e-05,
|
204 |
+
"loss": 0.5188,
|
205 |
+
"step": 56
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.58,
|
209 |
+
"grad_norm": 5.010158538818359,
|
210 |
+
"learning_rate": 2e-05,
|
211 |
+
"loss": 0.6704,
|
212 |
+
"step": 58
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.6,
|
216 |
+
"grad_norm": 6.04047155380249,
|
217 |
+
"learning_rate": 2e-05,
|
218 |
+
"loss": 0.8029,
|
219 |
+
"step": 60
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.62,
|
223 |
+
"grad_norm": 3.4900665283203125,
|
224 |
+
"learning_rate": 2e-05,
|
225 |
+
"loss": 0.9319,
|
226 |
+
"step": 62
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.64,
|
230 |
+
"grad_norm": 1.4386566877365112,
|
231 |
+
"learning_rate": 2e-05,
|
232 |
+
"loss": 0.392,
|
233 |
+
"step": 64
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.66,
|
237 |
+
"grad_norm": 2.075533866882324,
|
238 |
+
"learning_rate": 2e-05,
|
239 |
+
"loss": 1.0757,
|
240 |
+
"step": 66
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.68,
|
244 |
+
"grad_norm": 5.302359104156494,
|
245 |
+
"learning_rate": 2e-05,
|
246 |
+
"loss": 1.017,
|
247 |
+
"step": 68
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.7,
|
251 |
+
"grad_norm": 3.4823062419891357,
|
252 |
+
"learning_rate": 2e-05,
|
253 |
+
"loss": 1.1879,
|
254 |
+
"step": 70
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.72,
|
258 |
+
"grad_norm": 0.5839654803276062,
|
259 |
+
"learning_rate": 2e-05,
|
260 |
+
"loss": 0.1357,
|
261 |
+
"step": 72
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.74,
|
265 |
+
"grad_norm": 3.582629442214966,
|
266 |
+
"learning_rate": 2e-05,
|
267 |
+
"loss": 1.5566,
|
268 |
+
"step": 74
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.76,
|
272 |
+
"grad_norm": 1.8023282289505005,
|
273 |
+
"learning_rate": 2e-05,
|
274 |
+
"loss": 0.513,
|
275 |
+
"step": 76
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.78,
|
279 |
+
"grad_norm": 0.7795921564102173,
|
280 |
+
"learning_rate": 2e-05,
|
281 |
+
"loss": 0.8818,
|
282 |
+
"step": 78
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.8,
|
286 |
+
"grad_norm": 2.052372932434082,
|
287 |
+
"learning_rate": 2e-05,
|
288 |
+
"loss": 0.6126,
|
289 |
+
"step": 80
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.82,
|
293 |
+
"grad_norm": 1.606740117073059,
|
294 |
+
"learning_rate": 2e-05,
|
295 |
+
"loss": 0.5911,
|
296 |
+
"step": 82
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.84,
|
300 |
+
"grad_norm": 1.955881953239441,
|
301 |
+
"learning_rate": 2e-05,
|
302 |
+
"loss": 0.6584,
|
303 |
+
"step": 84
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.86,
|
307 |
+
"grad_norm": 2.623976469039917,
|
308 |
+
"learning_rate": 2e-05,
|
309 |
+
"loss": 0.5529,
|
310 |
+
"step": 86
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.88,
|
314 |
+
"grad_norm": 2.0439250469207764,
|
315 |
+
"learning_rate": 2e-05,
|
316 |
+
"loss": 0.9415,
|
317 |
+
"step": 88
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.9,
|
321 |
+
"grad_norm": 0.6250275373458862,
|
322 |
+
"learning_rate": 2e-05,
|
323 |
+
"loss": 0.1703,
|
324 |
+
"step": 90
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.92,
|
328 |
+
"grad_norm": 2.0065391063690186,
|
329 |
+
"learning_rate": 2e-05,
|
330 |
+
"loss": 0.6163,
|
331 |
+
"step": 92
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.94,
|
335 |
+
"grad_norm": 3.4686758518218994,
|
336 |
+
"learning_rate": 2e-05,
|
337 |
+
"loss": 1.1005,
|
338 |
+
"step": 94
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.96,
|
342 |
+
"grad_norm": 2.2346303462982178,
|
343 |
+
"learning_rate": 2e-05,
|
344 |
+
"loss": 0.8502,
|
345 |
+
"step": 96
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.98,
|
349 |
+
"grad_norm": 3.8489227294921875,
|
350 |
+
"learning_rate": 2e-05,
|
351 |
+
"loss": 1.3295,
|
352 |
+
"step": 98
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 1.0,
|
356 |
+
"grad_norm": 2.55889892578125,
|
357 |
+
"learning_rate": 2e-05,
|
358 |
+
"loss": 0.634,
|
359 |
+
"step": 100
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 1.0,
|
363 |
+
"step": 100,
|
364 |
+
"total_flos": 5692822418096128.0,
|
365 |
+
"train_loss": 0.71892333984375,
|
366 |
+
"train_runtime": 165.0653,
|
367 |
+
"train_samples_per_second": 2.423,
|
368 |
+
"train_steps_per_second": 0.606
|
369 |
+
}
|
370 |
+
],
|
371 |
+
"logging_steps": 2,
|
372 |
+
"max_steps": 100,
|
373 |
+
"num_input_tokens_seen": 0,
|
374 |
+
"num_train_epochs": 1,
|
375 |
+
"save_steps": 500,
|
376 |
+
"stateful_callbacks": {
|
377 |
+
"TrainerControl": {
|
378 |
+
"args": {
|
379 |
+
"should_epoch_stop": false,
|
380 |
+
"should_evaluate": false,
|
381 |
+
"should_log": false,
|
382 |
+
"should_save": false,
|
383 |
+
"should_training_stop": false
|
384 |
+
},
|
385 |
+
"attributes": {}
|
386 |
+
}
|
387 |
+
},
|
388 |
+
"total_flos": 5692822418096128.0,
|
389 |
+
"train_batch_size": 1,
|
390 |
+
"trial_name": null,
|
391 |
+
"trial_params": null
|
392 |
+
}
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round10.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:04c033bed86fd20d1bf5f45a017a7235285089bde6b845487e1d24f22528cb78
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round12.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3aa0ede5f53ed6efb703e4de0c88d026168e8444ff6938342da4a8d4b540d5df
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round15.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5c172229e49be46b6d28ec7d39d0dec151b6a3955e703aca7df6e3bab423f73b
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round17.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1781b39f8ac95a428751a6de041f2ef6a523a42fcaff84daefa48b7c5e687690
|
3 |
+
size 395787774
|
client_states_fedMultipqfullfreeze_homoAgg_NOCONT_bs4_saveoptim_lr2e-5_5e-5_sc315_4tasks_5rounds_fixitr100_T0125_decay099_SEED2/5_client_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:204e4484268b051e95249cd0b0d80d5fc1c8b309c758d33fca781ee24da8090f
|
3 |
+
size 395786922
|