Training in progress, epoch 5, checkpoint
Browse files- last-checkpoint/adapter_model.safetensors +1 -1
- last-checkpoint/global_step696/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step696/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step696/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step696/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step696/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 +667 -3
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 447329696
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d36bb55f9ac1df2f0d25ce2eeb81398e77802712a9578fe7f76ff563ea24649
|
| 3 |
size 447329696
|
last-checkpoint/global_step696/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:e4c7162362fba71e07486a810c54ff80de80f30a28d703d942d1b997fc7cb8f8
|
| 3 |
+
size 337225765
|
last-checkpoint/global_step696/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 sha256:443d6b0cb79a60396548ba9571603afc505a2e142118cf17253eb8cca82fab00
|
| 3 |
+
size 337225893
|
last-checkpoint/global_step696/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:f6c2721131a9b37b14e647c7842fa21c9aa750d5900edff2f7a8789d446cfd01
|
| 3 |
+
size 337225893
|
last-checkpoint/global_step696/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:ef5b68eb08ee3435b3171510705fb9555af98adb337cee17a512add33cdb6165
|
| 3 |
+
size 337225893
|
last-checkpoint/global_step696/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:019144f749e09b782b7717eb00845b206f2fa622efbcf032ef99a7cdb1d5e08b
|
| 3 |
+
size 2282578149
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step696
|
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 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b9eb01ae7bebb6b2a69fa94159506146b6a17d03c942454e369b782eddbb3636
|
| 3 |
size 15429
|
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 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:298868c6850dcaf75951547384b919258d47c4f21a1ad448e8ee66c8bb469ab8
|
| 3 |
size 15429
|
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 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c2a2670a19da7fa45f48be9e5fd15b26c73f5d6e0697c656b72c25cc7434cdf5
|
| 3 |
size 15429
|
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 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de3342513d449cf4f76bef1f9580a412a0169f63933fafeee84531d02a02cbab
|
| 3 |
size 15429
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1401
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:272c412581b4c5ddc60c3fc9367c287dd2275d591e95dd9031ee094fa9346284
|
| 3 |
size 1401
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
"best_global_step": null,
|
| 3 |
-
"best_metric": 0.
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 50,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -1668,6 +1668,670 @@
|
|
| 1668 |
"eval_samples_per_second": 31.219,
|
| 1669 |
"eval_steps_per_second": 1.988,
|
| 1670 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1671 |
}
|
| 1672 |
],
|
| 1673 |
"logging_steps": 5,
|
|
|
|
| 1 |
{
|
| 2 |
"best_global_step": null,
|
| 3 |
+
"best_metric": 0.009149392135441303,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.0,
|
| 6 |
"eval_steps": 50,
|
| 7 |
+
"global_step": 700,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1668 |
"eval_samples_per_second": 31.219,
|
| 1669 |
"eval_steps_per_second": 1.988,
|
| 1670 |
"step": 500
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 3.6104129263913824,
|
| 1674 |
+
"grad_norm": 0.02227591536939144,
|
| 1675 |
+
"learning_rate": 0.00011329185086309797,
|
| 1676 |
+
"logits/chosen": -5.440625190734863,
|
| 1677 |
+
"logits/rejected": -6.892187595367432,
|
| 1678 |
+
"logps/chosen": -263.04998779296875,
|
| 1679 |
+
"logps/rejected": -438.5,
|
| 1680 |
+
"loss": 0.0284,
|
| 1681 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1682 |
+
"rewards/chosen": -19.5625,
|
| 1683 |
+
"rewards/margins": 16.524999618530273,
|
| 1684 |
+
"rewards/rejected": -36.099998474121094,
|
| 1685 |
+
"step": 505
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 3.646319569120287,
|
| 1689 |
+
"grad_norm": 0.263701468706131,
|
| 1690 |
+
"learning_rate": 0.00011328024147954178,
|
| 1691 |
+
"logits/chosen": -6.275000095367432,
|
| 1692 |
+
"logits/rejected": -7.809374809265137,
|
| 1693 |
+
"logps/chosen": -318.5,
|
| 1694 |
+
"logps/rejected": -494.5,
|
| 1695 |
+
"loss": 0.0307,
|
| 1696 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1697 |
+
"rewards/chosen": -25.049999237060547,
|
| 1698 |
+
"rewards/margins": 17.087499618530273,
|
| 1699 |
+
"rewards/rejected": -42.13750076293945,
|
| 1700 |
+
"step": 510
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 3.682226211849192,
|
| 1704 |
+
"grad_norm": 0.08680078387260437,
|
| 1705 |
+
"learning_rate": 0.00011326849010673409,
|
| 1706 |
+
"logits/chosen": -7.175000190734863,
|
| 1707 |
+
"logits/rejected": -8.606249809265137,
|
| 1708 |
+
"logps/chosen": -344.70001220703125,
|
| 1709 |
+
"logps/rejected": -576.4000244140625,
|
| 1710 |
+
"loss": 0.0187,
|
| 1711 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1712 |
+
"rewards/chosen": -27.662500381469727,
|
| 1713 |
+
"rewards/margins": 22.403125762939453,
|
| 1714 |
+
"rewards/rejected": -50.0625,
|
| 1715 |
+
"step": 515
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 3.718132854578097,
|
| 1719 |
+
"grad_norm": 0.3637928068637848,
|
| 1720 |
+
"learning_rate": 0.00011325659677425894,
|
| 1721 |
+
"logits/chosen": -6.360937595367432,
|
| 1722 |
+
"logits/rejected": -8.0078125,
|
| 1723 |
+
"logps/chosen": -321.70001220703125,
|
| 1724 |
+
"logps/rejected": -571.2999877929688,
|
| 1725 |
+
"loss": 0.0563,
|
| 1726 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1727 |
+
"rewards/chosen": -25.375,
|
| 1728 |
+
"rewards/margins": 24.231250762939453,
|
| 1729 |
+
"rewards/rejected": -49.5625,
|
| 1730 |
+
"step": 520
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 3.754039497307002,
|
| 1734 |
+
"grad_norm": 0.06884779036045074,
|
| 1735 |
+
"learning_rate": 0.00011324456151205779,
|
| 1736 |
+
"logits/chosen": -5.998437404632568,
|
| 1737 |
+
"logits/rejected": -7.151562690734863,
|
| 1738 |
+
"logps/chosen": -310.5,
|
| 1739 |
+
"logps/rejected": -512.7999877929688,
|
| 1740 |
+
"loss": 0.0412,
|
| 1741 |
+
"rewards/accuracies": 0.984375,
|
| 1742 |
+
"rewards/chosen": -24.200000762939453,
|
| 1743 |
+
"rewards/margins": 19.4375,
|
| 1744 |
+
"rewards/rejected": -43.63750076293945,
|
| 1745 |
+
"step": 525
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 3.789946140035907,
|
| 1749 |
+
"grad_norm": 0.19923055171966553,
|
| 1750 |
+
"learning_rate": 0.0001132323843504294,
|
| 1751 |
+
"logits/chosen": -6.621874809265137,
|
| 1752 |
+
"logits/rejected": -7.451562404632568,
|
| 1753 |
+
"logps/chosen": -316.8999938964844,
|
| 1754 |
+
"logps/rejected": -483.79998779296875,
|
| 1755 |
+
"loss": 0.0418,
|
| 1756 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1757 |
+
"rewards/chosen": -24.6875,
|
| 1758 |
+
"rewards/margins": 16.034374237060547,
|
| 1759 |
+
"rewards/rejected": -40.724998474121094,
|
| 1760 |
+
"step": 530
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 3.8258527827648114,
|
| 1764 |
+
"grad_norm": 0.025405047461390495,
|
| 1765 |
+
"learning_rate": 0.00011322006532002976,
|
| 1766 |
+
"logits/chosen": -7.5390625,
|
| 1767 |
+
"logits/rejected": -8.484375,
|
| 1768 |
+
"logps/chosen": -325.1000061035156,
|
| 1769 |
+
"logps/rejected": -495.0,
|
| 1770 |
+
"loss": 0.0575,
|
| 1771 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 1772 |
+
"rewards/chosen": -25.862499237060547,
|
| 1773 |
+
"rewards/margins": 16.359375,
|
| 1774 |
+
"rewards/rejected": -42.224998474121094,
|
| 1775 |
+
"step": 535
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 3.8617594254937164,
|
| 1779 |
+
"grad_norm": 0.7414963841438293,
|
| 1780 |
+
"learning_rate": 0.00011320760445187202,
|
| 1781 |
+
"logits/chosen": -8.795312881469727,
|
| 1782 |
+
"logits/rejected": -10.037500381469727,
|
| 1783 |
+
"logps/chosen": -355.8999938964844,
|
| 1784 |
+
"logps/rejected": -577.0,
|
| 1785 |
+
"loss": 0.0196,
|
| 1786 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1787 |
+
"rewards/chosen": -28.868749618530273,
|
| 1788 |
+
"rewards/margins": 21.512500762939453,
|
| 1789 |
+
"rewards/rejected": -50.38750076293945,
|
| 1790 |
+
"step": 540
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 3.8976660682226214,
|
| 1794 |
+
"grad_norm": 2.0291175842285156,
|
| 1795 |
+
"learning_rate": 0.00011319500177732639,
|
| 1796 |
+
"logits/chosen": -8.653124809265137,
|
| 1797 |
+
"logits/rejected": -9.693750381469727,
|
| 1798 |
+
"logps/chosen": -344.0,
|
| 1799 |
+
"logps/rejected": -551.2999877929688,
|
| 1800 |
+
"loss": 0.0377,
|
| 1801 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1802 |
+
"rewards/chosen": -27.618749618530273,
|
| 1803 |
+
"rewards/margins": 19.924999237060547,
|
| 1804 |
+
"rewards/rejected": -47.537498474121094,
|
| 1805 |
+
"step": 545
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 3.933572710951526,
|
| 1809 |
+
"grad_norm": 0.06643925607204437,
|
| 1810 |
+
"learning_rate": 0.00011318225732812008,
|
| 1811 |
+
"logits/chosen": -6.587500095367432,
|
| 1812 |
+
"logits/rejected": -7.754687309265137,
|
| 1813 |
+
"logps/chosen": -280.04998779296875,
|
| 1814 |
+
"logps/rejected": -456.8999938964844,
|
| 1815 |
+
"loss": 0.033,
|
| 1816 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1817 |
+
"rewards/chosen": -21.100000381469727,
|
| 1818 |
+
"rewards/margins": 17.134374618530273,
|
| 1819 |
+
"rewards/rejected": -38.25,
|
| 1820 |
+
"step": 550
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 3.933572710951526,
|
| 1824 |
+
"eval_logits/chosen": -5.768382549285889,
|
| 1825 |
+
"eval_logits/rejected": -6.928308963775635,
|
| 1826 |
+
"eval_logps/chosen": -264.4117736816406,
|
| 1827 |
+
"eval_logps/rejected": -437.76470947265625,
|
| 1828 |
+
"eval_loss": 0.04973261430859566,
|
| 1829 |
+
"eval_rewards/accuracies": 0.9852941036224365,
|
| 1830 |
+
"eval_rewards/chosen": -19.34558868408203,
|
| 1831 |
+
"eval_rewards/margins": 16.566177368164062,
|
| 1832 |
+
"eval_rewards/rejected": -35.89706039428711,
|
| 1833 |
+
"eval_runtime": 8.691,
|
| 1834 |
+
"eval_samples_per_second": 30.721,
|
| 1835 |
+
"eval_steps_per_second": 1.956,
|
| 1836 |
+
"step": 550
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 3.969479353680431,
|
| 1840 |
+
"grad_norm": 0.709564208984375,
|
| 1841 |
+
"learning_rate": 0.00011316937113633724,
|
| 1842 |
+
"logits/chosen": -5.145312309265137,
|
| 1843 |
+
"logits/rejected": -6.15625,
|
| 1844 |
+
"logps/chosen": -254.3000030517578,
|
| 1845 |
+
"logps/rejected": -400.3999938964844,
|
| 1846 |
+
"loss": 0.0689,
|
| 1847 |
+
"rewards/accuracies": 0.965624988079071,
|
| 1848 |
+
"rewards/chosen": -18.596874237060547,
|
| 1849 |
+
"rewards/margins": 13.71875,
|
| 1850 |
+
"rewards/rejected": -32.318748474121094,
|
| 1851 |
+
"step": 555
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 4.0,
|
| 1855 |
+
"grad_norm": 0.014319206587970257,
|
| 1856 |
+
"learning_rate": 0.00011315634323441883,
|
| 1857 |
+
"logits/chosen": -4.669117450714111,
|
| 1858 |
+
"logits/rejected": -5.674632549285889,
|
| 1859 |
+
"logps/chosen": -250.88235473632812,
|
| 1860 |
+
"logps/rejected": -388.23529052734375,
|
| 1861 |
+
"loss": 0.0647,
|
| 1862 |
+
"rewards/accuracies": 0.9632353186607361,
|
| 1863 |
+
"rewards/chosen": -18.419116973876953,
|
| 1864 |
+
"rewards/margins": 12.941176414489746,
|
| 1865 |
+
"rewards/rejected": -31.33823585510254,
|
| 1866 |
+
"step": 560
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 4.0359066427289045,
|
| 1870 |
+
"grad_norm": 1.3408515453338623,
|
| 1871 |
+
"learning_rate": 0.0001131431736551626,
|
| 1872 |
+
"logits/chosen": -4.814062595367432,
|
| 1873 |
+
"logits/rejected": -6.176562309265137,
|
| 1874 |
+
"logps/chosen": -249.4499969482422,
|
| 1875 |
+
"logps/rejected": -412.79998779296875,
|
| 1876 |
+
"loss": 0.0462,
|
| 1877 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1878 |
+
"rewards/chosen": -18.274999618530273,
|
| 1879 |
+
"rewards/margins": 15.787500381469727,
|
| 1880 |
+
"rewards/rejected": -34.068748474121094,
|
| 1881 |
+
"step": 565
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 4.07181328545781,
|
| 1885 |
+
"grad_norm": 0.39796170592308044,
|
| 1886 |
+
"learning_rate": 0.00011312986243172293,
|
| 1887 |
+
"logits/chosen": -4.244531154632568,
|
| 1888 |
+
"logits/rejected": -6.751562595367432,
|
| 1889 |
+
"logps/chosen": -225.6999969482422,
|
| 1890 |
+
"logps/rejected": -486.6000061035156,
|
| 1891 |
+
"loss": 0.013,
|
| 1892 |
+
"rewards/accuracies": 0.996874988079071,
|
| 1893 |
+
"rewards/chosen": -15.662500381469727,
|
| 1894 |
+
"rewards/margins": 25.412500381469727,
|
| 1895 |
+
"rewards/rejected": -41.087501525878906,
|
| 1896 |
+
"step": 570
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 4.1077199281867145,
|
| 1900 |
+
"grad_norm": 2.608715772628784,
|
| 1901 |
+
"learning_rate": 0.00011311640959761081,
|
| 1902 |
+
"logits/chosen": -3.823437452316284,
|
| 1903 |
+
"logits/rejected": -7.4375,
|
| 1904 |
+
"logps/chosen": -240.14999389648438,
|
| 1905 |
+
"logps/rejected": -562.5999755859375,
|
| 1906 |
+
"loss": 0.053,
|
| 1907 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1908 |
+
"rewards/chosen": -17.225000381469727,
|
| 1909 |
+
"rewards/margins": 31.493749618530273,
|
| 1910 |
+
"rewards/rejected": -48.75,
|
| 1911 |
+
"step": 575
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 4.143626570915619,
|
| 1915 |
+
"grad_norm": 2.569322347640991,
|
| 1916 |
+
"learning_rate": 0.00011310281518669376,
|
| 1917 |
+
"logits/chosen": -5.467187404632568,
|
| 1918 |
+
"logits/rejected": -8.5078125,
|
| 1919 |
+
"logps/chosen": -370.6000061035156,
|
| 1920 |
+
"logps/rejected": -627.0,
|
| 1921 |
+
"loss": 0.0235,
|
| 1922 |
+
"rewards/accuracies": 0.984375,
|
| 1923 |
+
"rewards/chosen": -30.15625,
|
| 1924 |
+
"rewards/margins": 25.225000381469727,
|
| 1925 |
+
"rewards/rejected": -55.38750076293945,
|
| 1926 |
+
"step": 580
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 4.1795332136445245,
|
| 1930 |
+
"grad_norm": 7.271477699279785,
|
| 1931 |
+
"learning_rate": 0.00011308907923319566,
|
| 1932 |
+
"logits/chosen": -4.629687309265137,
|
| 1933 |
+
"logits/rejected": -7.557812690734863,
|
| 1934 |
+
"logps/chosen": -309.3500061035156,
|
| 1935 |
+
"logps/rejected": -587.4000244140625,
|
| 1936 |
+
"loss": 0.0519,
|
| 1937 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1938 |
+
"rewards/chosen": -24.274999618530273,
|
| 1939 |
+
"rewards/margins": 27.075000762939453,
|
| 1940 |
+
"rewards/rejected": -51.337501525878906,
|
| 1941 |
+
"step": 585
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 4.215439856373429,
|
| 1945 |
+
"grad_norm": 4.082700729370117,
|
| 1946 |
+
"learning_rate": 0.00011307520177169676,
|
| 1947 |
+
"logits/chosen": -4.9609375,
|
| 1948 |
+
"logits/rejected": -7.151562690734863,
|
| 1949 |
+
"logps/chosen": -320.79998779296875,
|
| 1950 |
+
"logps/rejected": -574.0,
|
| 1951 |
+
"loss": 0.0997,
|
| 1952 |
+
"rewards/accuracies": 0.971875011920929,
|
| 1953 |
+
"rewards/chosen": -25.225000381469727,
|
| 1954 |
+
"rewards/margins": 25.018749237060547,
|
| 1955 |
+
"rewards/rejected": -50.23749923706055,
|
| 1956 |
+
"step": 590
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 4.2513464991023335,
|
| 1960 |
+
"grad_norm": 1.2855074405670166,
|
| 1961 |
+
"learning_rate": 0.00011306118283713357,
|
| 1962 |
+
"logits/chosen": -3.370312452316284,
|
| 1963 |
+
"logits/rejected": -4.561718940734863,
|
| 1964 |
+
"logps/chosen": -268.54998779296875,
|
| 1965 |
+
"logps/rejected": -507.8999938964844,
|
| 1966 |
+
"loss": 0.0289,
|
| 1967 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1968 |
+
"rewards/chosen": -20.206249237060547,
|
| 1969 |
+
"rewards/margins": 23.3125,
|
| 1970 |
+
"rewards/rejected": -43.537498474121094,
|
| 1971 |
+
"step": 595
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 4.287253141831239,
|
| 1975 |
+
"grad_norm": 0.21659362316131592,
|
| 1976 |
+
"learning_rate": 0.00011304702246479876,
|
| 1977 |
+
"logits/chosen": -3.285937547683716,
|
| 1978 |
+
"logits/rejected": -4.146093845367432,
|
| 1979 |
+
"logps/chosen": -270.3500061035156,
|
| 1980 |
+
"logps/rejected": -488.20001220703125,
|
| 1981 |
+
"loss": 0.0174,
|
| 1982 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 1983 |
+
"rewards/chosen": -19.931249618530273,
|
| 1984 |
+
"rewards/margins": 21.193750381469727,
|
| 1985 |
+
"rewards/rejected": -41.125,
|
| 1986 |
+
"step": 600
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 4.287253141831239,
|
| 1990 |
+
"eval_logits/chosen": -3.882352828979492,
|
| 1991 |
+
"eval_logits/rejected": -5.0,
|
| 1992 |
+
"eval_logps/chosen": -272.76470947265625,
|
| 1993 |
+
"eval_logps/rejected": -492.8235168457031,
|
| 1994 |
+
"eval_loss": 0.014545433223247528,
|
| 1995 |
+
"eval_rewards/accuracies": 0.9926470518112183,
|
| 1996 |
+
"eval_rewards/chosen": -20.176469802856445,
|
| 1997 |
+
"eval_rewards/margins": 21.330883026123047,
|
| 1998 |
+
"eval_rewards/rejected": -41.52941131591797,
|
| 1999 |
+
"eval_runtime": 8.5511,
|
| 2000 |
+
"eval_samples_per_second": 31.224,
|
| 2001 |
+
"eval_steps_per_second": 1.988,
|
| 2002 |
+
"step": 600
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 4.3231597845601435,
|
| 2006 |
+
"grad_norm": 0.36118730902671814,
|
| 2007 |
+
"learning_rate": 0.00011303272069034098,
|
| 2008 |
+
"logits/chosen": -4.321875095367432,
|
| 2009 |
+
"logits/rejected": -5.332812309265137,
|
| 2010 |
+
"logps/chosen": -282.45001220703125,
|
| 2011 |
+
"logps/rejected": -488.0,
|
| 2012 |
+
"loss": 0.0449,
|
| 2013 |
+
"rewards/accuracies": 0.981249988079071,
|
| 2014 |
+
"rewards/chosen": -21.653125762939453,
|
| 2015 |
+
"rewards/margins": 19.787500381469727,
|
| 2016 |
+
"rewards/rejected": -41.45000076293945,
|
| 2017 |
+
"step": 605
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 4.359066427289049,
|
| 2021 |
+
"grad_norm": 0.5386459231376648,
|
| 2022 |
+
"learning_rate": 0.00011301827754976498,
|
| 2023 |
+
"logits/chosen": -5.6015625,
|
| 2024 |
+
"logits/rejected": -6.984375,
|
| 2025 |
+
"logps/chosen": -302.29998779296875,
|
| 2026 |
+
"logps/rejected": -557.7000122070312,
|
| 2027 |
+
"loss": 0.0153,
|
| 2028 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 2029 |
+
"rewards/chosen": -23.518749237060547,
|
| 2030 |
+
"rewards/margins": 24.784374237060547,
|
| 2031 |
+
"rewards/rejected": -48.29999923706055,
|
| 2032 |
+
"step": 610
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 4.3949730700179535,
|
| 2036 |
+
"grad_norm": 4.88869571685791,
|
| 2037 |
+
"learning_rate": 0.00011300369307943137,
|
| 2038 |
+
"logits/chosen": -6.215624809265137,
|
| 2039 |
+
"logits/rejected": -7.949999809265137,
|
| 2040 |
+
"logps/chosen": -270.1000061035156,
|
| 2041 |
+
"logps/rejected": -572.2999877929688,
|
| 2042 |
+
"loss": 0.0224,
|
| 2043 |
+
"rewards/accuracies": 0.996874988079071,
|
| 2044 |
+
"rewards/chosen": -19.809375762939453,
|
| 2045 |
+
"rewards/margins": 29.512500762939453,
|
| 2046 |
+
"rewards/rejected": -49.3125,
|
| 2047 |
+
"step": 615
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 4.430879712746858,
|
| 2051 |
+
"grad_norm": 2.2598838806152344,
|
| 2052 |
+
"learning_rate": 0.00011298896731605649,
|
| 2053 |
+
"logits/chosen": -3.9453125,
|
| 2054 |
+
"logits/rejected": -5.546875,
|
| 2055 |
+
"logps/chosen": -203.35000610351562,
|
| 2056 |
+
"logps/rejected": -407.79998779296875,
|
| 2057 |
+
"loss": 0.0373,
|
| 2058 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 2059 |
+
"rewards/chosen": -13.240625381469727,
|
| 2060 |
+
"rewards/margins": 20.003124237060547,
|
| 2061 |
+
"rewards/rejected": -33.26250076293945,
|
| 2062 |
+
"step": 620
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 4.466786355475763,
|
| 2066 |
+
"grad_norm": 0.47177407145500183,
|
| 2067 |
+
"learning_rate": 0.00011297410029671247,
|
| 2068 |
+
"logits/chosen": -3.74609375,
|
| 2069 |
+
"logits/rejected": -5.087500095367432,
|
| 2070 |
+
"logps/chosen": -281.70001220703125,
|
| 2071 |
+
"logps/rejected": -448.79998779296875,
|
| 2072 |
+
"loss": 0.04,
|
| 2073 |
+
"rewards/accuracies": 0.981249988079071,
|
| 2074 |
+
"rewards/chosen": -21.34375,
|
| 2075 |
+
"rewards/margins": 16.043750762939453,
|
| 2076 |
+
"rewards/rejected": -37.42499923706055,
|
| 2077 |
+
"step": 625
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 4.502692998204668,
|
| 2081 |
+
"grad_norm": 0.2046031653881073,
|
| 2082 |
+
"learning_rate": 0.00011295909205882698,
|
| 2083 |
+
"logits/chosen": -1.002783179283142,
|
| 2084 |
+
"logits/rejected": -3.4195313453674316,
|
| 2085 |
+
"logps/chosen": -189.8249969482422,
|
| 2086 |
+
"logps/rejected": -390.1000061035156,
|
| 2087 |
+
"loss": 0.0272,
|
| 2088 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 2089 |
+
"rewards/chosen": -11.790624618530273,
|
| 2090 |
+
"rewards/margins": 19.309375762939453,
|
| 2091 |
+
"rewards/rejected": -31.09375,
|
| 2092 |
+
"step": 630
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 4.5385996409335725,
|
| 2096 |
+
"grad_norm": 0.29450154304504395,
|
| 2097 |
+
"learning_rate": 0.00011294394264018326,
|
| 2098 |
+
"logits/chosen": -2.082812547683716,
|
| 2099 |
+
"logits/rejected": -4.514062404632568,
|
| 2100 |
+
"logps/chosen": -224.8000030517578,
|
| 2101 |
+
"logps/rejected": -433.79998779296875,
|
| 2102 |
+
"loss": 0.0158,
|
| 2103 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 2104 |
+
"rewards/chosen": -15.709375381469727,
|
| 2105 |
+
"rewards/margins": 19.975000381469727,
|
| 2106 |
+
"rewards/rejected": -35.70000076293945,
|
| 2107 |
+
"step": 635
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 4.574506283662478,
|
| 2111 |
+
"grad_norm": 0.15016689896583557,
|
| 2112 |
+
"learning_rate": 0.00011292865207891994,
|
| 2113 |
+
"logits/chosen": -3.026562452316284,
|
| 2114 |
+
"logits/rejected": -5.426562309265137,
|
| 2115 |
+
"logps/chosen": -247.1999969482422,
|
| 2116 |
+
"logps/rejected": -477.1000061035156,
|
| 2117 |
+
"loss": 0.0044,
|
| 2118 |
+
"rewards/accuracies": 0.996874988079071,
|
| 2119 |
+
"rewards/chosen": -17.771875381469727,
|
| 2120 |
+
"rewards/margins": 22.262500762939453,
|
| 2121 |
+
"rewards/rejected": -40.025001525878906,
|
| 2122 |
+
"step": 640
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 4.6104129263913824,
|
| 2126 |
+
"grad_norm": 3.1011385917663574,
|
| 2127 |
+
"learning_rate": 0.00011291322041353101,
|
| 2128 |
+
"logits/chosen": -3.1812500953674316,
|
| 2129 |
+
"logits/rejected": -5.546875,
|
| 2130 |
+
"logps/chosen": -231.9499969482422,
|
| 2131 |
+
"logps/rejected": -464.5,
|
| 2132 |
+
"loss": 0.0109,
|
| 2133 |
+
"rewards/accuracies": 0.996874988079071,
|
| 2134 |
+
"rewards/chosen": -16.149999618530273,
|
| 2135 |
+
"rewards/margins": 22.75,
|
| 2136 |
+
"rewards/rejected": -38.912498474121094,
|
| 2137 |
+
"step": 645
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 4.646319569120287,
|
| 2141 |
+
"grad_norm": 0.06376684457063675,
|
| 2142 |
+
"learning_rate": 0.00011289764768286565,
|
| 2143 |
+
"logits/chosen": -4.444531440734863,
|
| 2144 |
+
"logits/rejected": -6.487500190734863,
|
| 2145 |
+
"logps/chosen": -285.45001220703125,
|
| 2146 |
+
"logps/rejected": -519.7999877929688,
|
| 2147 |
+
"loss": 0.0149,
|
| 2148 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2149 |
+
"rewards/chosen": -21.674999237060547,
|
| 2150 |
+
"rewards/margins": 22.8125,
|
| 2151 |
+
"rewards/rejected": -44.525001525878906,
|
| 2152 |
+
"step": 650
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 4.646319569120287,
|
| 2156 |
+
"eval_logits/chosen": -4.110294342041016,
|
| 2157 |
+
"eval_logits/rejected": -6.588235378265381,
|
| 2158 |
+
"eval_logps/chosen": -257.6470642089844,
|
| 2159 |
+
"eval_logps/rejected": -516.5882568359375,
|
| 2160 |
+
"eval_loss": 0.029071472585201263,
|
| 2161 |
+
"eval_rewards/accuracies": 0.9889705777168274,
|
| 2162 |
+
"eval_rewards/chosen": -18.647058486938477,
|
| 2163 |
+
"eval_rewards/margins": 25.28676414489746,
|
| 2164 |
+
"eval_rewards/rejected": -43.94117736816406,
|
| 2165 |
+
"eval_runtime": 8.8159,
|
| 2166 |
+
"eval_samples_per_second": 30.286,
|
| 2167 |
+
"eval_steps_per_second": 1.928,
|
| 2168 |
+
"step": 650
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 4.682226211849192,
|
| 2172 |
+
"grad_norm": 3.26318621635437,
|
| 2173 |
+
"learning_rate": 0.00011288193392612822,
|
| 2174 |
+
"logits/chosen": -3.928906202316284,
|
| 2175 |
+
"logits/rejected": -6.315625190734863,
|
| 2176 |
+
"logps/chosen": -248.64999389648438,
|
| 2177 |
+
"logps/rejected": -503.79998779296875,
|
| 2178 |
+
"loss": 0.056,
|
| 2179 |
+
"rewards/accuracies": 0.9781249761581421,
|
| 2180 |
+
"rewards/chosen": -18.081249237060547,
|
| 2181 |
+
"rewards/margins": 24.456249237060547,
|
| 2182 |
+
"rewards/rejected": -42.537498474121094,
|
| 2183 |
+
"step": 655
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 4.718132854578097,
|
| 2187 |
+
"grad_norm": 0.18917639553546906,
|
| 2188 |
+
"learning_rate": 0.00011286607918287803,
|
| 2189 |
+
"logits/chosen": -2.7222657203674316,
|
| 2190 |
+
"logits/rejected": -5.206250190734863,
|
| 2191 |
+
"logps/chosen": -258.1499938964844,
|
| 2192 |
+
"logps/rejected": -481.79998779296875,
|
| 2193 |
+
"loss": 0.0583,
|
| 2194 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 2195 |
+
"rewards/chosen": -19.225000381469727,
|
| 2196 |
+
"rewards/margins": 21.606250762939453,
|
| 2197 |
+
"rewards/rejected": -40.849998474121094,
|
| 2198 |
+
"step": 660
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 4.7540394973070015,
|
| 2202 |
+
"grad_norm": 0.9138699173927307,
|
| 2203 |
+
"learning_rate": 0.00011285008349302943,
|
| 2204 |
+
"logits/chosen": -2.116406202316284,
|
| 2205 |
+
"logits/rejected": -4.259375095367432,
|
| 2206 |
+
"logps/chosen": -294.20001220703125,
|
| 2207 |
+
"logps/rejected": -505.20001220703125,
|
| 2208 |
+
"loss": 0.0192,
|
| 2209 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 2210 |
+
"rewards/chosen": -23.012500762939453,
|
| 2211 |
+
"rewards/margins": 20.296875,
|
| 2212 |
+
"rewards/rejected": -43.337501525878906,
|
| 2213 |
+
"step": 665
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 4.789946140035907,
|
| 2217 |
+
"grad_norm": 0.5865360498428345,
|
| 2218 |
+
"learning_rate": 0.00011283394689685153,
|
| 2219 |
+
"logits/chosen": -3.46875,
|
| 2220 |
+
"logits/rejected": -5.800000190734863,
|
| 2221 |
+
"logps/chosen": -292.0,
|
| 2222 |
+
"logps/rejected": -535.2999877929688,
|
| 2223 |
+
"loss": 0.0238,
|
| 2224 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 2225 |
+
"rewards/chosen": -22.268749237060547,
|
| 2226 |
+
"rewards/margins": 23.831249237060547,
|
| 2227 |
+
"rewards/rejected": -46.04999923706055,
|
| 2228 |
+
"step": 670
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 4.825852782764811,
|
| 2232 |
+
"grad_norm": 1.3726475238800049,
|
| 2233 |
+
"learning_rate": 0.0001128176694349682,
|
| 2234 |
+
"logits/chosen": -3.3148436546325684,
|
| 2235 |
+
"logits/rejected": -5.324999809265137,
|
| 2236 |
+
"logps/chosen": -286.29998779296875,
|
| 2237 |
+
"logps/rejected": -542.9000244140625,
|
| 2238 |
+
"loss": 0.0127,
|
| 2239 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 2240 |
+
"rewards/chosen": -22.024999618530273,
|
| 2241 |
+
"rewards/margins": 24.868749618530273,
|
| 2242 |
+
"rewards/rejected": -46.912498474121094,
|
| 2243 |
+
"step": 675
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 4.861759425493716,
|
| 2247 |
+
"grad_norm": 10.441712379455566,
|
| 2248 |
+
"learning_rate": 0.00011280125114835791,
|
| 2249 |
+
"logits/chosen": -2.067578077316284,
|
| 2250 |
+
"logits/rejected": -4.528124809265137,
|
| 2251 |
+
"logps/chosen": -216.75,
|
| 2252 |
+
"logps/rejected": -488.70001220703125,
|
| 2253 |
+
"loss": 0.0218,
|
| 2254 |
+
"rewards/accuracies": 0.9906250238418579,
|
| 2255 |
+
"rewards/chosen": -15.034375190734863,
|
| 2256 |
+
"rewards/margins": 26.6875,
|
| 2257 |
+
"rewards/rejected": -41.712501525878906,
|
| 2258 |
+
"step": 680
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 4.897666068222621,
|
| 2262 |
+
"grad_norm": 0.9471856355667114,
|
| 2263 |
+
"learning_rate": 0.00011278469207835369,
|
| 2264 |
+
"logits/chosen": -0.960888683795929,
|
| 2265 |
+
"logits/rejected": -3.39453125,
|
| 2266 |
+
"logps/chosen": -179.25,
|
| 2267 |
+
"logps/rejected": -433.1000061035156,
|
| 2268 |
+
"loss": 0.0337,
|
| 2269 |
+
"rewards/accuracies": 0.996874988079071,
|
| 2270 |
+
"rewards/chosen": -11.198437690734863,
|
| 2271 |
+
"rewards/margins": 24.59375,
|
| 2272 |
+
"rewards/rejected": -35.787498474121094,
|
| 2273 |
+
"step": 685
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 4.933572710951526,
|
| 2277 |
+
"grad_norm": 4.058782577514648,
|
| 2278 |
+
"learning_rate": 0.000112767992266643,
|
| 2279 |
+
"logits/chosen": -2.7562499046325684,
|
| 2280 |
+
"logits/rejected": -4.982031345367432,
|
| 2281 |
+
"logps/chosen": -250.60000610351562,
|
| 2282 |
+
"logps/rejected": -498.20001220703125,
|
| 2283 |
+
"loss": 0.0548,
|
| 2284 |
+
"rewards/accuracies": 0.987500011920929,
|
| 2285 |
+
"rewards/chosen": -18.268749237060547,
|
| 2286 |
+
"rewards/margins": 23.912500381469727,
|
| 2287 |
+
"rewards/rejected": -42.1875,
|
| 2288 |
+
"step": 690
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 4.9694793536804305,
|
| 2292 |
+
"grad_norm": 5.3591084480285645,
|
| 2293 |
+
"learning_rate": 0.00011275115175526756,
|
| 2294 |
+
"logits/chosen": -3.2984375953674316,
|
| 2295 |
+
"logits/rejected": -5.337500095367432,
|
| 2296 |
+
"logps/chosen": -273.29998779296875,
|
| 2297 |
+
"logps/rejected": -526.4000244140625,
|
| 2298 |
+
"loss": 0.0574,
|
| 2299 |
+
"rewards/accuracies": 0.984375,
|
| 2300 |
+
"rewards/chosen": -20.40625,
|
| 2301 |
+
"rewards/margins": 24.493749618530273,
|
| 2302 |
+
"rewards/rejected": -44.900001525878906,
|
| 2303 |
+
"step": 695
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 5.0,
|
| 2307 |
+
"grad_norm": 3.2741596698760986,
|
| 2308 |
+
"learning_rate": 0.00011273417058662334,
|
| 2309 |
+
"logits/chosen": -1.62890625,
|
| 2310 |
+
"logits/rejected": -3.163602828979492,
|
| 2311 |
+
"logps/chosen": -178.4705810546875,
|
| 2312 |
+
"logps/rejected": -387.76470947265625,
|
| 2313 |
+
"loss": 0.0274,
|
| 2314 |
+
"rewards/accuracies": 0.9852941036224365,
|
| 2315 |
+
"rewards/chosen": -10.939338684082031,
|
| 2316 |
+
"rewards/margins": 20.0,
|
| 2317 |
+
"rewards/rejected": -30.941177368164062,
|
| 2318 |
+
"step": 700
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 5.0,
|
| 2322 |
+
"eval_logits/chosen": -1.3389246463775635,
|
| 2323 |
+
"eval_logits/rejected": -2.8189337253570557,
|
| 2324 |
+
"eval_logps/chosen": -160.88235473632812,
|
| 2325 |
+
"eval_logps/rejected": -377.29412841796875,
|
| 2326 |
+
"eval_loss": 0.009149392135441303,
|
| 2327 |
+
"eval_rewards/accuracies": 1.0,
|
| 2328 |
+
"eval_rewards/chosen": -8.79411792755127,
|
| 2329 |
+
"eval_rewards/margins": 20.941177368164062,
|
| 2330 |
+
"eval_rewards/rejected": -29.757352828979492,
|
| 2331 |
+
"eval_runtime": 8.8399,
|
| 2332 |
+
"eval_samples_per_second": 30.204,
|
| 2333 |
+
"eval_steps_per_second": 1.923,
|
| 2334 |
+
"step": 700
|
| 2335 |
}
|
| 2336 |
],
|
| 2337 |
"logging_steps": 5,
|