Commit
•
3df4e32
1
Parent(s):
48f377f
Training in progress, step 7724, checkpoint
Browse files- last-checkpoint/adapter_model.safetensors +1 -1
- last-checkpoint/global_step7724/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step7724/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step7724/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step7724/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step7724/zero_pp_rank_0_mp_rank_00_model_states.pt +3 -0
- last-checkpoint/global_step7724/zero_pp_rank_1_mp_rank_00_model_states.pt +3 -0
- last-checkpoint/global_step7724/zero_pp_rank_2_mp_rank_00_model_states.pt +3 -0
- last-checkpoint/global_step7724/zero_pp_rank_3_mp_rank_00_model_states.pt +3 -0
- last-checkpoint/latest +1 -1
- last-checkpoint/rng_state_0.pth +1 -1
- last-checkpoint/rng_state_1.pth +1 -1
- last-checkpoint/rng_state_2.pth +1 -1
- last-checkpoint/rng_state_3.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +2276 -4
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 10107504
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc3067f6f49803d6483ed250997366f38d0ab2f09b2da22d2fabe6ac0fb75ff6
|
| 3 |
size 10107504
|
last-checkpoint/global_step7724/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:97c6b7bff0e93774276654dfabe2789287685eff81eb0356b3c4e432d008879b
|
| 3 |
+
size 15142384
|
last-checkpoint/global_step7724/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e0ae9c9e58dc038648826e0b8f756489ad36eeb01e206fc42eeb3371aa7ab4f8
|
| 3 |
+
size 15142384
|
last-checkpoint/global_step7724/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:873d603f9ba9580b5e130bff47347dcc9654a60665ff32cf88b22531d21ad096
|
| 3 |
+
size 15142384
|
last-checkpoint/global_step7724/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b48442fffd98f620e79aec539490b8f3738cb7abd39f3ed80c2070d65eac7eb7
|
| 3 |
+
size 15142384
|
last-checkpoint/global_step7724/zero_pp_rank_0_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99fd6c868aa11063fc90b4a235b658c1f06846579371a1ca435a3752fc94024d
|
| 3 |
+
size 133406
|
last-checkpoint/global_step7724/zero_pp_rank_1_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a4b34b591a64db0bd036adac493e5bd479a400986798fb848eab6c0451bc8b0
|
| 3 |
+
size 133406
|
last-checkpoint/global_step7724/zero_pp_rank_2_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f591073404fd55b339bcc2665c6fbfbcea9d2cbdc3cec4ef877565ba3e794c9e
|
| 3 |
+
size 133406
|
last-checkpoint/global_step7724/zero_pp_rank_3_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:454aca661de8e2cda3725c47af8a9a9cdeac92329976320a37ebc99b58ff8fbe
|
| 3 |
+
size 133406
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step7724
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14960
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:365955c51abdb92c80c42d93201c5209f863b3153268eb4a953da4966f1e43a5
|
| 3 |
size 14960
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14960
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:833cfceec5f4c1e2e2fd595b8054119e44fc059867ce43ae83edc5c5ad6c4340
|
| 3 |
size 14960
|
last-checkpoint/rng_state_2.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14960
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bf0517c56512d4ae7a3361a1e8dacb66d486d479cf260ca1d6f51f6ee19825e3
|
| 3 |
size 14960
|
last-checkpoint/rng_state_3.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14960
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ed8eaad9e0b09f1cddb62096ef671eb06bf7a4be830ed6580db5b8e23e266ae
|
| 3 |
size 14960
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3270c748caccc6a16e6d4f08adacc13be568ae6a10d730ba03d05068ed26e9d
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 5,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -76523,6 +76523,2278 @@
|
|
| 76523 |
"eval_samples_per_second": 3.173,
|
| 76524 |
"eval_steps_per_second": 0.216,
|
| 76525 |
"step": 7500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 76526 |
}
|
| 76527 |
],
|
| 76528 |
"logging_steps": 1,
|
|
@@ -76537,12 +78809,12 @@
|
|
| 76537 |
"should_evaluate": false,
|
| 76538 |
"should_log": false,
|
| 76539 |
"should_save": true,
|
| 76540 |
-
"should_training_stop":
|
| 76541 |
},
|
| 76542 |
"attributes": {}
|
| 76543 |
}
|
| 76544 |
},
|
| 76545 |
-
"total_flos":
|
| 76546 |
"train_batch_size": 4,
|
| 76547 |
"trial_name": null,
|
| 76548 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9999676343981616,
|
| 5 |
"eval_steps": 5,
|
| 6 |
+
"global_step": 7724,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 76523 |
"eval_samples_per_second": 3.173,
|
| 76524 |
"eval_steps_per_second": 0.216,
|
| 76525 |
"step": 7500
|
| 76526 |
+
},
|
| 76527 |
+
{
|
| 76528 |
+
"epoch": 0.971097517558339,
|
| 76529 |
+
"grad_norm": 2.078256325407057,
|
| 76530 |
+
"learning_rate": 3.171737588503407e-09,
|
| 76531 |
+
"loss": 2.4272,
|
| 76532 |
+
"step": 7501
|
| 76533 |
+
},
|
| 76534 |
+
{
|
| 76535 |
+
"epoch": 0.9712269799656925,
|
| 76536 |
+
"grad_norm": 1.366999089040965,
|
| 76537 |
+
"learning_rate": 3.1433791070100555e-09,
|
| 76538 |
+
"loss": 2.229,
|
| 76539 |
+
"step": 7502
|
| 76540 |
+
},
|
| 76541 |
+
{
|
| 76542 |
+
"epoch": 0.971356442373046,
|
| 76543 |
+
"grad_norm": 1.8412842996905656,
|
| 76544 |
+
"learning_rate": 3.1151476523359064e-09,
|
| 76545 |
+
"loss": 2.374,
|
| 76546 |
+
"step": 7503
|
| 76547 |
+
},
|
| 76548 |
+
{
|
| 76549 |
+
"epoch": 0.9714859047803994,
|
| 76550 |
+
"grad_norm": 2.4036313021269984,
|
| 76551 |
+
"learning_rate": 3.0870432302479435e-09,
|
| 76552 |
+
"loss": 2.552,
|
| 76553 |
+
"step": 7504
|
| 76554 |
+
},
|
| 76555 |
+
{
|
| 76556 |
+
"epoch": 0.9716153671877529,
|
| 76557 |
+
"grad_norm": 1.3319207705837204,
|
| 76558 |
+
"learning_rate": 3.059065846487061e-09,
|
| 76559 |
+
"loss": 2.4368,
|
| 76560 |
+
"step": 7505
|
| 76561 |
+
},
|
| 76562 |
+
{
|
| 76563 |
+
"epoch": 0.9716153671877529,
|
| 76564 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76565 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76566 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76567 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76568 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76569 |
+
"eval_PRM Precision": 1.0,
|
| 76570 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76571 |
+
"eval_PRM Specificty": 1.0,
|
| 76572 |
+
"eval_loss": 1.4964932203292847,
|
| 76573 |
+
"eval_runtime": 14.552,
|
| 76574 |
+
"eval_samples_per_second": 3.024,
|
| 76575 |
+
"eval_steps_per_second": 0.206,
|
| 76576 |
+
"step": 7505
|
| 76577 |
+
},
|
| 76578 |
+
{
|
| 76579 |
+
"epoch": 0.9717448295951063,
|
| 76580 |
+
"grad_norm": 2.604357541325846,
|
| 76581 |
+
"learning_rate": 3.031215506768062e-09,
|
| 76582 |
+
"loss": 2.4111,
|
| 76583 |
+
"step": 7506
|
| 76584 |
+
},
|
| 76585 |
+
{
|
| 76586 |
+
"epoch": 0.9718742920024598,
|
| 76587 |
+
"grad_norm": 1.7566724147974662,
|
| 76588 |
+
"learning_rate": 3.003492216780077e-09,
|
| 76589 |
+
"loss": 2.4978,
|
| 76590 |
+
"step": 7507
|
| 76591 |
+
},
|
| 76592 |
+
{
|
| 76593 |
+
"epoch": 0.9720037544098132,
|
| 76594 |
+
"grad_norm": 1.8860731801998685,
|
| 76595 |
+
"learning_rate": 2.975895982186075e-09,
|
| 76596 |
+
"loss": 2.1368,
|
| 76597 |
+
"step": 7508
|
| 76598 |
+
},
|
| 76599 |
+
{
|
| 76600 |
+
"epoch": 0.9721332168171667,
|
| 76601 |
+
"grad_norm": 1.336728000590379,
|
| 76602 |
+
"learning_rate": 2.948426808623145e-09,
|
| 76603 |
+
"loss": 2.3132,
|
| 76604 |
+
"step": 7509
|
| 76605 |
+
},
|
| 76606 |
+
{
|
| 76607 |
+
"epoch": 0.9722626792245201,
|
| 76608 |
+
"grad_norm": 1.6261016992599011,
|
| 76609 |
+
"learning_rate": 2.9210847017024922e-09,
|
| 76610 |
+
"loss": 2.2693,
|
| 76611 |
+
"step": 7510
|
| 76612 |
+
},
|
| 76613 |
+
{
|
| 76614 |
+
"epoch": 0.9722626792245201,
|
| 76615 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76616 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76617 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76618 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76619 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76620 |
+
"eval_PRM Precision": 1.0,
|
| 76621 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76622 |
+
"eval_PRM Specificty": 1.0,
|
| 76623 |
+
"eval_loss": 1.495072841644287,
|
| 76624 |
+
"eval_runtime": 14.6337,
|
| 76625 |
+
"eval_samples_per_second": 3.007,
|
| 76626 |
+
"eval_steps_per_second": 0.205,
|
| 76627 |
+
"step": 7510
|
| 76628 |
+
},
|
| 76629 |
+
{
|
| 76630 |
+
"epoch": 0.9723921416318736,
|
| 76631 |
+
"grad_norm": 1.7337031603610027,
|
| 76632 |
+
"learning_rate": 2.893869667009233e-09,
|
| 76633 |
+
"loss": 2.573,
|
| 76634 |
+
"step": 7511
|
| 76635 |
+
},
|
| 76636 |
+
{
|
| 76637 |
+
"epoch": 0.9725216040392272,
|
| 76638 |
+
"grad_norm": 1.516881266320023,
|
| 76639 |
+
"learning_rate": 2.866781710102601e-09,
|
| 76640 |
+
"loss": 2.4393,
|
| 76641 |
+
"step": 7512
|
| 76642 |
+
},
|
| 76643 |
+
{
|
| 76644 |
+
"epoch": 0.9726510664465806,
|
| 76645 |
+
"grad_norm": 1.778019440023385,
|
| 76646 |
+
"learning_rate": 2.839820836515947e-09,
|
| 76647 |
+
"loss": 2.4497,
|
| 76648 |
+
"step": 7513
|
| 76649 |
+
},
|
| 76650 |
+
{
|
| 76651 |
+
"epoch": 0.9727805288539341,
|
| 76652 |
+
"grad_norm": 1.649562667349022,
|
| 76653 |
+
"learning_rate": 2.812987051756394e-09,
|
| 76654 |
+
"loss": 2.3696,
|
| 76655 |
+
"step": 7514
|
| 76656 |
+
},
|
| 76657 |
+
{
|
| 76658 |
+
"epoch": 0.9729099912612875,
|
| 76659 |
+
"grad_norm": 1.6709675287516073,
|
| 76660 |
+
"learning_rate": 2.78628036130546e-09,
|
| 76661 |
+
"loss": 2.7583,
|
| 76662 |
+
"step": 7515
|
| 76663 |
+
},
|
| 76664 |
+
{
|
| 76665 |
+
"epoch": 0.9729099912612875,
|
| 76666 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76667 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76668 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76669 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76670 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76671 |
+
"eval_PRM Precision": 1.0,
|
| 76672 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76673 |
+
"eval_PRM Specificty": 1.0,
|
| 76674 |
+
"eval_loss": 1.4964045286178589,
|
| 76675 |
+
"eval_runtime": 14.1338,
|
| 76676 |
+
"eval_samples_per_second": 3.113,
|
| 76677 |
+
"eval_steps_per_second": 0.212,
|
| 76678 |
+
"step": 7515
|
| 76679 |
+
},
|
| 76680 |
+
{
|
| 76681 |
+
"epoch": 0.973039453668641,
|
| 76682 |
+
"grad_norm": 1.208623897904988,
|
| 76683 |
+
"learning_rate": 2.7597007706184344e-09,
|
| 76684 |
+
"loss": 2.2378,
|
| 76685 |
+
"step": 7516
|
| 76686 |
+
},
|
| 76687 |
+
{
|
| 76688 |
+
"epoch": 0.9731689160759944,
|
| 76689 |
+
"grad_norm": 1.3760642333243103,
|
| 76690 |
+
"learning_rate": 2.7332482851247926e-09,
|
| 76691 |
+
"loss": 2.4258,
|
| 76692 |
+
"step": 7517
|
| 76693 |
+
},
|
| 76694 |
+
{
|
| 76695 |
+
"epoch": 0.9732983784833479,
|
| 76696 |
+
"grad_norm": 3.037334589617975,
|
| 76697 |
+
"learning_rate": 2.7069229102279217e-09,
|
| 76698 |
+
"loss": 2.3164,
|
| 76699 |
+
"step": 7518
|
| 76700 |
+
},
|
| 76701 |
+
{
|
| 76702 |
+
"epoch": 0.9734278408907013,
|
| 76703 |
+
"grad_norm": 1.5660663366500944,
|
| 76704 |
+
"learning_rate": 2.680724651305325e-09,
|
| 76705 |
+
"loss": 2.1655,
|
| 76706 |
+
"step": 7519
|
| 76707 |
+
},
|
| 76708 |
+
{
|
| 76709 |
+
"epoch": 0.9735573032980548,
|
| 76710 |
+
"grad_norm": 2.3157849481641852,
|
| 76711 |
+
"learning_rate": 2.6546535137086244e-09,
|
| 76712 |
+
"loss": 2.5334,
|
| 76713 |
+
"step": 7520
|
| 76714 |
+
},
|
| 76715 |
+
{
|
| 76716 |
+
"epoch": 0.9735573032980548,
|
| 76717 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76718 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76719 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76720 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76721 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76722 |
+
"eval_PRM Precision": 1.0,
|
| 76723 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76724 |
+
"eval_PRM Specificty": 1.0,
|
| 76725 |
+
"eval_loss": 1.4964932203292847,
|
| 76726 |
+
"eval_runtime": 14.4152,
|
| 76727 |
+
"eval_samples_per_second": 3.052,
|
| 76728 |
+
"eval_steps_per_second": 0.208,
|
| 76729 |
+
"step": 7520
|
| 76730 |
+
},
|
| 76731 |
+
{
|
| 76732 |
+
"epoch": 0.9736867657054082,
|
| 76733 |
+
"grad_norm": 1.6032188100029758,
|
| 76734 |
+
"learning_rate": 2.6287095027632824e-09,
|
| 76735 |
+
"loss": 2.3384,
|
| 76736 |
+
"step": 7521
|
| 76737 |
+
},
|
| 76738 |
+
{
|
| 76739 |
+
"epoch": 0.9738162281127618,
|
| 76740 |
+
"grad_norm": 1.6009070614900751,
|
| 76741 |
+
"learning_rate": 2.6028926237689488e-09,
|
| 76742 |
+
"loss": 2.4028,
|
| 76743 |
+
"step": 7522
|
| 76744 |
+
},
|
| 76745 |
+
{
|
| 76746 |
+
"epoch": 0.9739456905201153,
|
| 76747 |
+
"grad_norm": 1.935103516636611,
|
| 76748 |
+
"learning_rate": 2.577202881999183e-09,
|
| 76749 |
+
"loss": 2.2874,
|
| 76750 |
+
"step": 7523
|
| 76751 |
+
},
|
| 76752 |
+
{
|
| 76753 |
+
"epoch": 0.9740751529274687,
|
| 76754 |
+
"grad_norm": 1.8155538775672537,
|
| 76755 |
+
"learning_rate": 2.5516402827016617e-09,
|
| 76756 |
+
"loss": 2.3831,
|
| 76757 |
+
"step": 7524
|
| 76758 |
+
},
|
| 76759 |
+
{
|
| 76760 |
+
"epoch": 0.9742046153348222,
|
| 76761 |
+
"grad_norm": 1.0483947077967555,
|
| 76762 |
+
"learning_rate": 2.5262048310980418e-09,
|
| 76763 |
+
"loss": 2.1833,
|
| 76764 |
+
"step": 7525
|
| 76765 |
+
},
|
| 76766 |
+
{
|
| 76767 |
+
"epoch": 0.9742046153348222,
|
| 76768 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76769 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76770 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76771 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76772 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76773 |
+
"eval_PRM Precision": 1.0,
|
| 76774 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76775 |
+
"eval_PRM Specificty": 1.0,
|
| 76776 |
+
"eval_loss": 1.495805263519287,
|
| 76777 |
+
"eval_runtime": 14.1041,
|
| 76778 |
+
"eval_samples_per_second": 3.12,
|
| 76779 |
+
"eval_steps_per_second": 0.213,
|
| 76780 |
+
"step": 7525
|
| 76781 |
+
},
|
| 76782 |
+
{
|
| 76783 |
+
"epoch": 0.9743340777421756,
|
| 76784 |
+
"grad_norm": 1.552266890430507,
|
| 76785 |
+
"learning_rate": 2.500896532384098e-09,
|
| 76786 |
+
"loss": 2.1396,
|
| 76787 |
+
"step": 7526
|
| 76788 |
+
},
|
| 76789 |
+
{
|
| 76790 |
+
"epoch": 0.9744635401495291,
|
| 76791 |
+
"grad_norm": 4.162519199111511,
|
| 76792 |
+
"learning_rate": 2.4757153917295136e-09,
|
| 76793 |
+
"loss": 3.0793,
|
| 76794 |
+
"step": 7527
|
| 76795 |
+
},
|
| 76796 |
+
{
|
| 76797 |
+
"epoch": 0.9745930025568825,
|
| 76798 |
+
"grad_norm": 1.4020772771060606,
|
| 76799 |
+
"learning_rate": 2.4506614142780216e-09,
|
| 76800 |
+
"loss": 2.2017,
|
| 76801 |
+
"step": 7528
|
| 76802 |
+
},
|
| 76803 |
+
{
|
| 76804 |
+
"epoch": 0.974722464964236,
|
| 76805 |
+
"grad_norm": 2.257257302308101,
|
| 76806 |
+
"learning_rate": 2.4257346051474034e-09,
|
| 76807 |
+
"loss": 2.3549,
|
| 76808 |
+
"step": 7529
|
| 76809 |
+
},
|
| 76810 |
+
{
|
| 76811 |
+
"epoch": 0.9748519273715894,
|
| 76812 |
+
"grad_norm": 2.304261799764049,
|
| 76813 |
+
"learning_rate": 2.400934969429558e-09,
|
| 76814 |
+
"loss": 2.4275,
|
| 76815 |
+
"step": 7530
|
| 76816 |
+
},
|
| 76817 |
+
{
|
| 76818 |
+
"epoch": 0.9748519273715894,
|
| 76819 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76820 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76821 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76822 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76823 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76824 |
+
"eval_PRM Precision": 1.0,
|
| 76825 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76826 |
+
"eval_PRM Specificty": 1.0,
|
| 76827 |
+
"eval_loss": 1.495072841644287,
|
| 76828 |
+
"eval_runtime": 13.7187,
|
| 76829 |
+
"eval_samples_per_second": 3.207,
|
| 76830 |
+
"eval_steps_per_second": 0.219,
|
| 76831 |
+
"step": 7530
|
| 76832 |
+
},
|
| 76833 |
+
{
|
| 76834 |
+
"epoch": 0.9749813897789429,
|
| 76835 |
+
"grad_norm": 1.358154053410943,
|
| 76836 |
+
"learning_rate": 2.376262512190225e-09,
|
| 76837 |
+
"loss": 2.2069,
|
| 76838 |
+
"step": 7531
|
| 76839 |
+
},
|
| 76840 |
+
{
|
| 76841 |
+
"epoch": 0.9751108521862965,
|
| 76842 |
+
"grad_norm": 1.9448873246460772,
|
| 76843 |
+
"learning_rate": 2.351717238469331e-09,
|
| 76844 |
+
"loss": 2.4498,
|
| 76845 |
+
"step": 7532
|
| 76846 |
+
},
|
| 76847 |
+
{
|
| 76848 |
+
"epoch": 0.9752403145936499,
|
| 76849 |
+
"grad_norm": 2.323208439137572,
|
| 76850 |
+
"learning_rate": 2.3272991532807126e-09,
|
| 76851 |
+
"loss": 2.5518,
|
| 76852 |
+
"step": 7533
|
| 76853 |
+
},
|
| 76854 |
+
{
|
| 76855 |
+
"epoch": 0.9753697770010034,
|
| 76856 |
+
"grad_norm": 1.9000478401818919,
|
| 76857 |
+
"learning_rate": 2.3030082616121858e-09,
|
| 76858 |
+
"loss": 2.2683,
|
| 76859 |
+
"step": 7534
|
| 76860 |
+
},
|
| 76861 |
+
{
|
| 76862 |
+
"epoch": 0.9754992394083568,
|
| 76863 |
+
"grad_norm": 2.0009142246658937,
|
| 76864 |
+
"learning_rate": 2.2788445684256836e-09,
|
| 76865 |
+
"loss": 2.2664,
|
| 76866 |
+
"step": 7535
|
| 76867 |
+
},
|
| 76868 |
+
{
|
| 76869 |
+
"epoch": 0.9754992394083568,
|
| 76870 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76871 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76872 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76873 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76874 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76875 |
+
"eval_PRM Precision": 1.0,
|
| 76876 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76877 |
+
"eval_PRM Specificty": 1.0,
|
| 76878 |
+
"eval_loss": 1.4978693723678589,
|
| 76879 |
+
"eval_runtime": 14.3015,
|
| 76880 |
+
"eval_samples_per_second": 3.077,
|
| 76881 |
+
"eval_steps_per_second": 0.21,
|
| 76882 |
+
"step": 7535
|
| 76883 |
+
},
|
| 76884 |
+
{
|
| 76885 |
+
"epoch": 0.9756287018157103,
|
| 76886 |
+
"grad_norm": 1.3166239883775896,
|
| 76887 |
+
"learning_rate": 2.254808078657189e-09,
|
| 76888 |
+
"loss": 2.3547,
|
| 76889 |
+
"step": 7536
|
| 76890 |
+
},
|
| 76891 |
+
{
|
| 76892 |
+
"epoch": 0.9757581642230637,
|
| 76893 |
+
"grad_norm": 3.8384585910489326,
|
| 76894 |
+
"learning_rate": 2.230898797216663e-09,
|
| 76895 |
+
"loss": 2.3691,
|
| 76896 |
+
"step": 7537
|
| 76897 |
+
},
|
| 76898 |
+
{
|
| 76899 |
+
"epoch": 0.9758876266304172,
|
| 76900 |
+
"grad_norm": 1.5529322232040648,
|
| 76901 |
+
"learning_rate": 2.207116728987976e-09,
|
| 76902 |
+
"loss": 2.4724,
|
| 76903 |
+
"step": 7538
|
| 76904 |
+
},
|
| 76905 |
+
{
|
| 76906 |
+
"epoch": 0.9760170890377706,
|
| 76907 |
+
"grad_norm": 1.2127774713992632,
|
| 76908 |
+
"learning_rate": 2.1834618788291183e-09,
|
| 76909 |
+
"loss": 2.2393,
|
| 76910 |
+
"step": 7539
|
| 76911 |
+
},
|
| 76912 |
+
{
|
| 76913 |
+
"epoch": 0.9761465514451241,
|
| 76914 |
+
"grad_norm": 2.1174450047835274,
|
| 76915 |
+
"learning_rate": 2.1599342515720573e-09,
|
| 76916 |
+
"loss": 2.335,
|
| 76917 |
+
"step": 7540
|
| 76918 |
+
},
|
| 76919 |
+
{
|
| 76920 |
+
"epoch": 0.9761465514451241,
|
| 76921 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76922 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76923 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76924 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76925 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76926 |
+
"eval_PRM Precision": 1.0,
|
| 76927 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76928 |
+
"eval_PRM Specificty": 1.0,
|
| 76929 |
+
"eval_loss": 1.493696689605713,
|
| 76930 |
+
"eval_runtime": 14.3755,
|
| 76931 |
+
"eval_samples_per_second": 3.061,
|
| 76932 |
+
"eval_steps_per_second": 0.209,
|
| 76933 |
+
"step": 7540
|
| 76934 |
+
},
|
| 76935 |
+
{
|
| 76936 |
+
"epoch": 0.9762760138524776,
|
| 76937 |
+
"grad_norm": 2.483136951823826,
|
| 76938 |
+
"learning_rate": 2.136533852022879e-09,
|
| 76939 |
+
"loss": 2.4524,
|
| 76940 |
+
"step": 7541
|
| 76941 |
+
},
|
| 76942 |
+
{
|
| 76943 |
+
"epoch": 0.9764054762598311,
|
| 76944 |
+
"grad_norm": 1.5511048057136474,
|
| 76945 |
+
"learning_rate": 2.11326068496151e-09,
|
| 76946 |
+
"loss": 2.4346,
|
| 76947 |
+
"step": 7542
|
| 76948 |
+
},
|
| 76949 |
+
{
|
| 76950 |
+
"epoch": 0.9765349386671845,
|
| 76951 |
+
"grad_norm": 1.8425095487094147,
|
| 76952 |
+
"learning_rate": 2.0901147551419948e-09,
|
| 76953 |
+
"loss": 2.3542,
|
| 76954 |
+
"step": 7543
|
| 76955 |
+
},
|
| 76956 |
+
{
|
| 76957 |
+
"epoch": 0.976664401074538,
|
| 76958 |
+
"grad_norm": 1.591283586442896,
|
| 76959 |
+
"learning_rate": 2.0670960672923567e-09,
|
| 76960 |
+
"loss": 2.2773,
|
| 76961 |
+
"step": 7544
|
| 76962 |
+
},
|
| 76963 |
+
{
|
| 76964 |
+
"epoch": 0.9767938634818915,
|
| 76965 |
+
"grad_norm": 3.219894890994954,
|
| 76966 |
+
"learning_rate": 2.044204626114599e-09,
|
| 76967 |
+
"loss": 2.686,
|
| 76968 |
+
"step": 7545
|
| 76969 |
+
},
|
| 76970 |
+
{
|
| 76971 |
+
"epoch": 0.9767938634818915,
|
| 76972 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 76973 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 76974 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 76975 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 76976 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 76977 |
+
"eval_PRM Precision": 1.0,
|
| 76978 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 76979 |
+
"eval_PRM Specificty": 1.0,
|
| 76980 |
+
"eval_loss": 1.4943181276321411,
|
| 76981 |
+
"eval_runtime": 14.3109,
|
| 76982 |
+
"eval_samples_per_second": 3.075,
|
| 76983 |
+
"eval_steps_per_second": 0.21,
|
| 76984 |
+
"step": 7545
|
| 76985 |
+
},
|
| 76986 |
+
{
|
| 76987 |
+
"epoch": 0.9769233258892449,
|
| 76988 |
+
"grad_norm": 1.6948323938636074,
|
| 76989 |
+
"learning_rate": 2.021440436284841e-09,
|
| 76990 |
+
"loss": 2.5454,
|
| 76991 |
+
"step": 7546
|
| 76992 |
+
},
|
| 76993 |
+
{
|
| 76994 |
+
"epoch": 0.9770527882965984,
|
| 76995 |
+
"grad_norm": 2.0809224454371016,
|
| 76996 |
+
"learning_rate": 1.9988035024529756e-09,
|
| 76997 |
+
"loss": 2.2063,
|
| 76998 |
+
"step": 7547
|
| 76999 |
+
},
|
| 77000 |
+
{
|
| 77001 |
+
"epoch": 0.9771822507039518,
|
| 77002 |
+
"grad_norm": 1.8854420394243903,
|
| 77003 |
+
"learning_rate": 1.97629382924322e-09,
|
| 77004 |
+
"loss": 2.2332,
|
| 77005 |
+
"step": 7548
|
| 77006 |
+
},
|
| 77007 |
+
{
|
| 77008 |
+
"epoch": 0.9773117131113053,
|
| 77009 |
+
"grad_norm": 2.5643896611272123,
|
| 77010 |
+
"learning_rate": 1.9539114212535625e-09,
|
| 77011 |
+
"loss": 2.293,
|
| 77012 |
+
"step": 7549
|
| 77013 |
+
},
|
| 77014 |
+
{
|
| 77015 |
+
"epoch": 0.9774411755186587,
|
| 77016 |
+
"grad_norm": 1.1925655779849842,
|
| 77017 |
+
"learning_rate": 1.931656283056041e-09,
|
| 77018 |
+
"loss": 2.1436,
|
| 77019 |
+
"step": 7550
|
| 77020 |
+
},
|
| 77021 |
+
{
|
| 77022 |
+
"epoch": 0.9774411755186587,
|
| 77023 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77024 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77025 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77026 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77027 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77028 |
+
"eval_PRM Precision": 1.0,
|
| 77029 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77030 |
+
"eval_PRM Specificty": 1.0,
|
| 77031 |
+
"eval_loss": 1.495649814605713,
|
| 77032 |
+
"eval_runtime": 14.3042,
|
| 77033 |
+
"eval_samples_per_second": 3.076,
|
| 77034 |
+
"eval_steps_per_second": 0.21,
|
| 77035 |
+
"step": 7550
|
| 77036 |
+
},
|
| 77037 |
+
{
|
| 77038 |
+
"epoch": 0.9775706379260123,
|
| 77039 |
+
"grad_norm": 6.1310483327674925,
|
| 77040 |
+
"learning_rate": 1.90952841919681e-09,
|
| 77041 |
+
"loss": 2.8223,
|
| 77042 |
+
"step": 7551
|
| 77043 |
+
},
|
| 77044 |
+
{
|
| 77045 |
+
"epoch": 0.9777001003333657,
|
| 77046 |
+
"grad_norm": 1.8490057934229254,
|
| 77047 |
+
"learning_rate": 1.887527834195796e-09,
|
| 77048 |
+
"loss": 2.2754,
|
| 77049 |
+
"step": 7552
|
| 77050 |
+
},
|
| 77051 |
+
{
|
| 77052 |
+
"epoch": 0.9778295627407192,
|
| 77053 |
+
"grad_norm": 1.9545709254230654,
|
| 77054 |
+
"learning_rate": 1.8656545325471122e-09,
|
| 77055 |
+
"loss": 2.3342,
|
| 77056 |
+
"step": 7553
|
| 77057 |
+
},
|
| 77058 |
+
{
|
| 77059 |
+
"epoch": 0.9779590251480726,
|
| 77060 |
+
"grad_norm": 1.7527908462995896,
|
| 77061 |
+
"learning_rate": 1.8439085187189205e-09,
|
| 77062 |
+
"loss": 2.324,
|
| 77063 |
+
"step": 7554
|
| 77064 |
+
},
|
| 77065 |
+
{
|
| 77066 |
+
"epoch": 0.9780884875554261,
|
| 77067 |
+
"grad_norm": 1.8819808845260853,
|
| 77068 |
+
"learning_rate": 1.822289797153154e-09,
|
| 77069 |
+
"loss": 2.3325,
|
| 77070 |
+
"step": 7555
|
| 77071 |
+
},
|
| 77072 |
+
{
|
| 77073 |
+
"epoch": 0.9780884875554261,
|
| 77074 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77075 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77076 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77077 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77078 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77079 |
+
"eval_PRM Precision": 1.0,
|
| 77080 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77081 |
+
"eval_PRM Specificty": 1.0,
|
| 77082 |
+
"eval_loss": 1.4964932203292847,
|
| 77083 |
+
"eval_runtime": 13.6691,
|
| 77084 |
+
"eval_samples_per_second": 3.219,
|
| 77085 |
+
"eval_steps_per_second": 0.219,
|
| 77086 |
+
"step": 7555
|
| 77087 |
+
},
|
| 77088 |
+
{
|
| 77089 |
+
"epoch": 0.9782179499627796,
|
| 77090 |
+
"grad_norm": 1.9981064278922835,
|
| 77091 |
+
"learning_rate": 1.8007983722660021e-09,
|
| 77092 |
+
"loss": 2.519,
|
| 77093 |
+
"step": 7556
|
| 77094 |
+
},
|
| 77095 |
+
{
|
| 77096 |
+
"epoch": 0.978347412370133,
|
| 77097 |
+
"grad_norm": 2.209991202418474,
|
| 77098 |
+
"learning_rate": 1.7794342484474259e-09,
|
| 77099 |
+
"loss": 2.4434,
|
| 77100 |
+
"step": 7557
|
| 77101 |
+
},
|
| 77102 |
+
{
|
| 77103 |
+
"epoch": 0.9784768747774865,
|
| 77104 |
+
"grad_norm": 2.282220337288573,
|
| 77105 |
+
"learning_rate": 1.758197430061434e-09,
|
| 77106 |
+
"loss": 2.5867,
|
| 77107 |
+
"step": 7558
|
| 77108 |
+
},
|
| 77109 |
+
{
|
| 77110 |
+
"epoch": 0.9786063371848399,
|
| 77111 |
+
"grad_norm": 2.386537577649605,
|
| 77112 |
+
"learning_rate": 1.7370879214462232e-09,
|
| 77113 |
+
"loss": 2.3914,
|
| 77114 |
+
"step": 7559
|
| 77115 |
+
},
|
| 77116 |
+
{
|
| 77117 |
+
"epoch": 0.9787357995921934,
|
| 77118 |
+
"grad_norm": 1.4211663268329082,
|
| 77119 |
+
"learning_rate": 1.7161057269138303e-09,
|
| 77120 |
+
"loss": 2.387,
|
| 77121 |
+
"step": 7560
|
| 77122 |
+
},
|
| 77123 |
+
{
|
| 77124 |
+
"epoch": 0.9787357995921934,
|
| 77125 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77126 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77127 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77128 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77129 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77130 |
+
"eval_PRM Precision": 1.0,
|
| 77131 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77132 |
+
"eval_PRM Specificty": 1.0,
|
| 77133 |
+
"eval_loss": 1.494384765625,
|
| 77134 |
+
"eval_runtime": 14.0268,
|
| 77135 |
+
"eval_samples_per_second": 3.137,
|
| 77136 |
+
"eval_steps_per_second": 0.214,
|
| 77137 |
+
"step": 7560
|
| 77138 |
+
},
|
| 77139 |
+
{
|
| 77140 |
+
"epoch": 0.9788652619995469,
|
| 77141 |
+
"grad_norm": 2.1706050298282404,
|
| 77142 |
+
"learning_rate": 1.6952508507501324e-09,
|
| 77143 |
+
"loss": 2.1416,
|
| 77144 |
+
"step": 7561
|
| 77145 |
+
},
|
| 77146 |
+
{
|
| 77147 |
+
"epoch": 0.9789947244069004,
|
| 77148 |
+
"grad_norm": 1.4697119638840281,
|
| 77149 |
+
"learning_rate": 1.6745232972152637e-09,
|
| 77150 |
+
"loss": 2.3052,
|
| 77151 |
+
"step": 7562
|
| 77152 |
+
},
|
| 77153 |
+
{
|
| 77154 |
+
"epoch": 0.9791241868142538,
|
| 77155 |
+
"grad_norm": 1.9887583015605543,
|
| 77156 |
+
"learning_rate": 1.6539230705432675e-09,
|
| 77157 |
+
"loss": 2.7405,
|
| 77158 |
+
"step": 7563
|
| 77159 |
+
},
|
| 77160 |
+
{
|
| 77161 |
+
"epoch": 0.9792536492216073,
|
| 77162 |
+
"grad_norm": 1.3002785682904727,
|
| 77163 |
+
"learning_rate": 1.6334501749421667e-09,
|
| 77164 |
+
"loss": 2.3696,
|
| 77165 |
+
"step": 7564
|
| 77166 |
+
},
|
| 77167 |
+
{
|
| 77168 |
+
"epoch": 0.9793831116289607,
|
| 77169 |
+
"grad_norm": 4.01701817403821,
|
| 77170 |
+
"learning_rate": 1.613104614593894e-09,
|
| 77171 |
+
"loss": 2.3389,
|
| 77172 |
+
"step": 7565
|
| 77173 |
+
},
|
| 77174 |
+
{
|
| 77175 |
+
"epoch": 0.9793831116289607,
|
| 77176 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77177 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77178 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77179 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77180 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77181 |
+
"eval_PRM Precision": 1.0,
|
| 77182 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77183 |
+
"eval_PRM Specificty": 1.0,
|
| 77184 |
+
"eval_loss": 1.498579502105713,
|
| 77185 |
+
"eval_runtime": 15.3295,
|
| 77186 |
+
"eval_samples_per_second": 2.87,
|
| 77187 |
+
"eval_steps_per_second": 0.196,
|
| 77188 |
+
"step": 7565
|
| 77189 |
+
},
|
| 77190 |
+
{
|
| 77191 |
+
"epoch": 0.9795125740363142,
|
| 77192 |
+
"grad_norm": 2.82389856272306,
|
| 77193 |
+
"learning_rate": 1.5928863936544998e-09,
|
| 77194 |
+
"loss": 2.4026,
|
| 77195 |
+
"step": 7566
|
| 77196 |
+
},
|
| 77197 |
+
{
|
| 77198 |
+
"epoch": 0.9796420364436677,
|
| 77199 |
+
"grad_norm": 1.9015822937119669,
|
| 77200 |
+
"learning_rate": 1.5727955162539444e-09,
|
| 77201 |
+
"loss": 2.387,
|
| 77202 |
+
"step": 7567
|
| 77203 |
+
},
|
| 77204 |
+
{
|
| 77205 |
+
"epoch": 0.9797714988510211,
|
| 77206 |
+
"grad_norm": 1.1406419283583769,
|
| 77207 |
+
"learning_rate": 1.5528319864961672e-09,
|
| 77208 |
+
"loss": 2.1597,
|
| 77209 |
+
"step": 7568
|
| 77210 |
+
},
|
| 77211 |
+
{
|
| 77212 |
+
"epoch": 0.9799009612583746,
|
| 77213 |
+
"grad_norm": 2.1355981721634114,
|
| 77214 |
+
"learning_rate": 1.5329958084592255e-09,
|
| 77215 |
+
"loss": 2.3271,
|
| 77216 |
+
"step": 7569
|
| 77217 |
+
},
|
| 77218 |
+
{
|
| 77219 |
+
"epoch": 0.980030423665728,
|
| 77220 |
+
"grad_norm": 1.4628505282297621,
|
| 77221 |
+
"learning_rate": 1.5132869861948785e-09,
|
| 77222 |
+
"loss": 2.4343,
|
| 77223 |
+
"step": 7570
|
| 77224 |
+
},
|
| 77225 |
+
{
|
| 77226 |
+
"epoch": 0.980030423665728,
|
| 77227 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77228 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77229 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77230 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77231 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77232 |
+
"eval_PRM Precision": 1.0,
|
| 77233 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77234 |
+
"eval_PRM Specificty": 1.0,
|
| 77235 |
+
"eval_loss": 1.4957829713821411,
|
| 77236 |
+
"eval_runtime": 13.8423,
|
| 77237 |
+
"eval_samples_per_second": 3.179,
|
| 77238 |
+
"eval_steps_per_second": 0.217,
|
| 77239 |
+
"step": 7570
|
| 77240 |
+
},
|
| 77241 |
+
{
|
| 77242 |
+
"epoch": 0.9801598860730816,
|
| 77243 |
+
"grad_norm": 1.5826016277513821,
|
| 77244 |
+
"learning_rate": 1.4937055237292802e-09,
|
| 77245 |
+
"loss": 2.3635,
|
| 77246 |
+
"step": 7571
|
| 77247 |
+
},
|
| 77248 |
+
{
|
| 77249 |
+
"epoch": 0.980289348480435,
|
| 77250 |
+
"grad_norm": 1.2967653618589454,
|
| 77251 |
+
"learning_rate": 1.4742514250621481e-09,
|
| 77252 |
+
"loss": 2.2004,
|
| 77253 |
+
"step": 7572
|
| 77254 |
+
},
|
| 77255 |
+
{
|
| 77256 |
+
"epoch": 0.9804188108877885,
|
| 77257 |
+
"grad_norm": 1.348762418289272,
|
| 77258 |
+
"learning_rate": 1.454924694167456e-09,
|
| 77259 |
+
"loss": 2.439,
|
| 77260 |
+
"step": 7573
|
| 77261 |
+
},
|
| 77262 |
+
{
|
| 77263 |
+
"epoch": 0.9805482732951419,
|
| 77264 |
+
"grad_norm": 2.0059993316066187,
|
| 77265 |
+
"learning_rate": 1.4357253349931571e-09,
|
| 77266 |
+
"loss": 2.3491,
|
| 77267 |
+
"step": 7574
|
| 77268 |
+
},
|
| 77269 |
+
{
|
| 77270 |
+
"epoch": 0.9806777357024954,
|
| 77271 |
+
"grad_norm": 1.0696062980009842,
|
| 77272 |
+
"learning_rate": 1.4166533514609754e-09,
|
| 77273 |
+
"loss": 2.189,
|
| 77274 |
+
"step": 7575
|
| 77275 |
+
},
|
| 77276 |
+
{
|
| 77277 |
+
"epoch": 0.9806777357024954,
|
| 77278 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77279 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77280 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77281 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77282 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77283 |
+
"eval_PRM Precision": 1.0,
|
| 77284 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77285 |
+
"eval_PRM Specificty": 1.0,
|
| 77286 |
+
"eval_loss": 1.4971369504928589,
|
| 77287 |
+
"eval_runtime": 14.5855,
|
| 77288 |
+
"eval_samples_per_second": 3.017,
|
| 77289 |
+
"eval_steps_per_second": 0.206,
|
| 77290 |
+
"step": 7575
|
| 77291 |
+
},
|
| 77292 |
+
{
|
| 77293 |
+
"epoch": 0.9808071981098488,
|
| 77294 |
+
"grad_norm": 1.0643332651729815,
|
| 77295 |
+
"learning_rate": 1.3977087474667527e-09,
|
| 77296 |
+
"loss": 2.2942,
|
| 77297 |
+
"step": 7576
|
| 77298 |
+
},
|
| 77299 |
+
{
|
| 77300 |
+
"epoch": 0.9809366605172023,
|
| 77301 |
+
"grad_norm": 1.3466758091724915,
|
| 77302 |
+
"learning_rate": 1.3788915268804492e-09,
|
| 77303 |
+
"loss": 2.3136,
|
| 77304 |
+
"step": 7577
|
| 77305 |
+
},
|
| 77306 |
+
{
|
| 77307 |
+
"epoch": 0.9810661229245558,
|
| 77308 |
+
"grad_norm": 2.204496975992815,
|
| 77309 |
+
"learning_rate": 1.3602016935457262e-09,
|
| 77310 |
+
"loss": 2.4861,
|
| 77311 |
+
"step": 7578
|
| 77312 |
+
},
|
| 77313 |
+
{
|
| 77314 |
+
"epoch": 0.9811955853319092,
|
| 77315 |
+
"grad_norm": 1.414782482719819,
|
| 77316 |
+
"learning_rate": 1.3416392512804327e-09,
|
| 77317 |
+
"loss": 2.283,
|
| 77318 |
+
"step": 7579
|
| 77319 |
+
},
|
| 77320 |
+
{
|
| 77321 |
+
"epoch": 0.9813250477392627,
|
| 77322 |
+
"grad_norm": 2.985776601234628,
|
| 77323 |
+
"learning_rate": 1.3232042038762577e-09,
|
| 77324 |
+
"loss": 2.5293,
|
| 77325 |
+
"step": 7580
|
| 77326 |
+
},
|
| 77327 |
+
{
|
| 77328 |
+
"epoch": 0.9813250477392627,
|
| 77329 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77330 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77331 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77332 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77333 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77334 |
+
"eval_PRM Precision": 1.0,
|
| 77335 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77336 |
+
"eval_PRM Specificty": 1.0,
|
| 77337 |
+
"eval_loss": 1.4950283765792847,
|
| 77338 |
+
"eval_runtime": 14.9875,
|
| 77339 |
+
"eval_samples_per_second": 2.936,
|
| 77340 |
+
"eval_steps_per_second": 0.2,
|
| 77341 |
+
"step": 7580
|
| 77342 |
+
},
|
| 77343 |
+
{
|
| 77344 |
+
"epoch": 0.9814545101466162,
|
| 77345 |
+
"grad_norm": 2.130158472795342,
|
| 77346 |
+
"learning_rate": 1.3048965550990083e-09,
|
| 77347 |
+
"loss": 2.3813,
|
| 77348 |
+
"step": 7581
|
| 77349 |
+
},
|
| 77350 |
+
{
|
| 77351 |
+
"epoch": 0.9815839725539697,
|
| 77352 |
+
"grad_norm": 1.623329970728458,
|
| 77353 |
+
"learning_rate": 1.2867163086884015e-09,
|
| 77354 |
+
"loss": 2.3645,
|
| 77355 |
+
"step": 7582
|
| 77356 |
+
},
|
| 77357 |
+
{
|
| 77358 |
+
"epoch": 0.9817134349613231,
|
| 77359 |
+
"grad_norm": 1.9459265474325154,
|
| 77360 |
+
"learning_rate": 1.268663468358064e-09,
|
| 77361 |
+
"loss": 2.3744,
|
| 77362 |
+
"step": 7583
|
| 77363 |
+
},
|
| 77364 |
+
{
|
| 77365 |
+
"epoch": 0.9818428973686766,
|
| 77366 |
+
"grad_norm": 2.1195756554460385,
|
| 77367 |
+
"learning_rate": 1.2507380377956012e-09,
|
| 77368 |
+
"loss": 2.4797,
|
| 77369 |
+
"step": 7584
|
| 77370 |
+
},
|
| 77371 |
+
{
|
| 77372 |
+
"epoch": 0.98197235977603,
|
| 77373 |
+
"grad_norm": 1.7386627470623568,
|
| 77374 |
+
"learning_rate": 1.2329400206628068e-09,
|
| 77375 |
+
"loss": 2.4109,
|
| 77376 |
+
"step": 7585
|
| 77377 |
+
},
|
| 77378 |
+
{
|
| 77379 |
+
"epoch": 0.98197235977603,
|
| 77380 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77381 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77382 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77383 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77384 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77385 |
+
"eval_PRM Precision": 1.0,
|
| 77386 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77387 |
+
"eval_PRM Specificty": 1.0,
|
| 77388 |
+
"eval_loss": 1.4935635328292847,
|
| 77389 |
+
"eval_runtime": 13.6963,
|
| 77390 |
+
"eval_samples_per_second": 3.213,
|
| 77391 |
+
"eval_steps_per_second": 0.219,
|
| 77392 |
+
"step": 7585
|
| 77393 |
+
},
|
| 77394 |
+
{
|
| 77395 |
+
"epoch": 0.9821018221833835,
|
| 77396 |
+
"grad_norm": 1.1887944261227992,
|
| 77397 |
+
"learning_rate": 1.2152694205951054e-09,
|
| 77398 |
+
"loss": 2.2488,
|
| 77399 |
+
"step": 7586
|
| 77400 |
+
},
|
| 77401 |
+
{
|
| 77402 |
+
"epoch": 0.9822312845907369,
|
| 77403 |
+
"grad_norm": 2.586722476127736,
|
| 77404 |
+
"learning_rate": 1.1977262412022489e-09,
|
| 77405 |
+
"loss": 2.4194,
|
| 77406 |
+
"step": 7587
|
| 77407 |
+
},
|
| 77408 |
+
{
|
| 77409 |
+
"epoch": 0.9823607469980904,
|
| 77410 |
+
"grad_norm": 2.4786217471919127,
|
| 77411 |
+
"learning_rate": 1.18031048606769e-09,
|
| 77412 |
+
"loss": 2.8074,
|
| 77413 |
+
"step": 7588
|
| 77414 |
+
},
|
| 77415 |
+
{
|
| 77416 |
+
"epoch": 0.9824902094054438,
|
| 77417 |
+
"grad_norm": 3.7031030228112605,
|
| 77418 |
+
"learning_rate": 1.1630221587489304e-09,
|
| 77419 |
+
"loss": 2.2295,
|
| 77420 |
+
"step": 7589
|
| 77421 |
+
},
|
| 77422 |
+
{
|
| 77423 |
+
"epoch": 0.9826196718127974,
|
| 77424 |
+
"grad_norm": 2.26698201879527,
|
| 77425 |
+
"learning_rate": 1.1458612627775201e-09,
|
| 77426 |
+
"loss": 2.2422,
|
| 77427 |
+
"step": 7590
|
| 77428 |
+
},
|
| 77429 |
+
{
|
| 77430 |
+
"epoch": 0.9826196718127974,
|
| 77431 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77432 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77433 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77434 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77435 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77436 |
+
"eval_PRM Precision": 1.0,
|
| 77437 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77438 |
+
"eval_PRM Specificty": 1.0,
|
| 77439 |
+
"eval_loss": 1.4950283765792847,
|
| 77440 |
+
"eval_runtime": 14.7278,
|
| 77441 |
+
"eval_samples_per_second": 2.988,
|
| 77442 |
+
"eval_steps_per_second": 0.204,
|
| 77443 |
+
"step": 7590
|
| 77444 |
+
},
|
| 77445 |
+
{
|
| 77446 |
+
"epoch": 0.9827491342201509,
|
| 77447 |
+
"grad_norm": 1.8635644563124136,
|
| 77448 |
+
"learning_rate": 1.1288278016588493e-09,
|
| 77449 |
+
"loss": 2.2876,
|
| 77450 |
+
"step": 7591
|
| 77451 |
+
},
|
| 77452 |
+
{
|
| 77453 |
+
"epoch": 0.9828785966275043,
|
| 77454 |
+
"grad_norm": 3.6206585566509584,
|
| 77455 |
+
"learning_rate": 1.1119217788724268e-09,
|
| 77456 |
+
"loss": 2.5964,
|
| 77457 |
+
"step": 7592
|
| 77458 |
+
},
|
| 77459 |
+
{
|
| 77460 |
+
"epoch": 0.9830080590348578,
|
| 77461 |
+
"grad_norm": 1.868015361402047,
|
| 77462 |
+
"learning_rate": 1.095143197871601e-09,
|
| 77463 |
+
"loss": 2.5551,
|
| 77464 |
+
"step": 7593
|
| 77465 |
+
},
|
| 77466 |
+
{
|
| 77467 |
+
"epoch": 0.9831375214422112,
|
| 77468 |
+
"grad_norm": 1.266705226144988,
|
| 77469 |
+
"learning_rate": 1.078492062083769e-09,
|
| 77470 |
+
"loss": 2.3367,
|
| 77471 |
+
"step": 7594
|
| 77472 |
+
},
|
| 77473 |
+
{
|
| 77474 |
+
"epoch": 0.9832669838495647,
|
| 77475 |
+
"grad_norm": 1.8780513790793911,
|
| 77476 |
+
"learning_rate": 1.0619683749102384e-09,
|
| 77477 |
+
"loss": 2.2266,
|
| 77478 |
+
"step": 7595
|
| 77479 |
+
},
|
| 77480 |
+
{
|
| 77481 |
+
"epoch": 0.9832669838495647,
|
| 77482 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77483 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77484 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77485 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77486 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77487 |
+
"eval_PRM Precision": 1.0,
|
| 77488 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77489 |
+
"eval_PRM Specificty": 1.0,
|
| 77490 |
+
"eval_loss": 1.4957386255264282,
|
| 77491 |
+
"eval_runtime": 14.1659,
|
| 77492 |
+
"eval_samples_per_second": 3.106,
|
| 77493 |
+
"eval_steps_per_second": 0.212,
|
| 77494 |
+
"step": 7595
|
| 77495 |
+
},
|
| 77496 |
+
{
|
| 77497 |
+
"epoch": 0.9833964462569181,
|
| 77498 |
+
"grad_norm": 2.314347414973597,
|
| 77499 |
+
"learning_rate": 1.045572139726364e-09,
|
| 77500 |
+
"loss": 2.3955,
|
| 77501 |
+
"step": 7596
|
| 77502 |
+
},
|
| 77503 |
+
{
|
| 77504 |
+
"epoch": 0.9835259086642716,
|
| 77505 |
+
"grad_norm": 2.6351190170336465,
|
| 77506 |
+
"learning_rate": 1.029303359881273e-09,
|
| 77507 |
+
"loss": 2.2913,
|
| 77508 |
+
"step": 7597
|
| 77509 |
+
},
|
| 77510 |
+
{
|
| 77511 |
+
"epoch": 0.983655371071625,
|
| 77512 |
+
"grad_norm": 1.7546872803255078,
|
| 77513 |
+
"learning_rate": 1.0131620386983482e-09,
|
| 77514 |
+
"loss": 2.4309,
|
| 77515 |
+
"step": 7598
|
| 77516 |
+
},
|
| 77517 |
+
{
|
| 77518 |
+
"epoch": 0.9837848334789785,
|
| 77519 |
+
"grad_norm": 2.3215434065106426,
|
| 77520 |
+
"learning_rate": 9.971481794746745e-10,
|
| 77521 |
+
"loss": 2.1787,
|
| 77522 |
+
"step": 7599
|
| 77523 |
+
},
|
| 77524 |
+
{
|
| 77525 |
+
"epoch": 0.9839142958863321,
|
| 77526 |
+
"grad_norm": 1.2198424416141735,
|
| 77527 |
+
"learning_rate": 9.812617854814549e-10,
|
| 77528 |
+
"loss": 2.3247,
|
| 77529 |
+
"step": 7600
|
| 77530 |
+
},
|
| 77531 |
+
{
|
| 77532 |
+
"epoch": 0.9839142958863321,
|
| 77533 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77534 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77535 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77536 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77537 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77538 |
+
"eval_PRM Precision": 1.0,
|
| 77539 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77540 |
+
"eval_PRM Specificty": 1.0,
|
| 77541 |
+
"eval_loss": 1.4950283765792847,
|
| 77542 |
+
"eval_runtime": 14.5217,
|
| 77543 |
+
"eval_samples_per_second": 3.03,
|
| 77544 |
+
"eval_steps_per_second": 0.207,
|
| 77545 |
+
"step": 7600
|
| 77546 |
+
},
|
| 77547 |
+
{
|
| 77548 |
+
"epoch": 0.9840437582936855,
|
| 77549 |
+
"grad_norm": 1.519794772751305,
|
| 77550 |
+
"learning_rate": 9.655028599638017e-10,
|
| 77551 |
+
"loss": 2.5454,
|
| 77552 |
+
"step": 7601
|
| 77553 |
+
},
|
| 77554 |
+
{
|
| 77555 |
+
"epoch": 0.984173220701039,
|
| 77556 |
+
"grad_norm": 1.2600965157647652,
|
| 77557 |
+
"learning_rate": 9.498714061408065e-10,
|
| 77558 |
+
"loss": 2.21,
|
| 77559 |
+
"step": 7602
|
| 77560 |
+
},
|
| 77561 |
+
{
|
| 77562 |
+
"epoch": 0.9843026831083924,
|
| 77563 |
+
"grad_norm": 2.838040649791305,
|
| 77564 |
+
"learning_rate": 9.343674272054709e-10,
|
| 77565 |
+
"loss": 2.4016,
|
| 77566 |
+
"step": 7603
|
| 77567 |
+
},
|
| 77568 |
+
{
|
| 77569 |
+
"epoch": 0.9844321455157459,
|
| 77570 |
+
"grad_norm": 1.6408515614253465,
|
| 77571 |
+
"learning_rate": 9.189909263248448e-10,
|
| 77572 |
+
"loss": 2.3752,
|
| 77573 |
+
"step": 7604
|
| 77574 |
+
},
|
| 77575 |
+
{
|
| 77576 |
+
"epoch": 0.9845616079230993,
|
| 77577 |
+
"grad_norm": 2.1381858784410843,
|
| 77578 |
+
"learning_rate": 9.037419066398878e-10,
|
| 77579 |
+
"loss": 2.5369,
|
| 77580 |
+
"step": 7605
|
| 77581 |
+
},
|
| 77582 |
+
{
|
| 77583 |
+
"epoch": 0.9845616079230993,
|
| 77584 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77585 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77586 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77587 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77588 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77589 |
+
"eval_PRM Precision": 1.0,
|
| 77590 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77591 |
+
"eval_PRM Specificty": 1.0,
|
| 77592 |
+
"eval_loss": 1.4979802370071411,
|
| 77593 |
+
"eval_runtime": 13.3075,
|
| 77594 |
+
"eval_samples_per_second": 3.306,
|
| 77595 |
+
"eval_steps_per_second": 0.225,
|
| 77596 |
+
"step": 7605
|
| 77597 |
+
},
|
| 77598 |
+
{
|
| 77599 |
+
"epoch": 0.9846910703304528,
|
| 77600 |
+
"grad_norm": 1.5820801596214173,
|
| 77601 |
+
"learning_rate": 8.886203712654695e-10,
|
| 77602 |
+
"loss": 2.5869,
|
| 77603 |
+
"step": 7606
|
| 77604 |
+
},
|
| 77605 |
+
{
|
| 77606 |
+
"epoch": 0.9848205327378062,
|
| 77607 |
+
"grad_norm": 3.27107927874882,
|
| 77608 |
+
"learning_rate": 8.736263232905079e-10,
|
| 77609 |
+
"loss": 2.4082,
|
| 77610 |
+
"step": 7607
|
| 77611 |
+
},
|
| 77612 |
+
{
|
| 77613 |
+
"epoch": 0.9849499951451597,
|
| 77614 |
+
"grad_norm": 1.6374668708983278,
|
| 77615 |
+
"learning_rate": 8.587597657778307e-10,
|
| 77616 |
+
"loss": 2.4065,
|
| 77617 |
+
"step": 7608
|
| 77618 |
+
},
|
| 77619 |
+
{
|
| 77620 |
+
"epoch": 0.9850794575525131,
|
| 77621 |
+
"grad_norm": 1.727336438934298,
|
| 77622 |
+
"learning_rate": 8.440207017641755e-10,
|
| 77623 |
+
"loss": 2.5813,
|
| 77624 |
+
"step": 7609
|
| 77625 |
+
},
|
| 77626 |
+
{
|
| 77627 |
+
"epoch": 0.9852089199598667,
|
| 77628 |
+
"grad_norm": 1.6646088684556826,
|
| 77629 |
+
"learning_rate": 8.294091342604671e-10,
|
| 77630 |
+
"loss": 2.4685,
|
| 77631 |
+
"step": 7610
|
| 77632 |
+
},
|
| 77633 |
+
{
|
| 77634 |
+
"epoch": 0.9852089199598667,
|
| 77635 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77636 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77637 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77638 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77639 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77640 |
+
"eval_PRM Precision": 1.0,
|
| 77641 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77642 |
+
"eval_PRM Specificty": 1.0,
|
| 77643 |
+
"eval_loss": 1.4964932203292847,
|
| 77644 |
+
"eval_runtime": 14.4666,
|
| 77645 |
+
"eval_samples_per_second": 3.041,
|
| 77646 |
+
"eval_steps_per_second": 0.207,
|
| 77647 |
+
"step": 7610
|
| 77648 |
+
},
|
| 77649 |
+
{
|
| 77650 |
+
"epoch": 0.9853383823672202,
|
| 77651 |
+
"grad_norm": 1.6183130409645539,
|
| 77652 |
+
"learning_rate": 8.149250662512626e-10,
|
| 77653 |
+
"loss": 2.2349,
|
| 77654 |
+
"step": 7611
|
| 77655 |
+
},
|
| 77656 |
+
{
|
| 77657 |
+
"epoch": 0.9854678447745736,
|
| 77658 |
+
"grad_norm": 1.9724083536469152,
|
| 77659 |
+
"learning_rate": 8.005685006952368e-10,
|
| 77660 |
+
"loss": 2.4695,
|
| 77661 |
+
"step": 7612
|
| 77662 |
+
},
|
| 77663 |
+
{
|
| 77664 |
+
"epoch": 0.9855973071819271,
|
| 77665 |
+
"grad_norm": 1.3922509704379329,
|
| 77666 |
+
"learning_rate": 7.863394405251135e-10,
|
| 77667 |
+
"loss": 2.374,
|
| 77668 |
+
"step": 7613
|
| 77669 |
+
},
|
| 77670 |
+
{
|
| 77671 |
+
"epoch": 0.9857267695892805,
|
| 77672 |
+
"grad_norm": 2.1223017834335094,
|
| 77673 |
+
"learning_rate": 7.722378886473176e-10,
|
| 77674 |
+
"loss": 2.1777,
|
| 77675 |
+
"step": 7614
|
| 77676 |
+
},
|
| 77677 |
+
{
|
| 77678 |
+
"epoch": 0.985856231996634,
|
| 77679 |
+
"grad_norm": 2.718200331729673,
|
| 77680 |
+
"learning_rate": 7.582638479424614e-10,
|
| 77681 |
+
"loss": 2.4619,
|
| 77682 |
+
"step": 7615
|
| 77683 |
+
},
|
| 77684 |
+
{
|
| 77685 |
+
"epoch": 0.985856231996634,
|
| 77686 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77687 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77688 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77689 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77690 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77691 |
+
"eval_PRM Precision": 1.0,
|
| 77692 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77693 |
+
"eval_PRM Specificty": 1.0,
|
| 77694 |
+
"eval_loss": 1.493607997894287,
|
| 77695 |
+
"eval_runtime": 15.0409,
|
| 77696 |
+
"eval_samples_per_second": 2.925,
|
| 77697 |
+
"eval_steps_per_second": 0.199,
|
| 77698 |
+
"step": 7615
|
| 77699 |
+
},
|
| 77700 |
+
{
|
| 77701 |
+
"epoch": 0.9859856944039874,
|
| 77702 |
+
"grad_norm": 1.8789229257617968,
|
| 77703 |
+
"learning_rate": 7.444173212650673e-10,
|
| 77704 |
+
"loss": 2.3359,
|
| 77705 |
+
"step": 7616
|
| 77706 |
+
},
|
| 77707 |
+
{
|
| 77708 |
+
"epoch": 0.9861151568113409,
|
| 77709 |
+
"grad_norm": 2.58106500301192,
|
| 77710 |
+
"learning_rate": 7.306983114434979e-10,
|
| 77711 |
+
"loss": 2.5012,
|
| 77712 |
+
"step": 7617
|
| 77713 |
+
},
|
| 77714 |
+
{
|
| 77715 |
+
"epoch": 0.9862446192186943,
|
| 77716 |
+
"grad_norm": 1.7565557791717235,
|
| 77717 |
+
"learning_rate": 7.171068212801641e-10,
|
| 77718 |
+
"loss": 2.3738,
|
| 77719 |
+
"step": 7618
|
| 77720 |
+
},
|
| 77721 |
+
{
|
| 77722 |
+
"epoch": 0.9863740816260478,
|
| 77723 |
+
"grad_norm": 2.554639240543607,
|
| 77724 |
+
"learning_rate": 7.036428535513867e-10,
|
| 77725 |
+
"loss": 2.5596,
|
| 77726 |
+
"step": 7619
|
| 77727 |
+
},
|
| 77728 |
+
{
|
| 77729 |
+
"epoch": 0.9865035440334013,
|
| 77730 |
+
"grad_norm": 2.2603757514179184,
|
| 77731 |
+
"learning_rate": 6.903064110074659e-10,
|
| 77732 |
+
"loss": 2.4717,
|
| 77733 |
+
"step": 7620
|
| 77734 |
+
},
|
| 77735 |
+
{
|
| 77736 |
+
"epoch": 0.9865035440334013,
|
| 77737 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77738 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77739 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77740 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77741 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77742 |
+
"eval_PRM Precision": 1.0,
|
| 77743 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77744 |
+
"eval_PRM Specificty": 1.0,
|
| 77745 |
+
"eval_loss": 1.4978693723678589,
|
| 77746 |
+
"eval_runtime": 14.3112,
|
| 77747 |
+
"eval_samples_per_second": 3.075,
|
| 77748 |
+
"eval_steps_per_second": 0.21,
|
| 77749 |
+
"step": 7620
|
| 77750 |
+
},
|
| 77751 |
+
{
|
| 77752 |
+
"epoch": 0.9866330064407548,
|
| 77753 |
+
"grad_norm": 1.848935015750131,
|
| 77754 |
+
"learning_rate": 6.770974963726807e-10,
|
| 77755 |
+
"loss": 2.4548,
|
| 77756 |
+
"step": 7621
|
| 77757 |
+
},
|
| 77758 |
+
{
|
| 77759 |
+
"epoch": 0.9867624688481083,
|
| 77760 |
+
"grad_norm": 1.4956095265091809,
|
| 77761 |
+
"learning_rate": 6.640161123451505e-10,
|
| 77762 |
+
"loss": 2.1816,
|
| 77763 |
+
"step": 7622
|
| 77764 |
+
},
|
| 77765 |
+
{
|
| 77766 |
+
"epoch": 0.9868919312554617,
|
| 77767 |
+
"grad_norm": 1.630094864183574,
|
| 77768 |
+
"learning_rate": 6.510622615970436e-10,
|
| 77769 |
+
"loss": 2.3577,
|
| 77770 |
+
"step": 7623
|
| 77771 |
+
},
|
| 77772 |
+
{
|
| 77773 |
+
"epoch": 0.9870213936628152,
|
| 77774 |
+
"grad_norm": 3.1326216222556083,
|
| 77775 |
+
"learning_rate": 6.382359467744375e-10,
|
| 77776 |
+
"loss": 2.3452,
|
| 77777 |
+
"step": 7624
|
| 77778 |
+
},
|
| 77779 |
+
{
|
| 77780 |
+
"epoch": 0.9871508560701686,
|
| 77781 |
+
"grad_norm": 2.2543400031695144,
|
| 77782 |
+
"learning_rate": 6.255371704973895e-10,
|
| 77783 |
+
"loss": 2.4229,
|
| 77784 |
+
"step": 7625
|
| 77785 |
+
},
|
| 77786 |
+
{
|
| 77787 |
+
"epoch": 0.9871508560701686,
|
| 77788 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77789 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77790 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77791 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77792 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77793 |
+
"eval_PRM Precision": 1.0,
|
| 77794 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77795 |
+
"eval_PRM Specificty": 1.0,
|
| 77796 |
+
"eval_loss": 1.4950283765792847,
|
| 77797 |
+
"eval_runtime": 15.7043,
|
| 77798 |
+
"eval_samples_per_second": 2.802,
|
| 77799 |
+
"eval_steps_per_second": 0.191,
|
| 77800 |
+
"step": 7625
|
| 77801 |
+
},
|
| 77802 |
+
{
|
| 77803 |
+
"epoch": 0.9872803184775221,
|
| 77804 |
+
"grad_norm": 1.4943666856649598,
|
| 77805 |
+
"learning_rate": 6.129659353597967e-10,
|
| 77806 |
+
"loss": 2.377,
|
| 77807 |
+
"step": 7626
|
| 77808 |
+
},
|
| 77809 |
+
{
|
| 77810 |
+
"epoch": 0.9874097808848755,
|
| 77811 |
+
"grad_norm": 1.6970512279219263,
|
| 77812 |
+
"learning_rate": 6.005222439297437e-10,
|
| 77813 |
+
"loss": 2.458,
|
| 77814 |
+
"step": 7627
|
| 77815 |
+
},
|
| 77816 |
+
{
|
| 77817 |
+
"epoch": 0.987539243292229,
|
| 77818 |
+
"grad_norm": 1.360629472544963,
|
| 77819 |
+
"learning_rate": 5.882060987490168e-10,
|
| 77820 |
+
"loss": 2.2747,
|
| 77821 |
+
"step": 7628
|
| 77822 |
+
},
|
| 77823 |
+
{
|
| 77824 |
+
"epoch": 0.9876687056995825,
|
| 77825 |
+
"grad_norm": 1.947585735123565,
|
| 77826 |
+
"learning_rate": 5.760175023334507e-10,
|
| 77827 |
+
"loss": 2.4478,
|
| 77828 |
+
"step": 7629
|
| 77829 |
+
},
|
| 77830 |
+
{
|
| 77831 |
+
"epoch": 0.987798168106936,
|
| 77832 |
+
"grad_norm": 1.506017763428392,
|
| 77833 |
+
"learning_rate": 5.639564571727902e-10,
|
| 77834 |
+
"loss": 2.4949,
|
| 77835 |
+
"step": 7630
|
| 77836 |
+
},
|
| 77837 |
+
{
|
| 77838 |
+
"epoch": 0.987798168106936,
|
| 77839 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77840 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77841 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77842 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77843 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77844 |
+
"eval_PRM Precision": 1.0,
|
| 77845 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77846 |
+
"eval_PRM Specificty": 1.0,
|
| 77847 |
+
"eval_loss": 1.4949840307235718,
|
| 77848 |
+
"eval_runtime": 14.2747,
|
| 77849 |
+
"eval_samples_per_second": 3.082,
|
| 77850 |
+
"eval_steps_per_second": 0.21,
|
| 77851 |
+
"step": 7630
|
| 77852 |
+
},
|
| 77853 |
+
{
|
| 77854 |
+
"epoch": 0.9879276305142894,
|
| 77855 |
+
"grad_norm": 2.42281583678435,
|
| 77856 |
+
"learning_rate": 5.520229657307585e-10,
|
| 77857 |
+
"loss": 2.3638,
|
| 77858 |
+
"step": 7631
|
| 77859 |
+
},
|
| 77860 |
+
{
|
| 77861 |
+
"epoch": 0.9880570929216429,
|
| 77862 |
+
"grad_norm": 1.5200091251742374,
|
| 77863 |
+
"learning_rate": 5.402170304449894e-10,
|
| 77864 |
+
"loss": 2.3445,
|
| 77865 |
+
"step": 7632
|
| 77866 |
+
},
|
| 77867 |
+
{
|
| 77868 |
+
"epoch": 0.9881865553289964,
|
| 77869 |
+
"grad_norm": 3.56315214653513,
|
| 77870 |
+
"learning_rate": 5.285386537271647e-10,
|
| 77871 |
+
"loss": 2.4712,
|
| 77872 |
+
"step": 7633
|
| 77873 |
+
},
|
| 77874 |
+
{
|
| 77875 |
+
"epoch": 0.9883160177363498,
|
| 77876 |
+
"grad_norm": 1.4993946555982482,
|
| 77877 |
+
"learning_rate": 5.169878379628069e-10,
|
| 77878 |
+
"loss": 2.3345,
|
| 77879 |
+
"step": 7634
|
| 77880 |
+
},
|
| 77881 |
+
{
|
| 77882 |
+
"epoch": 0.9884454801437033,
|
| 77883 |
+
"grad_norm": 1.3766723430853567,
|
| 77884 |
+
"learning_rate": 5.055645855112784e-10,
|
| 77885 |
+
"loss": 2.2922,
|
| 77886 |
+
"step": 7635
|
| 77887 |
+
},
|
| 77888 |
+
{
|
| 77889 |
+
"epoch": 0.9884454801437033,
|
| 77890 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77891 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77892 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77893 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77894 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77895 |
+
"eval_PRM Precision": 1.0,
|
| 77896 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77897 |
+
"eval_PRM Specificty": 1.0,
|
| 77898 |
+
"eval_loss": 1.4942737817764282,
|
| 77899 |
+
"eval_runtime": 13.9902,
|
| 77900 |
+
"eval_samples_per_second": 3.145,
|
| 77901 |
+
"eval_steps_per_second": 0.214,
|
| 77902 |
+
"step": 7635
|
| 77903 |
+
},
|
| 77904 |
+
{
|
| 77905 |
+
"epoch": 0.9885749425510567,
|
| 77906 |
+
"grad_norm": 2.765105962883385,
|
| 77907 |
+
"learning_rate": 4.942688987061989e-10,
|
| 77908 |
+
"loss": 2.948,
|
| 77909 |
+
"step": 7636
|
| 77910 |
+
},
|
| 77911 |
+
{
|
| 77912 |
+
"epoch": 0.9887044049584102,
|
| 77913 |
+
"grad_norm": 1.7884727220792906,
|
| 77914 |
+
"learning_rate": 4.831007798548199e-10,
|
| 77915 |
+
"loss": 2.4609,
|
| 77916 |
+
"step": 7637
|
| 77917 |
+
},
|
| 77918 |
+
{
|
| 77919 |
+
"epoch": 0.9888338673657636,
|
| 77920 |
+
"grad_norm": 2.585771279521546,
|
| 77921 |
+
"learning_rate": 4.720602312385109e-10,
|
| 77922 |
+
"loss": 2.498,
|
| 77923 |
+
"step": 7638
|
| 77924 |
+
},
|
| 77925 |
+
{
|
| 77926 |
+
"epoch": 0.9889633297731172,
|
| 77927 |
+
"grad_norm": 1.6577037565047226,
|
| 77928 |
+
"learning_rate": 4.611472551124818e-10,
|
| 77929 |
+
"loss": 2.323,
|
| 77930 |
+
"step": 7639
|
| 77931 |
+
},
|
| 77932 |
+
{
|
| 77933 |
+
"epoch": 0.9890927921804706,
|
| 77934 |
+
"grad_norm": 2.742440349149222,
|
| 77935 |
+
"learning_rate": 4.5036185370592176e-10,
|
| 77936 |
+
"loss": 2.7051,
|
| 77937 |
+
"step": 7640
|
| 77938 |
+
},
|
| 77939 |
+
{
|
| 77940 |
+
"epoch": 0.9890927921804706,
|
| 77941 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77942 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77943 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77944 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77945 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77946 |
+
"eval_PRM Precision": 1.0,
|
| 77947 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77948 |
+
"eval_PRM Specificty": 1.0,
|
| 77949 |
+
"eval_loss": 1.4964932203292847,
|
| 77950 |
+
"eval_runtime": 14.2938,
|
| 77951 |
+
"eval_samples_per_second": 3.078,
|
| 77952 |
+
"eval_steps_per_second": 0.21,
|
| 77953 |
+
"step": 7640
|
| 77954 |
+
},
|
| 77955 |
+
{
|
| 77956 |
+
"epoch": 0.9892222545878241,
|
| 77957 |
+
"grad_norm": 1.899570315318355,
|
| 77958 |
+
"learning_rate": 4.3970402922199895e-10,
|
| 77959 |
+
"loss": 2.3279,
|
| 77960 |
+
"step": 7641
|
| 77961 |
+
},
|
| 77962 |
+
{
|
| 77963 |
+
"epoch": 0.9893517169951775,
|
| 77964 |
+
"grad_norm": 1.429668343151256,
|
| 77965 |
+
"learning_rate": 4.291737838378607e-10,
|
| 77966 |
+
"loss": 2.2539,
|
| 77967 |
+
"step": 7642
|
| 77968 |
+
},
|
| 77969 |
+
{
|
| 77970 |
+
"epoch": 0.989481179402531,
|
| 77971 |
+
"grad_norm": 1.666726346466479,
|
| 77972 |
+
"learning_rate": 4.187711197043559e-10,
|
| 77973 |
+
"loss": 2.2537,
|
| 77974 |
+
"step": 7643
|
| 77975 |
+
},
|
| 77976 |
+
{
|
| 77977 |
+
"epoch": 0.9896106418098845,
|
| 77978 |
+
"grad_norm": 1.7583614698230214,
|
| 77979 |
+
"learning_rate": 4.0849603894659027e-10,
|
| 77980 |
+
"loss": 2.3398,
|
| 77981 |
+
"step": 7644
|
| 77982 |
+
},
|
| 77983 |
+
{
|
| 77984 |
+
"epoch": 0.9897401042172379,
|
| 77985 |
+
"grad_norm": 1.218907723659834,
|
| 77986 |
+
"learning_rate": 3.9834854366344044e-10,
|
| 77987 |
+
"loss": 2.1571,
|
| 77988 |
+
"step": 7645
|
| 77989 |
+
},
|
| 77990 |
+
{
|
| 77991 |
+
"epoch": 0.9897401042172379,
|
| 77992 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 77993 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 77994 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 77995 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 77996 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 77997 |
+
"eval_PRM Precision": 1.0,
|
| 77998 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 77999 |
+
"eval_PRM Specificty": 1.0,
|
| 78000 |
+
"eval_loss": 1.4964932203292847,
|
| 78001 |
+
"eval_runtime": 13.6451,
|
| 78002 |
+
"eval_samples_per_second": 3.225,
|
| 78003 |
+
"eval_steps_per_second": 0.22,
|
| 78004 |
+
"step": 7645
|
| 78005 |
+
},
|
| 78006 |
+
{
|
| 78007 |
+
"epoch": 0.9898695666245914,
|
| 78008 |
+
"grad_norm": 2.16683363153041,
|
| 78009 |
+
"learning_rate": 3.8832863592762336e-10,
|
| 78010 |
+
"loss": 2.4543,
|
| 78011 |
+
"step": 7646
|
| 78012 |
+
},
|
| 78013 |
+
{
|
| 78014 |
+
"epoch": 0.9899990290319448,
|
| 78015 |
+
"grad_norm": 1.8880776975893787,
|
| 78016 |
+
"learning_rate": 3.7843631778604337e-10,
|
| 78017 |
+
"loss": 2.314,
|
| 78018 |
+
"step": 7647
|
| 78019 |
+
},
|
| 78020 |
+
{
|
| 78021 |
+
"epoch": 0.9901284914392983,
|
| 78022 |
+
"grad_norm": 1.913928145061539,
|
| 78023 |
+
"learning_rate": 3.686715912593064e-10,
|
| 78024 |
+
"loss": 2.385,
|
| 78025 |
+
"step": 7648
|
| 78026 |
+
},
|
| 78027 |
+
{
|
| 78028 |
+
"epoch": 0.9902579538466518,
|
| 78029 |
+
"grad_norm": 2.598615507596037,
|
| 78030 |
+
"learning_rate": 3.5903445834213625e-10,
|
| 78031 |
+
"loss": 2.4478,
|
| 78032 |
+
"step": 7649
|
| 78033 |
+
},
|
| 78034 |
+
{
|
| 78035 |
+
"epoch": 0.9903874162540053,
|
| 78036 |
+
"grad_norm": 2.781411612118827,
|
| 78037 |
+
"learning_rate": 3.4952492100302773e-10,
|
| 78038 |
+
"loss": 2.5112,
|
| 78039 |
+
"step": 7650
|
| 78040 |
+
},
|
| 78041 |
+
{
|
| 78042 |
+
"epoch": 0.9903874162540053,
|
| 78043 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78044 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78045 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78046 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78047 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78048 |
+
"eval_PRM Precision": 1.0,
|
| 78049 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78050 |
+
"eval_PRM Specificty": 1.0,
|
| 78051 |
+
"eval_loss": 1.4964488744735718,
|
| 78052 |
+
"eval_runtime": 14.049,
|
| 78053 |
+
"eval_samples_per_second": 3.132,
|
| 78054 |
+
"eval_steps_per_second": 0.214,
|
| 78055 |
+
"step": 7650
|
| 78056 |
+
},
|
| 78057 |
+
{
|
| 78058 |
+
"epoch": 0.9905168786613587,
|
| 78059 |
+
"grad_norm": 4.920852837984811,
|
| 78060 |
+
"learning_rate": 3.4014298118452415e-10,
|
| 78061 |
+
"loss": 2.8179,
|
| 78062 |
+
"step": 7651
|
| 78063 |
+
},
|
| 78064 |
+
{
|
| 78065 |
+
"epoch": 0.9906463410687122,
|
| 78066 |
+
"grad_norm": 1.602683531452493,
|
| 78067 |
+
"learning_rate": 3.308886408031481e-10,
|
| 78068 |
+
"loss": 2.4128,
|
| 78069 |
+
"step": 7652
|
| 78070 |
+
},
|
| 78071 |
+
{
|
| 78072 |
+
"epoch": 0.9907758034760656,
|
| 78073 |
+
"grad_norm": 2.1060923661609112,
|
| 78074 |
+
"learning_rate": 3.217619017492624e-10,
|
| 78075 |
+
"loss": 2.374,
|
| 78076 |
+
"step": 7653
|
| 78077 |
+
},
|
| 78078 |
+
{
|
| 78079 |
+
"epoch": 0.9909052658834191,
|
| 78080 |
+
"grad_norm": 1.6204569374519608,
|
| 78081 |
+
"learning_rate": 3.127627658870702e-10,
|
| 78082 |
+
"loss": 2.3461,
|
| 78083 |
+
"step": 7654
|
| 78084 |
+
},
|
| 78085 |
+
{
|
| 78086 |
+
"epoch": 0.9910347282907725,
|
| 78087 |
+
"grad_norm": 1.1008634986922998,
|
| 78088 |
+
"learning_rate": 3.038912350549622e-10,
|
| 78089 |
+
"loss": 2.2966,
|
| 78090 |
+
"step": 7655
|
| 78091 |
+
},
|
| 78092 |
+
{
|
| 78093 |
+
"epoch": 0.9910347282907725,
|
| 78094 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78095 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78096 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78097 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78098 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78099 |
+
"eval_PRM Precision": 1.0,
|
| 78100 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78101 |
+
"eval_PRM Specificty": 1.0,
|
| 78102 |
+
"eval_loss": 1.4950505495071411,
|
| 78103 |
+
"eval_runtime": 13.6045,
|
| 78104 |
+
"eval_samples_per_second": 3.234,
|
| 78105 |
+
"eval_steps_per_second": 0.221,
|
| 78106 |
+
"step": 7655
|
| 78107 |
+
},
|
| 78108 |
+
{
|
| 78109 |
+
"epoch": 0.991164190698126,
|
| 78110 |
+
"grad_norm": 2.1093204697180243,
|
| 78111 |
+
"learning_rate": 2.9514731106509985e-10,
|
| 78112 |
+
"loss": 2.3687,
|
| 78113 |
+
"step": 7656
|
| 78114 |
+
},
|
| 78115 |
+
{
|
| 78116 |
+
"epoch": 0.9912936531054795,
|
| 78117 |
+
"grad_norm": 1.9171259210801308,
|
| 78118 |
+
"learning_rate": 2.8653099570355447e-10,
|
| 78119 |
+
"loss": 2.4336,
|
| 78120 |
+
"step": 7657
|
| 78121 |
+
},
|
| 78122 |
+
{
|
| 78123 |
+
"epoch": 0.9914231155128329,
|
| 78124 |
+
"grad_norm": 2.3993710563557116,
|
| 78125 |
+
"learning_rate": 2.7804229073044584e-10,
|
| 78126 |
+
"loss": 2.3958,
|
| 78127 |
+
"step": 7658
|
| 78128 |
+
},
|
| 78129 |
+
{
|
| 78130 |
+
"epoch": 0.9915525779201865,
|
| 78131 |
+
"grad_norm": 1.2834745849996845,
|
| 78132 |
+
"learning_rate": 2.696811978797342e-10,
|
| 78133 |
+
"loss": 2.4238,
|
| 78134 |
+
"step": 7659
|
| 78135 |
+
},
|
| 78136 |
+
{
|
| 78137 |
+
"epoch": 0.9916820403275399,
|
| 78138 |
+
"grad_norm": 2.040530792128796,
|
| 78139 |
+
"learning_rate": 2.614477188593589e-10,
|
| 78140 |
+
"loss": 2.2217,
|
| 78141 |
+
"step": 7660
|
| 78142 |
+
},
|
| 78143 |
+
{
|
| 78144 |
+
"epoch": 0.9916820403275399,
|
| 78145 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78146 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78147 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78148 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78149 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78150 |
+
"eval_PRM Precision": 1.0,
|
| 78151 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78152 |
+
"eval_PRM Specificty": 1.0,
|
| 78153 |
+
"eval_loss": 1.498579502105713,
|
| 78154 |
+
"eval_runtime": 13.8118,
|
| 78155 |
+
"eval_samples_per_second": 3.186,
|
| 78156 |
+
"eval_steps_per_second": 0.217,
|
| 78157 |
+
"step": 7660
|
| 78158 |
+
},
|
| 78159 |
+
{
|
| 78160 |
+
"epoch": 0.9918115027348934,
|
| 78161 |
+
"grad_norm": 1.146395381273546,
|
| 78162 |
+
"learning_rate": 2.5334185535109967e-10,
|
| 78163 |
+
"loss": 2.1646,
|
| 78164 |
+
"step": 7661
|
| 78165 |
+
},
|
| 78166 |
+
{
|
| 78167 |
+
"epoch": 0.9919409651422468,
|
| 78168 |
+
"grad_norm": 1.1207392650082522,
|
| 78169 |
+
"learning_rate": 2.453636090108541e-10,
|
| 78170 |
+
"loss": 2.3267,
|
| 78171 |
+
"step": 7662
|
| 78172 |
+
},
|
| 78173 |
+
{
|
| 78174 |
+
"epoch": 0.9920704275496003,
|
| 78175 |
+
"grad_norm": 1.749398980875084,
|
| 78176 |
+
"learning_rate": 2.375129814682908e-10,
|
| 78177 |
+
"loss": 2.3125,
|
| 78178 |
+
"step": 7663
|
| 78179 |
+
},
|
| 78180 |
+
{
|
| 78181 |
+
"epoch": 0.9921998899569537,
|
| 78182 |
+
"grad_norm": 1.4514255061424015,
|
| 78183 |
+
"learning_rate": 2.2978997432705764e-10,
|
| 78184 |
+
"loss": 2.2827,
|
| 78185 |
+
"step": 7664
|
| 78186 |
+
},
|
| 78187 |
+
{
|
| 78188 |
+
"epoch": 0.9923293523643072,
|
| 78189 |
+
"grad_norm": 1.3744947085311505,
|
| 78190 |
+
"learning_rate": 2.2219458916471205e-10,
|
| 78191 |
+
"loss": 2.1985,
|
| 78192 |
+
"step": 7665
|
| 78193 |
+
},
|
| 78194 |
+
{
|
| 78195 |
+
"epoch": 0.9923293523643072,
|
| 78196 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78197 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78198 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78199 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78200 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78201 |
+
"eval_PRM Precision": 1.0,
|
| 78202 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78203 |
+
"eval_PRM Specificty": 1.0,
|
| 78204 |
+
"eval_loss": 1.4971369504928589,
|
| 78205 |
+
"eval_runtime": 13.8034,
|
| 78206 |
+
"eval_samples_per_second": 3.188,
|
| 78207 |
+
"eval_steps_per_second": 0.217,
|
| 78208 |
+
"step": 7665
|
| 78209 |
+
},
|
| 78210 |
+
{
|
| 78211 |
+
"epoch": 0.9924588147716606,
|
| 78212 |
+
"grad_norm": 1.9129455404788904,
|
| 78213 |
+
"learning_rate": 2.1472682753279073e-10,
|
| 78214 |
+
"loss": 2.4282,
|
| 78215 |
+
"step": 7666
|
| 78216 |
+
},
|
| 78217 |
+
{
|
| 78218 |
+
"epoch": 0.9925882771790141,
|
| 78219 |
+
"grad_norm": 1.6119870892933774,
|
| 78220 |
+
"learning_rate": 2.0738669095680952e-10,
|
| 78221 |
+
"loss": 2.4197,
|
| 78222 |
+
"step": 7667
|
| 78223 |
+
},
|
| 78224 |
+
{
|
| 78225 |
+
"epoch": 0.9927177395863676,
|
| 78226 |
+
"grad_norm": 2.025417941338564,
|
| 78227 |
+
"learning_rate": 2.0017418093598583e-10,
|
| 78228 |
+
"loss": 2.5481,
|
| 78229 |
+
"step": 7668
|
| 78230 |
+
},
|
| 78231 |
+
{
|
| 78232 |
+
"epoch": 0.9928472019937211,
|
| 78233 |
+
"grad_norm": 1.7189407225275821,
|
| 78234 |
+
"learning_rate": 1.9308929894372441e-10,
|
| 78235 |
+
"loss": 2.249,
|
| 78236 |
+
"step": 7669
|
| 78237 |
+
},
|
| 78238 |
+
{
|
| 78239 |
+
"epoch": 0.9929766644010746,
|
| 78240 |
+
"grad_norm": 2.6274370671024756,
|
| 78241 |
+
"learning_rate": 1.8613204642720096e-10,
|
| 78242 |
+
"loss": 2.5369,
|
| 78243 |
+
"step": 7670
|
| 78244 |
+
},
|
| 78245 |
+
{
|
| 78246 |
+
"epoch": 0.9929766644010746,
|
| 78247 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78248 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78249 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78250 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78251 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78252 |
+
"eval_PRM Precision": 1.0,
|
| 78253 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78254 |
+
"eval_PRM Specificty": 1.0,
|
| 78255 |
+
"eval_loss": 1.4986461400985718,
|
| 78256 |
+
"eval_runtime": 14.0005,
|
| 78257 |
+
"eval_samples_per_second": 3.143,
|
| 78258 |
+
"eval_steps_per_second": 0.214,
|
| 78259 |
+
"step": 7670
|
| 78260 |
+
},
|
| 78261 |
+
{
|
| 78262 |
+
"epoch": 0.993106126808428,
|
| 78263 |
+
"grad_norm": 1.733854755515088,
|
| 78264 |
+
"learning_rate": 1.793024248076397e-10,
|
| 78265 |
+
"loss": 2.4133,
|
| 78266 |
+
"step": 7671
|
| 78267 |
+
},
|
| 78268 |
+
{
|
| 78269 |
+
"epoch": 0.9932355892157815,
|
| 78270 |
+
"grad_norm": 3.5557057390425304,
|
| 78271 |
+
"learning_rate": 1.726004354801053e-10,
|
| 78272 |
+
"loss": 2.5322,
|
| 78273 |
+
"step": 7672
|
| 78274 |
+
},
|
| 78275 |
+
{
|
| 78276 |
+
"epoch": 0.9933650516231349,
|
| 78277 |
+
"grad_norm": 2.5351565687700592,
|
| 78278 |
+
"learning_rate": 1.660260798136415e-10,
|
| 78279 |
+
"loss": 2.5747,
|
| 78280 |
+
"step": 7673
|
| 78281 |
+
},
|
| 78282 |
+
{
|
| 78283 |
+
"epoch": 0.9934945140304884,
|
| 78284 |
+
"grad_norm": 2.155793966215189,
|
| 78285 |
+
"learning_rate": 1.5957935915113242e-10,
|
| 78286 |
+
"loss": 2.3379,
|
| 78287 |
+
"step": 7674
|
| 78288 |
+
},
|
| 78289 |
+
{
|
| 78290 |
+
"epoch": 0.9936239764378418,
|
| 78291 |
+
"grad_norm": 2.561628615143274,
|
| 78292 |
+
"learning_rate": 1.532602748094414e-10,
|
| 78293 |
+
"loss": 2.3025,
|
| 78294 |
+
"step": 7675
|
| 78295 |
+
},
|
| 78296 |
+
{
|
| 78297 |
+
"epoch": 0.9936239764378418,
|
| 78298 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78299 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78300 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78301 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78302 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78303 |
+
"eval_PRM Precision": 1.0,
|
| 78304 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78305 |
+
"eval_PRM Specificty": 1.0,
|
| 78306 |
+
"eval_loss": 1.4943625926971436,
|
| 78307 |
+
"eval_runtime": 14.2777,
|
| 78308 |
+
"eval_samples_per_second": 3.082,
|
| 78309 |
+
"eval_steps_per_second": 0.21,
|
| 78310 |
+
"step": 7675
|
| 78311 |
+
},
|
| 78312 |
+
{
|
| 78313 |
+
"epoch": 0.9937534388451953,
|
| 78314 |
+
"grad_norm": 1.4586702756577856,
|
| 78315 |
+
"learning_rate": 1.470688280794108e-10,
|
| 78316 |
+
"loss": 2.4033,
|
| 78317 |
+
"step": 7676
|
| 78318 |
+
},
|
| 78319 |
+
{
|
| 78320 |
+
"epoch": 0.9938829012525487,
|
| 78321 |
+
"grad_norm": 1.7218032271015382,
|
| 78322 |
+
"learning_rate": 1.410050202257929e-10,
|
| 78323 |
+
"loss": 2.3477,
|
| 78324 |
+
"step": 7677
|
| 78325 |
+
},
|
| 78326 |
+
{
|
| 78327 |
+
"epoch": 0.9940123636599023,
|
| 78328 |
+
"grad_norm": 2.066099568411916,
|
| 78329 |
+
"learning_rate": 1.3506885248718016e-10,
|
| 78330 |
+
"loss": 2.2959,
|
| 78331 |
+
"step": 7678
|
| 78332 |
+
},
|
| 78333 |
+
{
|
| 78334 |
+
"epoch": 0.9941418260672558,
|
| 78335 |
+
"grad_norm": 2.098283026570332,
|
| 78336 |
+
"learning_rate": 1.2926032607621375e-10,
|
| 78337 |
+
"loss": 2.5974,
|
| 78338 |
+
"step": 7679
|
| 78339 |
+
},
|
| 78340 |
+
{
|
| 78341 |
+
"epoch": 0.9942712884746092,
|
| 78342 |
+
"grad_norm": 1.6535605307389196,
|
| 78343 |
+
"learning_rate": 1.2357944217937513e-10,
|
| 78344 |
+
"loss": 2.1594,
|
| 78345 |
+
"step": 7680
|
| 78346 |
+
},
|
| 78347 |
+
{
|
| 78348 |
+
"epoch": 0.9942712884746092,
|
| 78349 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78350 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78351 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78352 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78353 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78354 |
+
"eval_PRM Precision": 1.0,
|
| 78355 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78356 |
+
"eval_PRM Specificty": 1.0,
|
| 78357 |
+
"eval_loss": 1.49853515625,
|
| 78358 |
+
"eval_runtime": 13.6992,
|
| 78359 |
+
"eval_samples_per_second": 3.212,
|
| 78360 |
+
"eval_steps_per_second": 0.219,
|
| 78361 |
+
"step": 7680
|
| 78362 |
+
},
|
| 78363 |
+
{
|
| 78364 |
+
"epoch": 0.9944007508819627,
|
| 78365 |
+
"grad_norm": 1.6459518108342641,
|
| 78366 |
+
"learning_rate": 1.1802620195712487e-10,
|
| 78367 |
+
"loss": 2.2615,
|
| 78368 |
+
"step": 7681
|
| 78369 |
+
},
|
| 78370 |
+
{
|
| 78371 |
+
"epoch": 0.9945302132893161,
|
| 78372 |
+
"grad_norm": 2.1204057298785104,
|
| 78373 |
+
"learning_rate": 1.1260060654376403e-10,
|
| 78374 |
+
"loss": 2.3254,
|
| 78375 |
+
"step": 7682
|
| 78376 |
+
},
|
| 78377 |
+
{
|
| 78378 |
+
"epoch": 0.9946596756966696,
|
| 78379 |
+
"grad_norm": 1.4357120737136115,
|
| 78380 |
+
"learning_rate": 1.0730265704764208e-10,
|
| 78381 |
+
"loss": 2.2004,
|
| 78382 |
+
"step": 7683
|
| 78383 |
+
},
|
| 78384 |
+
{
|
| 78385 |
+
"epoch": 0.994789138104023,
|
| 78386 |
+
"grad_norm": 1.3728528906654596,
|
| 78387 |
+
"learning_rate": 1.0213235455094895e-10,
|
| 78388 |
+
"loss": 2.1951,
|
| 78389 |
+
"step": 7684
|
| 78390 |
+
},
|
| 78391 |
+
{
|
| 78392 |
+
"epoch": 0.9949186005113765,
|
| 78393 |
+
"grad_norm": 3.21900535857529,
|
| 78394 |
+
"learning_rate": 9.708970010978424e-11,
|
| 78395 |
+
"loss": 2.3359,
|
| 78396 |
+
"step": 7685
|
| 78397 |
+
},
|
| 78398 |
+
{
|
| 78399 |
+
"epoch": 0.9949186005113765,
|
| 78400 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78401 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78402 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78403 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78404 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78405 |
+
"eval_PRM Precision": 1.0,
|
| 78406 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78407 |
+
"eval_PRM Specificty": 1.0,
|
| 78408 |
+
"eval_loss": 1.4950062036514282,
|
| 78409 |
+
"eval_runtime": 14.3312,
|
| 78410 |
+
"eval_samples_per_second": 3.07,
|
| 78411 |
+
"eval_steps_per_second": 0.209,
|
| 78412 |
+
"step": 7685
|
| 78413 |
+
},
|
| 78414 |
+
{
|
| 78415 |
+
"epoch": 0.9950480629187299,
|
| 78416 |
+
"grad_norm": 1.6686792950383713,
|
| 78417 |
+
"learning_rate": 9.217469475429619e-11,
|
| 78418 |
+
"loss": 2.2742,
|
| 78419 |
+
"step": 7686
|
| 78420 |
+
},
|
| 78421 |
+
{
|
| 78422 |
+
"epoch": 0.9951775253260834,
|
| 78423 |
+
"grad_norm": 2.2413072500408857,
|
| 78424 |
+
"learning_rate": 8.738733948840395e-11,
|
| 78425 |
+
"loss": 2.1741,
|
| 78426 |
+
"step": 7687
|
| 78427 |
+
},
|
| 78428 |
+
{
|
| 78429 |
+
"epoch": 0.995306987733437,
|
| 78430 |
+
"grad_norm": 2.9388080891161135,
|
| 78431 |
+
"learning_rate": 8.272763529007521e-11,
|
| 78432 |
+
"loss": 2.5015,
|
| 78433 |
+
"step": 7688
|
| 78434 |
+
},
|
| 78435 |
+
{
|
| 78436 |
+
"epoch": 0.9954364501407904,
|
| 78437 |
+
"grad_norm": 1.2448282419855086,
|
| 78438 |
+
"learning_rate": 7.819558311111808e-11,
|
| 78439 |
+
"loss": 2.158,
|
| 78440 |
+
"step": 7689
|
| 78441 |
+
},
|
| 78442 |
+
{
|
| 78443 |
+
"epoch": 0.9955659125481439,
|
| 78444 |
+
"grad_norm": 1.5770981587173023,
|
| 78445 |
+
"learning_rate": 7.379118387731977e-11,
|
| 78446 |
+
"loss": 2.605,
|
| 78447 |
+
"step": 7690
|
| 78448 |
+
},
|
| 78449 |
+
{
|
| 78450 |
+
"epoch": 0.9955659125481439,
|
| 78451 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78452 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78453 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78454 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78455 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78456 |
+
"eval_PRM Precision": 1.0,
|
| 78457 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78458 |
+
"eval_PRM Specificty": 1.0,
|
| 78459 |
+
"eval_loss": 1.4972034692764282,
|
| 78460 |
+
"eval_runtime": 14.7955,
|
| 78461 |
+
"eval_samples_per_second": 2.974,
|
| 78462 |
+
"eval_steps_per_second": 0.203,
|
| 78463 |
+
"step": 7690
|
| 78464 |
+
},
|
| 78465 |
+
{
|
| 78466 |
+
"epoch": 0.9956953749554973,
|
| 78467 |
+
"grad_norm": 2.2313159470985133,
|
| 78468 |
+
"learning_rate": 6.951443848830785e-11,
|
| 78469 |
+
"loss": 2.2695,
|
| 78470 |
+
"step": 7691
|
| 78471 |
+
},
|
| 78472 |
+
{
|
| 78473 |
+
"epoch": 0.9958248373628508,
|
| 78474 |
+
"grad_norm": 1.583978575410211,
|
| 78475 |
+
"learning_rate": 6.53653478177585e-11,
|
| 78476 |
+
"loss": 2.5193,
|
| 78477 |
+
"step": 7692
|
| 78478 |
+
},
|
| 78479 |
+
{
|
| 78480 |
+
"epoch": 0.9959542997702042,
|
| 78481 |
+
"grad_norm": 2.6329864704078796,
|
| 78482 |
+
"learning_rate": 6.134391271318818e-11,
|
| 78483 |
+
"loss": 2.1763,
|
| 78484 |
+
"step": 7693
|
| 78485 |
+
},
|
| 78486 |
+
{
|
| 78487 |
+
"epoch": 0.9960837621775577,
|
| 78488 |
+
"grad_norm": 1.9048592700851108,
|
| 78489 |
+
"learning_rate": 5.745013399609256e-11,
|
| 78490 |
+
"loss": 2.2927,
|
| 78491 |
+
"step": 7694
|
| 78492 |
+
},
|
| 78493 |
+
{
|
| 78494 |
+
"epoch": 0.9962132245849111,
|
| 78495 |
+
"grad_norm": 1.2076390899334741,
|
| 78496 |
+
"learning_rate": 5.3684012461738265e-11,
|
| 78497 |
+
"loss": 2.2676,
|
| 78498 |
+
"step": 7695
|
| 78499 |
+
},
|
| 78500 |
+
{
|
| 78501 |
+
"epoch": 0.9962132245849111,
|
| 78502 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78503 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78504 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78505 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78506 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78507 |
+
"eval_PRM Precision": 1.0,
|
| 78508 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78509 |
+
"eval_PRM Specificty": 1.0,
|
| 78510 |
+
"eval_loss": 1.4951171875,
|
| 78511 |
+
"eval_runtime": 15.159,
|
| 78512 |
+
"eval_samples_per_second": 2.903,
|
| 78513 |
+
"eval_steps_per_second": 0.198,
|
| 78514 |
+
"step": 7695
|
| 78515 |
+
},
|
| 78516 |
+
{
|
| 78517 |
+
"epoch": 0.9963426869922646,
|
| 78518 |
+
"grad_norm": 1.6253165998282486,
|
| 78519 |
+
"learning_rate": 5.004554887957924e-11,
|
| 78520 |
+
"loss": 2.2419,
|
| 78521 |
+
"step": 7696
|
| 78522 |
+
},
|
| 78523 |
+
{
|
| 78524 |
+
"epoch": 0.996472149399618,
|
| 78525 |
+
"grad_norm": 2.151738149711087,
|
| 78526 |
+
"learning_rate": 4.6534743992701635e-11,
|
| 78527 |
+
"loss": 2.6377,
|
| 78528 |
+
"step": 7697
|
| 78529 |
+
},
|
| 78530 |
+
{
|
| 78531 |
+
"epoch": 0.9966016118069716,
|
| 78532 |
+
"grad_norm": 1.9143121114234054,
|
| 78533 |
+
"learning_rate": 4.315159851837891e-11,
|
| 78534 |
+
"loss": 2.3135,
|
| 78535 |
+
"step": 7698
|
| 78536 |
+
},
|
| 78537 |
+
{
|
| 78538 |
+
"epoch": 0.996731074214325,
|
| 78539 |
+
"grad_norm": 1.5023196597769006,
|
| 78540 |
+
"learning_rate": 3.989611314765551e-11,
|
| 78541 |
+
"loss": 2.3765,
|
| 78542 |
+
"step": 7699
|
| 78543 |
+
},
|
| 78544 |
+
{
|
| 78545 |
+
"epoch": 0.9968605366216785,
|
| 78546 |
+
"grad_norm": 1.9325526285591823,
|
| 78547 |
+
"learning_rate": 3.6768288545485644e-11,
|
| 78548 |
+
"loss": 2.4561,
|
| 78549 |
+
"step": 7700
|
| 78550 |
+
},
|
| 78551 |
+
{
|
| 78552 |
+
"epoch": 0.9968605366216785,
|
| 78553 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78554 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78555 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78556 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78557 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78558 |
+
"eval_PRM Precision": 1.0,
|
| 78559 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78560 |
+
"eval_PRM Specificty": 1.0,
|
| 78561 |
+
"eval_loss": 1.4986239671707153,
|
| 78562 |
+
"eval_runtime": 14.5617,
|
| 78563 |
+
"eval_samples_per_second": 3.022,
|
| 78564 |
+
"eval_steps_per_second": 0.206,
|
| 78565 |
+
"step": 7700
|
| 78566 |
+
},
|
| 78567 |
+
{
|
| 78568 |
+
"epoch": 0.996989999029032,
|
| 78569 |
+
"grad_norm": 1.4172984725227213,
|
| 78570 |
+
"learning_rate": 3.376812535080265e-11,
|
| 78571 |
+
"loss": 2.373,
|
| 78572 |
+
"step": 7701
|
| 78573 |
+
},
|
| 78574 |
+
{
|
| 78575 |
+
"epoch": 0.9971194614363854,
|
| 78576 |
+
"grad_norm": 1.3930326525241075,
|
| 78577 |
+
"learning_rate": 3.0895624176519036e-11,
|
| 78578 |
+
"loss": 2.3347,
|
| 78579 |
+
"step": 7702
|
| 78580 |
+
},
|
| 78581 |
+
{
|
| 78582 |
+
"epoch": 0.9972489238437389,
|
| 78583 |
+
"grad_norm": 1.3253972058225258,
|
| 78584 |
+
"learning_rate": 2.8150785609318276e-11,
|
| 78585 |
+
"loss": 2.4019,
|
| 78586 |
+
"step": 7703
|
| 78587 |
+
},
|
| 78588 |
+
{
|
| 78589 |
+
"epoch": 0.9973783862510923,
|
| 78590 |
+
"grad_norm": 1.5408348561147462,
|
| 78591 |
+
"learning_rate": 2.5533610209932392e-11,
|
| 78592 |
+
"loss": 2.3169,
|
| 78593 |
+
"step": 7704
|
| 78594 |
+
},
|
| 78595 |
+
{
|
| 78596 |
+
"epoch": 0.9975078486584458,
|
| 78597 |
+
"grad_norm": 2.3013034684937415,
|
| 78598 |
+
"learning_rate": 2.3044098512933766e-11,
|
| 78599 |
+
"loss": 2.5208,
|
| 78600 |
+
"step": 7705
|
| 78601 |
+
},
|
| 78602 |
+
{
|
| 78603 |
+
"epoch": 0.9975078486584458,
|
| 78604 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78605 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78606 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78607 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78608 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78609 |
+
"eval_PRM Precision": 1.0,
|
| 78610 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78611 |
+
"eval_PRM Specificty": 1.0,
|
| 78612 |
+
"eval_loss": 1.4978249073028564,
|
| 78613 |
+
"eval_runtime": 13.1607,
|
| 78614 |
+
"eval_samples_per_second": 3.343,
|
| 78615 |
+
"eval_steps_per_second": 0.228,
|
| 78616 |
+
"step": 7705
|
| 78617 |
+
},
|
| 78618 |
+
{
|
| 78619 |
+
"epoch": 0.9976373110657992,
|
| 78620 |
+
"grad_norm": 2.259931800473903,
|
| 78621 |
+
"learning_rate": 2.068225102694332e-11,
|
| 78622 |
+
"loss": 2.1882,
|
| 78623 |
+
"step": 7706
|
| 78624 |
+
},
|
| 78625 |
+
{
|
| 78626 |
+
"epoch": 0.9977667734731527,
|
| 78627 |
+
"grad_norm": 1.7890624569047229,
|
| 78628 |
+
"learning_rate": 1.8448068234283578e-11,
|
| 78629 |
+
"loss": 2.4595,
|
| 78630 |
+
"step": 7707
|
| 78631 |
+
},
|
| 78632 |
+
{
|
| 78633 |
+
"epoch": 0.9978962358805062,
|
| 78634 |
+
"grad_norm": 1.2401958555293557,
|
| 78635 |
+
"learning_rate": 1.634155059146436e-11,
|
| 78636 |
+
"loss": 2.2754,
|
| 78637 |
+
"step": 7708
|
| 78638 |
+
},
|
| 78639 |
+
{
|
| 78640 |
+
"epoch": 0.9980256982878597,
|
| 78641 |
+
"grad_norm": 1.0822891791340588,
|
| 78642 |
+
"learning_rate": 1.4362698528697094e-11,
|
| 78643 |
+
"loss": 2.4175,
|
| 78644 |
+
"step": 7709
|
| 78645 |
+
},
|
| 78646 |
+
{
|
| 78647 |
+
"epoch": 0.9981551606952132,
|
| 78648 |
+
"grad_norm": 1.6127689542819217,
|
| 78649 |
+
"learning_rate": 1.2511512450241737e-11,
|
| 78650 |
+
"loss": 2.4133,
|
| 78651 |
+
"step": 7710
|
| 78652 |
+
},
|
| 78653 |
+
{
|
| 78654 |
+
"epoch": 0.9981551606952132,
|
| 78655 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78656 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78657 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78658 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78659 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78660 |
+
"eval_PRM Precision": 1.0,
|
| 78661 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78662 |
+
"eval_PRM Specificty": 1.0,
|
| 78663 |
+
"eval_loss": 1.497114658355713,
|
| 78664 |
+
"eval_runtime": 14.9295,
|
| 78665 |
+
"eval_samples_per_second": 2.947,
|
| 78666 |
+
"eval_steps_per_second": 0.201,
|
| 78667 |
+
"step": 7710
|
| 78668 |
+
},
|
| 78669 |
+
{
|
| 78670 |
+
"epoch": 0.9982846231025666,
|
| 78671 |
+
"grad_norm": 1.7945901352090199,
|
| 78672 |
+
"learning_rate": 1.0787992734198616e-11,
|
| 78673 |
+
"loss": 2.4961,
|
| 78674 |
+
"step": 7711
|
| 78675 |
+
},
|
| 78676 |
+
{
|
| 78677 |
+
"epoch": 0.9984140855099201,
|
| 78678 |
+
"grad_norm": 1.3928437748878686,
|
| 78679 |
+
"learning_rate": 9.192139732716598e-12,
|
| 78680 |
+
"loss": 2.4438,
|
| 78681 |
+
"step": 7712
|
| 78682 |
+
},
|
| 78683 |
+
{
|
| 78684 |
+
"epoch": 0.9985435479172735,
|
| 78685 |
+
"grad_norm": 2.9852499683419422,
|
| 78686 |
+
"learning_rate": 7.723953771715531e-12,
|
| 78687 |
+
"loss": 2.3274,
|
| 78688 |
+
"step": 7713
|
| 78689 |
+
},
|
| 78690 |
+
{
|
| 78691 |
+
"epoch": 0.998673010324627,
|
| 78692 |
+
"grad_norm": 2.150419645821146,
|
| 78693 |
+
"learning_rate": 6.383435151094408e-12,
|
| 78694 |
+
"loss": 2.4382,
|
| 78695 |
+
"step": 7714
|
| 78696 |
+
},
|
| 78697 |
+
{
|
| 78698 |
+
"epoch": 0.9988024727319804,
|
| 78699 |
+
"grad_norm": 1.657145315275181,
|
| 78700 |
+
"learning_rate": 5.170584144661983e-12,
|
| 78701 |
+
"loss": 2.5605,
|
| 78702 |
+
"step": 7715
|
| 78703 |
+
},
|
| 78704 |
+
{
|
| 78705 |
+
"epoch": 0.9988024727319804,
|
| 78706 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78707 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78708 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78709 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78710 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78711 |
+
"eval_PRM Precision": 1.0,
|
| 78712 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78713 |
+
"eval_PRM Specificty": 1.0,
|
| 78714 |
+
"eval_loss": 1.4971591234207153,
|
| 78715 |
+
"eval_runtime": 15.0228,
|
| 78716 |
+
"eval_samples_per_second": 2.929,
|
| 78717 |
+
"eval_steps_per_second": 0.2,
|
| 78718 |
+
"step": 7715
|
| 78719 |
+
},
|
| 78720 |
+
{
|
| 78721 |
+
"epoch": 0.9989319351393339,
|
| 78722 |
+
"grad_norm": 1.4614906798288771,
|
| 78723 |
+
"learning_rate": 4.0854010002755466e-12,
|
| 78724 |
+
"loss": 2.2092,
|
| 78725 |
+
"step": 7716
|
| 78726 |
+
},
|
| 78727 |
+
{
|
| 78728 |
+
"epoch": 0.9990613975466874,
|
| 78729 |
+
"grad_norm": 2.668030707241662,
|
| 78730 |
+
"learning_rate": 3.12788593949398e-12,
|
| 78731 |
+
"loss": 2.4106,
|
| 78732 |
+
"step": 7717
|
| 78733 |
+
},
|
| 78734 |
+
{
|
| 78735 |
+
"epoch": 0.9991908599540409,
|
| 78736 |
+
"grad_norm": 1.4888698890193617,
|
| 78737 |
+
"learning_rate": 2.298039157994092e-12,
|
| 78738 |
+
"loss": 2.3706,
|
| 78739 |
+
"step": 7718
|
| 78740 |
+
},
|
| 78741 |
+
{
|
| 78742 |
+
"epoch": 0.9993203223613943,
|
| 78743 |
+
"grad_norm": 1.558004894188135,
|
| 78744 |
+
"learning_rate": 1.5958608251542828e-12,
|
| 78745 |
+
"loss": 2.3206,
|
| 78746 |
+
"step": 7719
|
| 78747 |
+
},
|
| 78748 |
+
{
|
| 78749 |
+
"epoch": 0.9994497847687478,
|
| 78750 |
+
"grad_norm": 2.4306575905372134,
|
| 78751 |
+
"learning_rate": 1.0213510846096564e-12,
|
| 78752 |
+
"loss": 2.4265,
|
| 78753 |
+
"step": 7720
|
| 78754 |
+
},
|
| 78755 |
+
{
|
| 78756 |
+
"epoch": 0.9994497847687478,
|
| 78757 |
+
"eval_PRM Accuracy": 0.2358490566037736,
|
| 78758 |
+
"eval_PRM F1": 0.047058823529411764,
|
| 78759 |
+
"eval_PRM F1 AUC": 0.5120481927710844,
|
| 78760 |
+
"eval_PRM F1 Neg": 0.36220472440944884,
|
| 78761 |
+
"eval_PRM NPV": 0.22115384615384615,
|
| 78762 |
+
"eval_PRM Precision": 1.0,
|
| 78763 |
+
"eval_PRM Recall": 0.024096385542168676,
|
| 78764 |
+
"eval_PRM Specificty": 1.0,
|
| 78765 |
+
"eval_loss": 1.4971591234207153,
|
| 78766 |
+
"eval_runtime": 14.2046,
|
| 78767 |
+
"eval_samples_per_second": 3.098,
|
| 78768 |
+
"eval_steps_per_second": 0.211,
|
| 78769 |
+
"step": 7720
|
| 78770 |
+
},
|
| 78771 |
+
{
|
| 78772 |
+
"epoch": 0.9995792471761012,
|
| 78773 |
+
"grad_norm": 2.458064036563779,
|
| 78774 |
+
"learning_rate": 5.745100535581305e-13,
|
| 78775 |
+
"loss": 2.3542,
|
| 78776 |
+
"step": 7721
|
| 78777 |
+
},
|
| 78778 |
+
{
|
| 78779 |
+
"epoch": 0.9997087095834547,
|
| 78780 |
+
"grad_norm": 2.295762242918334,
|
| 78781 |
+
"learning_rate": 2.5533782331554903e-13,
|
| 78782 |
+
"loss": 2.5386,
|
| 78783 |
+
"step": 7722
|
| 78784 |
+
},
|
| 78785 |
+
{
|
| 78786 |
+
"epoch": 0.9998381719908082,
|
| 78787 |
+
"grad_norm": 2.5005497572826556,
|
| 78788 |
+
"learning_rate": 6.383445910751463e-14,
|
| 78789 |
+
"loss": 2.3679,
|
| 78790 |
+
"step": 7723
|
| 78791 |
+
},
|
| 78792 |
+
{
|
| 78793 |
+
"epoch": 0.9999676343981616,
|
| 78794 |
+
"grad_norm": 2.2599512852855326,
|
| 78795 |
+
"learning_rate": 0.0,
|
| 78796 |
+
"loss": 2.4456,
|
| 78797 |
+
"step": 7724
|
| 78798 |
}
|
| 78799 |
],
|
| 78800 |
"logging_steps": 1,
|
|
|
|
| 78809 |
"should_evaluate": false,
|
| 78810 |
"should_log": false,
|
| 78811 |
"should_save": true,
|
| 78812 |
+
"should_training_stop": true
|
| 78813 |
},
|
| 78814 |
"attributes": {}
|
| 78815 |
}
|
| 78816 |
},
|
| 78817 |
+
"total_flos": 6493605880496128.0,
|
| 78818 |
"train_batch_size": 4,
|
| 78819 |
"trial_name": null,
|
| 78820 |
"trial_params": null
|