Training in progress, step 14000, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 613004648
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e7faa7408a314eb81f420ce1f2568d726ce6012f44614f1f2a60a7364bd5ef3f
|
| 3 |
size 613004648
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1226096954
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed1cf1b200aa0d0cd19c9252edcac54428c1ac0420162221afb866eff0c7bd48
|
| 3 |
size 1226096954
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96df60f7a61e3b5eb854abf0dcce2641f189b2c45835fcf9c1fef3b37282da2b
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bdb91814ca594bcc53b1d6ad5131ca9e1395fe1b33192ef261102fa12fca871c
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 6.
|
| 5 |
"eval_steps": 250,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -9939,6 +9939,770 @@
|
|
| 9939 |
"eval_spearman_manhattan": 0.752705655614685,
|
| 9940 |
"eval_steps_per_second": 37.444,
|
| 9941 |
"step": 13000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9942 |
}
|
| 9943 |
],
|
| 9944 |
"logging_steps": 10,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 6.560449859418932,
|
| 5 |
"eval_steps": 250,
|
| 6 |
+
"global_step": 14000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 9939 |
"eval_spearman_manhattan": 0.752705655614685,
|
| 9940 |
"eval_steps_per_second": 37.444,
|
| 9941 |
"step": 13000
|
| 9942 |
+
},
|
| 9943 |
+
{
|
| 9944 |
+
"epoch": 6.096532333645736,
|
| 9945 |
+
"grad_norm": 2.591386556625366,
|
| 9946 |
+
"learning_rate": 9.237933458294283e-06,
|
| 9947 |
+
"loss": 0.074,
|
| 9948 |
+
"step": 13010
|
| 9949 |
+
},
|
| 9950 |
+
{
|
| 9951 |
+
"epoch": 6.101218369259606,
|
| 9952 |
+
"grad_norm": 3.0938313007354736,
|
| 9953 |
+
"learning_rate": 9.23734770384255e-06,
|
| 9954 |
+
"loss": 0.0889,
|
| 9955 |
+
"step": 13020
|
| 9956 |
+
},
|
| 9957 |
+
{
|
| 9958 |
+
"epoch": 6.105904404873477,
|
| 9959 |
+
"grad_norm": 1.073983907699585,
|
| 9960 |
+
"learning_rate": 9.236761949390815e-06,
|
| 9961 |
+
"loss": 0.0702,
|
| 9962 |
+
"step": 13030
|
| 9963 |
+
},
|
| 9964 |
+
{
|
| 9965 |
+
"epoch": 6.110590440487348,
|
| 9966 |
+
"grad_norm": 1.7386090755462646,
|
| 9967 |
+
"learning_rate": 9.236176194939082e-06,
|
| 9968 |
+
"loss": 0.0965,
|
| 9969 |
+
"step": 13040
|
| 9970 |
+
},
|
| 9971 |
+
{
|
| 9972 |
+
"epoch": 6.115276476101219,
|
| 9973 |
+
"grad_norm": 2.1852550506591797,
|
| 9974 |
+
"learning_rate": 9.235590440487349e-06,
|
| 9975 |
+
"loss": 0.0818,
|
| 9976 |
+
"step": 13050
|
| 9977 |
+
},
|
| 9978 |
+
{
|
| 9979 |
+
"epoch": 6.119962511715089,
|
| 9980 |
+
"grad_norm": 2.9571776390075684,
|
| 9981 |
+
"learning_rate": 9.235004686035614e-06,
|
| 9982 |
+
"loss": 0.0909,
|
| 9983 |
+
"step": 13060
|
| 9984 |
+
},
|
| 9985 |
+
{
|
| 9986 |
+
"epoch": 6.12464854732896,
|
| 9987 |
+
"grad_norm": 1.568499207496643,
|
| 9988 |
+
"learning_rate": 9.234418931583881e-06,
|
| 9989 |
+
"loss": 0.0935,
|
| 9990 |
+
"step": 13070
|
| 9991 |
+
},
|
| 9992 |
+
{
|
| 9993 |
+
"epoch": 6.12933458294283,
|
| 9994 |
+
"grad_norm": 2.0286030769348145,
|
| 9995 |
+
"learning_rate": 9.233833177132146e-06,
|
| 9996 |
+
"loss": 0.0923,
|
| 9997 |
+
"step": 13080
|
| 9998 |
+
},
|
| 9999 |
+
{
|
| 10000 |
+
"epoch": 6.134020618556701,
|
| 10001 |
+
"grad_norm": 1.3567001819610596,
|
| 10002 |
+
"learning_rate": 9.233247422680413e-06,
|
| 10003 |
+
"loss": 0.0874,
|
| 10004 |
+
"step": 13090
|
| 10005 |
+
},
|
| 10006 |
+
{
|
| 10007 |
+
"epoch": 6.138706654170572,
|
| 10008 |
+
"grad_norm": 3.1520678997039795,
|
| 10009 |
+
"learning_rate": 9.23266166822868e-06,
|
| 10010 |
+
"loss": 0.0907,
|
| 10011 |
+
"step": 13100
|
| 10012 |
+
},
|
| 10013 |
+
{
|
| 10014 |
+
"epoch": 6.143392689784442,
|
| 10015 |
+
"grad_norm": 1.935091495513916,
|
| 10016 |
+
"learning_rate": 9.232075913776945e-06,
|
| 10017 |
+
"loss": 0.0927,
|
| 10018 |
+
"step": 13110
|
| 10019 |
+
},
|
| 10020 |
+
{
|
| 10021 |
+
"epoch": 6.148078725398313,
|
| 10022 |
+
"grad_norm": 2.9123198986053467,
|
| 10023 |
+
"learning_rate": 9.23149015932521e-06,
|
| 10024 |
+
"loss": 0.0986,
|
| 10025 |
+
"step": 13120
|
| 10026 |
+
},
|
| 10027 |
+
{
|
| 10028 |
+
"epoch": 6.152764761012183,
|
| 10029 |
+
"grad_norm": 2.592006206512451,
|
| 10030 |
+
"learning_rate": 9.230904404873478e-06,
|
| 10031 |
+
"loss": 0.0864,
|
| 10032 |
+
"step": 13130
|
| 10033 |
+
},
|
| 10034 |
+
{
|
| 10035 |
+
"epoch": 6.157450796626055,
|
| 10036 |
+
"grad_norm": 2.0214340686798096,
|
| 10037 |
+
"learning_rate": 9.230318650421743e-06,
|
| 10038 |
+
"loss": 0.1066,
|
| 10039 |
+
"step": 13140
|
| 10040 |
+
},
|
| 10041 |
+
{
|
| 10042 |
+
"epoch": 6.162136832239925,
|
| 10043 |
+
"grad_norm": 1.743891716003418,
|
| 10044 |
+
"learning_rate": 9.22973289597001e-06,
|
| 10045 |
+
"loss": 0.0871,
|
| 10046 |
+
"step": 13150
|
| 10047 |
+
},
|
| 10048 |
+
{
|
| 10049 |
+
"epoch": 6.166822867853796,
|
| 10050 |
+
"grad_norm": 2.368562698364258,
|
| 10051 |
+
"learning_rate": 9.229147141518277e-06,
|
| 10052 |
+
"loss": 0.0754,
|
| 10053 |
+
"step": 13160
|
| 10054 |
+
},
|
| 10055 |
+
{
|
| 10056 |
+
"epoch": 6.171508903467666,
|
| 10057 |
+
"grad_norm": 2.369436264038086,
|
| 10058 |
+
"learning_rate": 9.228561387066542e-06,
|
| 10059 |
+
"loss": 0.0788,
|
| 10060 |
+
"step": 13170
|
| 10061 |
+
},
|
| 10062 |
+
{
|
| 10063 |
+
"epoch": 6.176194939081537,
|
| 10064 |
+
"grad_norm": 1.3312957286834717,
|
| 10065 |
+
"learning_rate": 9.227975632614809e-06,
|
| 10066 |
+
"loss": 0.0816,
|
| 10067 |
+
"step": 13180
|
| 10068 |
+
},
|
| 10069 |
+
{
|
| 10070 |
+
"epoch": 6.180880974695408,
|
| 10071 |
+
"grad_norm": 1.15755295753479,
|
| 10072 |
+
"learning_rate": 9.227389878163074e-06,
|
| 10073 |
+
"loss": 0.0867,
|
| 10074 |
+
"step": 13190
|
| 10075 |
+
},
|
| 10076 |
+
{
|
| 10077 |
+
"epoch": 6.185567010309279,
|
| 10078 |
+
"grad_norm": 1.4255220890045166,
|
| 10079 |
+
"learning_rate": 9.226804123711341e-06,
|
| 10080 |
+
"loss": 0.085,
|
| 10081 |
+
"step": 13200
|
| 10082 |
+
},
|
| 10083 |
+
{
|
| 10084 |
+
"epoch": 6.190253045923149,
|
| 10085 |
+
"grad_norm": 2.264315605163574,
|
| 10086 |
+
"learning_rate": 9.226218369259608e-06,
|
| 10087 |
+
"loss": 0.0939,
|
| 10088 |
+
"step": 13210
|
| 10089 |
+
},
|
| 10090 |
+
{
|
| 10091 |
+
"epoch": 6.1949390815370196,
|
| 10092 |
+
"grad_norm": 1.314502239227295,
|
| 10093 |
+
"learning_rate": 9.225632614807873e-06,
|
| 10094 |
+
"loss": 0.0847,
|
| 10095 |
+
"step": 13220
|
| 10096 |
+
},
|
| 10097 |
+
{
|
| 10098 |
+
"epoch": 6.19962511715089,
|
| 10099 |
+
"grad_norm": 1.463553786277771,
|
| 10100 |
+
"learning_rate": 9.22504686035614e-06,
|
| 10101 |
+
"loss": 0.0733,
|
| 10102 |
+
"step": 13230
|
| 10103 |
+
},
|
| 10104 |
+
{
|
| 10105 |
+
"epoch": 6.204311152764761,
|
| 10106 |
+
"grad_norm": 2.397528648376465,
|
| 10107 |
+
"learning_rate": 9.224461105904405e-06,
|
| 10108 |
+
"loss": 0.1004,
|
| 10109 |
+
"step": 13240
|
| 10110 |
+
},
|
| 10111 |
+
{
|
| 10112 |
+
"epoch": 6.208997188378632,
|
| 10113 |
+
"grad_norm": 1.8068519830703735,
|
| 10114 |
+
"learning_rate": 9.223875351452672e-06,
|
| 10115 |
+
"loss": 0.081,
|
| 10116 |
+
"step": 13250
|
| 10117 |
+
},
|
| 10118 |
+
{
|
| 10119 |
+
"epoch": 6.208997188378632,
|
| 10120 |
+
"eval_loss": 0.04053681343793869,
|
| 10121 |
+
"eval_pearson_cosine": 0.7813660928551514,
|
| 10122 |
+
"eval_pearson_dot": 0.6355779688477607,
|
| 10123 |
+
"eval_pearson_euclidean": 0.7355234843709724,
|
| 10124 |
+
"eval_pearson_manhattan": 0.7352503996205417,
|
| 10125 |
+
"eval_runtime": 39.4829,
|
| 10126 |
+
"eval_samples_per_second": 37.991,
|
| 10127 |
+
"eval_spearman_cosine": 0.7853597304476542,
|
| 10128 |
+
"eval_spearman_dot": 0.6478331015151749,
|
| 10129 |
+
"eval_spearman_euclidean": 0.7514050247961205,
|
| 10130 |
+
"eval_spearman_manhattan": 0.751296576320389,
|
| 10131 |
+
"eval_steps_per_second": 37.991,
|
| 10132 |
+
"step": 13250
|
| 10133 |
+
},
|
| 10134 |
+
{
|
| 10135 |
+
"epoch": 6.213683223992502,
|
| 10136 |
+
"grad_norm": 2.0780019760131836,
|
| 10137 |
+
"learning_rate": 9.223289597000939e-06,
|
| 10138 |
+
"loss": 0.1108,
|
| 10139 |
+
"step": 13260
|
| 10140 |
+
},
|
| 10141 |
+
{
|
| 10142 |
+
"epoch": 6.218369259606373,
|
| 10143 |
+
"grad_norm": 1.3198286294937134,
|
| 10144 |
+
"learning_rate": 9.222703842549204e-06,
|
| 10145 |
+
"loss": 0.0809,
|
| 10146 |
+
"step": 13270
|
| 10147 |
+
},
|
| 10148 |
+
{
|
| 10149 |
+
"epoch": 6.223055295220243,
|
| 10150 |
+
"grad_norm": 1.4620928764343262,
|
| 10151 |
+
"learning_rate": 9.22211808809747e-06,
|
| 10152 |
+
"loss": 0.0803,
|
| 10153 |
+
"step": 13280
|
| 10154 |
+
},
|
| 10155 |
+
{
|
| 10156 |
+
"epoch": 6.227741330834115,
|
| 10157 |
+
"grad_norm": 2.1493215560913086,
|
| 10158 |
+
"learning_rate": 9.221532333645736e-06,
|
| 10159 |
+
"loss": 0.0925,
|
| 10160 |
+
"step": 13290
|
| 10161 |
+
},
|
| 10162 |
+
{
|
| 10163 |
+
"epoch": 6.232427366447985,
|
| 10164 |
+
"grad_norm": 2.0909435749053955,
|
| 10165 |
+
"learning_rate": 9.220946579194002e-06,
|
| 10166 |
+
"loss": 0.1007,
|
| 10167 |
+
"step": 13300
|
| 10168 |
+
},
|
| 10169 |
+
{
|
| 10170 |
+
"epoch": 6.237113402061856,
|
| 10171 |
+
"grad_norm": 1.8717360496520996,
|
| 10172 |
+
"learning_rate": 9.220360824742269e-06,
|
| 10173 |
+
"loss": 0.0907,
|
| 10174 |
+
"step": 13310
|
| 10175 |
+
},
|
| 10176 |
+
{
|
| 10177 |
+
"epoch": 6.241799437675726,
|
| 10178 |
+
"grad_norm": 2.817506790161133,
|
| 10179 |
+
"learning_rate": 9.219775070290535e-06,
|
| 10180 |
+
"loss": 0.1027,
|
| 10181 |
+
"step": 13320
|
| 10182 |
+
},
|
| 10183 |
+
{
|
| 10184 |
+
"epoch": 6.246485473289597,
|
| 10185 |
+
"grad_norm": 2.344900369644165,
|
| 10186 |
+
"learning_rate": 9.2191893158388e-06,
|
| 10187 |
+
"loss": 0.0993,
|
| 10188 |
+
"step": 13330
|
| 10189 |
+
},
|
| 10190 |
+
{
|
| 10191 |
+
"epoch": 6.251171508903468,
|
| 10192 |
+
"grad_norm": 1.5306053161621094,
|
| 10193 |
+
"learning_rate": 9.218603561387068e-06,
|
| 10194 |
+
"loss": 0.0717,
|
| 10195 |
+
"step": 13340
|
| 10196 |
+
},
|
| 10197 |
+
{
|
| 10198 |
+
"epoch": 6.2558575445173386,
|
| 10199 |
+
"grad_norm": 1.2907652854919434,
|
| 10200 |
+
"learning_rate": 9.218017806935333e-06,
|
| 10201 |
+
"loss": 0.0813,
|
| 10202 |
+
"step": 13350
|
| 10203 |
+
},
|
| 10204 |
+
{
|
| 10205 |
+
"epoch": 6.260543580131209,
|
| 10206 |
+
"grad_norm": 2.5026895999908447,
|
| 10207 |
+
"learning_rate": 9.2174320524836e-06,
|
| 10208 |
+
"loss": 0.107,
|
| 10209 |
+
"step": 13360
|
| 10210 |
+
},
|
| 10211 |
+
{
|
| 10212 |
+
"epoch": 6.2652296157450795,
|
| 10213 |
+
"grad_norm": 2.076305389404297,
|
| 10214 |
+
"learning_rate": 9.216846298031867e-06,
|
| 10215 |
+
"loss": 0.0725,
|
| 10216 |
+
"step": 13370
|
| 10217 |
+
},
|
| 10218 |
+
{
|
| 10219 |
+
"epoch": 6.26991565135895,
|
| 10220 |
+
"grad_norm": 2.985992908477783,
|
| 10221 |
+
"learning_rate": 9.216260543580132e-06,
|
| 10222 |
+
"loss": 0.1041,
|
| 10223 |
+
"step": 13380
|
| 10224 |
+
},
|
| 10225 |
+
{
|
| 10226 |
+
"epoch": 6.274601686972821,
|
| 10227 |
+
"grad_norm": 1.430763602256775,
|
| 10228 |
+
"learning_rate": 9.215674789128399e-06,
|
| 10229 |
+
"loss": 0.0941,
|
| 10230 |
+
"step": 13390
|
| 10231 |
+
},
|
| 10232 |
+
{
|
| 10233 |
+
"epoch": 6.279287722586692,
|
| 10234 |
+
"grad_norm": 2.2910046577453613,
|
| 10235 |
+
"learning_rate": 9.215089034676664e-06,
|
| 10236 |
+
"loss": 0.07,
|
| 10237 |
+
"step": 13400
|
| 10238 |
+
},
|
| 10239 |
+
{
|
| 10240 |
+
"epoch": 6.283973758200562,
|
| 10241 |
+
"grad_norm": 2.093707799911499,
|
| 10242 |
+
"learning_rate": 9.214503280224931e-06,
|
| 10243 |
+
"loss": 0.085,
|
| 10244 |
+
"step": 13410
|
| 10245 |
+
},
|
| 10246 |
+
{
|
| 10247 |
+
"epoch": 6.288659793814433,
|
| 10248 |
+
"grad_norm": 1.2367407083511353,
|
| 10249 |
+
"learning_rate": 9.213917525773196e-06,
|
| 10250 |
+
"loss": 0.0895,
|
| 10251 |
+
"step": 13420
|
| 10252 |
+
},
|
| 10253 |
+
{
|
| 10254 |
+
"epoch": 6.293345829428303,
|
| 10255 |
+
"grad_norm": 1.8991947174072266,
|
| 10256 |
+
"learning_rate": 9.213331771321463e-06,
|
| 10257 |
+
"loss": 0.0992,
|
| 10258 |
+
"step": 13430
|
| 10259 |
+
},
|
| 10260 |
+
{
|
| 10261 |
+
"epoch": 6.298031865042175,
|
| 10262 |
+
"grad_norm": 2.4072980880737305,
|
| 10263 |
+
"learning_rate": 9.212746016869728e-06,
|
| 10264 |
+
"loss": 0.1033,
|
| 10265 |
+
"step": 13440
|
| 10266 |
+
},
|
| 10267 |
+
{
|
| 10268 |
+
"epoch": 6.302717900656045,
|
| 10269 |
+
"grad_norm": 1.8667056560516357,
|
| 10270 |
+
"learning_rate": 9.212160262417995e-06,
|
| 10271 |
+
"loss": 0.0825,
|
| 10272 |
+
"step": 13450
|
| 10273 |
+
},
|
| 10274 |
+
{
|
| 10275 |
+
"epoch": 6.307403936269916,
|
| 10276 |
+
"grad_norm": 2.0823960304260254,
|
| 10277 |
+
"learning_rate": 9.21157450796626e-06,
|
| 10278 |
+
"loss": 0.095,
|
| 10279 |
+
"step": 13460
|
| 10280 |
+
},
|
| 10281 |
+
{
|
| 10282 |
+
"epoch": 6.312089971883786,
|
| 10283 |
+
"grad_norm": 1.7069936990737915,
|
| 10284 |
+
"learning_rate": 9.210988753514527e-06,
|
| 10285 |
+
"loss": 0.088,
|
| 10286 |
+
"step": 13470
|
| 10287 |
+
},
|
| 10288 |
+
{
|
| 10289 |
+
"epoch": 6.316776007497657,
|
| 10290 |
+
"grad_norm": 2.103296995162964,
|
| 10291 |
+
"learning_rate": 9.210402999062794e-06,
|
| 10292 |
+
"loss": 0.0867,
|
| 10293 |
+
"step": 13480
|
| 10294 |
+
},
|
| 10295 |
+
{
|
| 10296 |
+
"epoch": 6.321462043111528,
|
| 10297 |
+
"grad_norm": 1.0462993383407593,
|
| 10298 |
+
"learning_rate": 9.20981724461106e-06,
|
| 10299 |
+
"loss": 0.0813,
|
| 10300 |
+
"step": 13490
|
| 10301 |
+
},
|
| 10302 |
+
{
|
| 10303 |
+
"epoch": 6.3261480787253985,
|
| 10304 |
+
"grad_norm": 1.5300264358520508,
|
| 10305 |
+
"learning_rate": 9.209231490159326e-06,
|
| 10306 |
+
"loss": 0.0807,
|
| 10307 |
+
"step": 13500
|
| 10308 |
+
},
|
| 10309 |
+
{
|
| 10310 |
+
"epoch": 6.3261480787253985,
|
| 10311 |
+
"eval_loss": 0.040135517716407776,
|
| 10312 |
+
"eval_pearson_cosine": 0.7837547061964756,
|
| 10313 |
+
"eval_pearson_dot": 0.6449662712980881,
|
| 10314 |
+
"eval_pearson_euclidean": 0.7344489051322731,
|
| 10315 |
+
"eval_pearson_manhattan": 0.733928021220585,
|
| 10316 |
+
"eval_runtime": 40.9066,
|
| 10317 |
+
"eval_samples_per_second": 36.669,
|
| 10318 |
+
"eval_spearman_cosine": 0.787895557874529,
|
| 10319 |
+
"eval_spearman_dot": 0.6615144668773622,
|
| 10320 |
+
"eval_spearman_euclidean": 0.7513080603143243,
|
| 10321 |
+
"eval_spearman_manhattan": 0.7509655476151297,
|
| 10322 |
+
"eval_steps_per_second": 36.669,
|
| 10323 |
+
"step": 13500
|
| 10324 |
+
},
|
| 10325 |
+
{
|
| 10326 |
+
"epoch": 6.330834114339269,
|
| 10327 |
+
"grad_norm": 2.420490264892578,
|
| 10328 |
+
"learning_rate": 9.208645735707592e-06,
|
| 10329 |
+
"loss": 0.086,
|
| 10330 |
+
"step": 13510
|
| 10331 |
+
},
|
| 10332 |
+
{
|
| 10333 |
+
"epoch": 6.3355201499531395,
|
| 10334 |
+
"grad_norm": 2.568124771118164,
|
| 10335 |
+
"learning_rate": 9.208059981255859e-06,
|
| 10336 |
+
"loss": 0.0972,
|
| 10337 |
+
"step": 13520
|
| 10338 |
+
},
|
| 10339 |
+
{
|
| 10340 |
+
"epoch": 6.34020618556701,
|
| 10341 |
+
"grad_norm": 2.0971317291259766,
|
| 10342 |
+
"learning_rate": 9.207474226804124e-06,
|
| 10343 |
+
"loss": 0.0851,
|
| 10344 |
+
"step": 13530
|
| 10345 |
+
},
|
| 10346 |
+
{
|
| 10347 |
+
"epoch": 6.344892221180881,
|
| 10348 |
+
"grad_norm": 2.3092575073242188,
|
| 10349 |
+
"learning_rate": 9.20688847235239e-06,
|
| 10350 |
+
"loss": 0.1103,
|
| 10351 |
+
"step": 13540
|
| 10352 |
+
},
|
| 10353 |
+
{
|
| 10354 |
+
"epoch": 6.349578256794752,
|
| 10355 |
+
"grad_norm": 1.0552003383636475,
|
| 10356 |
+
"learning_rate": 9.206302717900658e-06,
|
| 10357 |
+
"loss": 0.0927,
|
| 10358 |
+
"step": 13550
|
| 10359 |
+
},
|
| 10360 |
+
{
|
| 10361 |
+
"epoch": 6.354264292408622,
|
| 10362 |
+
"grad_norm": 2.5293657779693604,
|
| 10363 |
+
"learning_rate": 9.205716963448923e-06,
|
| 10364 |
+
"loss": 0.0948,
|
| 10365 |
+
"step": 13560
|
| 10366 |
+
},
|
| 10367 |
+
{
|
| 10368 |
+
"epoch": 6.358950328022493,
|
| 10369 |
+
"grad_norm": 2.3516149520874023,
|
| 10370 |
+
"learning_rate": 9.20513120899719e-06,
|
| 10371 |
+
"loss": 0.0997,
|
| 10372 |
+
"step": 13570
|
| 10373 |
+
},
|
| 10374 |
+
{
|
| 10375 |
+
"epoch": 6.363636363636363,
|
| 10376 |
+
"grad_norm": 2.1005942821502686,
|
| 10377 |
+
"learning_rate": 9.204545454545455e-06,
|
| 10378 |
+
"loss": 0.0839,
|
| 10379 |
+
"step": 13580
|
| 10380 |
+
},
|
| 10381 |
+
{
|
| 10382 |
+
"epoch": 6.368322399250235,
|
| 10383 |
+
"grad_norm": 1.8200384378433228,
|
| 10384 |
+
"learning_rate": 9.203959700093722e-06,
|
| 10385 |
+
"loss": 0.0876,
|
| 10386 |
+
"step": 13590
|
| 10387 |
+
},
|
| 10388 |
+
{
|
| 10389 |
+
"epoch": 6.373008434864105,
|
| 10390 |
+
"grad_norm": 2.4939422607421875,
|
| 10391 |
+
"learning_rate": 9.203373945641987e-06,
|
| 10392 |
+
"loss": 0.0912,
|
| 10393 |
+
"step": 13600
|
| 10394 |
+
},
|
| 10395 |
+
{
|
| 10396 |
+
"epoch": 6.377694470477976,
|
| 10397 |
+
"grad_norm": 2.695819616317749,
|
| 10398 |
+
"learning_rate": 9.202788191190254e-06,
|
| 10399 |
+
"loss": 0.0797,
|
| 10400 |
+
"step": 13610
|
| 10401 |
+
},
|
| 10402 |
+
{
|
| 10403 |
+
"epoch": 6.382380506091846,
|
| 10404 |
+
"grad_norm": 2.2140467166900635,
|
| 10405 |
+
"learning_rate": 9.20220243673852e-06,
|
| 10406 |
+
"loss": 0.0908,
|
| 10407 |
+
"step": 13620
|
| 10408 |
+
},
|
| 10409 |
+
{
|
| 10410 |
+
"epoch": 6.387066541705717,
|
| 10411 |
+
"grad_norm": 1.4225996732711792,
|
| 10412 |
+
"learning_rate": 9.201616682286786e-06,
|
| 10413 |
+
"loss": 0.0888,
|
| 10414 |
+
"step": 13630
|
| 10415 |
+
},
|
| 10416 |
+
{
|
| 10417 |
+
"epoch": 6.391752577319588,
|
| 10418 |
+
"grad_norm": 1.8974356651306152,
|
| 10419 |
+
"learning_rate": 9.201030927835051e-06,
|
| 10420 |
+
"loss": 0.0721,
|
| 10421 |
+
"step": 13640
|
| 10422 |
+
},
|
| 10423 |
+
{
|
| 10424 |
+
"epoch": 6.3964386129334585,
|
| 10425 |
+
"grad_norm": 2.853360176086426,
|
| 10426 |
+
"learning_rate": 9.200445173383318e-06,
|
| 10427 |
+
"loss": 0.0998,
|
| 10428 |
+
"step": 13650
|
| 10429 |
+
},
|
| 10430 |
+
{
|
| 10431 |
+
"epoch": 6.401124648547329,
|
| 10432 |
+
"grad_norm": 1.7699189186096191,
|
| 10433 |
+
"learning_rate": 9.199859418931585e-06,
|
| 10434 |
+
"loss": 0.075,
|
| 10435 |
+
"step": 13660
|
| 10436 |
+
},
|
| 10437 |
+
{
|
| 10438 |
+
"epoch": 6.4058106841611995,
|
| 10439 |
+
"grad_norm": 1.9779934883117676,
|
| 10440 |
+
"learning_rate": 9.19927366447985e-06,
|
| 10441 |
+
"loss": 0.0854,
|
| 10442 |
+
"step": 13670
|
| 10443 |
+
},
|
| 10444 |
+
{
|
| 10445 |
+
"epoch": 6.41049671977507,
|
| 10446 |
+
"grad_norm": 1.0165759325027466,
|
| 10447 |
+
"learning_rate": 9.198687910028117e-06,
|
| 10448 |
+
"loss": 0.0694,
|
| 10449 |
+
"step": 13680
|
| 10450 |
+
},
|
| 10451 |
+
{
|
| 10452 |
+
"epoch": 6.415182755388941,
|
| 10453 |
+
"grad_norm": 2.2649970054626465,
|
| 10454 |
+
"learning_rate": 9.198102155576383e-06,
|
| 10455 |
+
"loss": 0.1138,
|
| 10456 |
+
"step": 13690
|
| 10457 |
+
},
|
| 10458 |
+
{
|
| 10459 |
+
"epoch": 6.419868791002812,
|
| 10460 |
+
"grad_norm": 1.683463454246521,
|
| 10461 |
+
"learning_rate": 9.19751640112465e-06,
|
| 10462 |
+
"loss": 0.0838,
|
| 10463 |
+
"step": 13700
|
| 10464 |
+
},
|
| 10465 |
+
{
|
| 10466 |
+
"epoch": 6.424554826616682,
|
| 10467 |
+
"grad_norm": 2.579838275909424,
|
| 10468 |
+
"learning_rate": 9.196930646672916e-06,
|
| 10469 |
+
"loss": 0.1073,
|
| 10470 |
+
"step": 13710
|
| 10471 |
+
},
|
| 10472 |
+
{
|
| 10473 |
+
"epoch": 6.429240862230553,
|
| 10474 |
+
"grad_norm": 2.041942834854126,
|
| 10475 |
+
"learning_rate": 9.196344892221182e-06,
|
| 10476 |
+
"loss": 0.0804,
|
| 10477 |
+
"step": 13720
|
| 10478 |
+
},
|
| 10479 |
+
{
|
| 10480 |
+
"epoch": 6.433926897844423,
|
| 10481 |
+
"grad_norm": 1.7387852668762207,
|
| 10482 |
+
"learning_rate": 9.195759137769449e-06,
|
| 10483 |
+
"loss": 0.0994,
|
| 10484 |
+
"step": 13730
|
| 10485 |
+
},
|
| 10486 |
+
{
|
| 10487 |
+
"epoch": 6.438612933458295,
|
| 10488 |
+
"grad_norm": 1.3456875085830688,
|
| 10489 |
+
"learning_rate": 9.195173383317714e-06,
|
| 10490 |
+
"loss": 0.0957,
|
| 10491 |
+
"step": 13740
|
| 10492 |
+
},
|
| 10493 |
+
{
|
| 10494 |
+
"epoch": 6.443298969072165,
|
| 10495 |
+
"grad_norm": 1.8419511318206787,
|
| 10496 |
+
"learning_rate": 9.194587628865979e-06,
|
| 10497 |
+
"loss": 0.0863,
|
| 10498 |
+
"step": 13750
|
| 10499 |
+
},
|
| 10500 |
+
{
|
| 10501 |
+
"epoch": 6.443298969072165,
|
| 10502 |
+
"eval_loss": 0.04049157723784447,
|
| 10503 |
+
"eval_pearson_cosine": 0.7813714988340337,
|
| 10504 |
+
"eval_pearson_dot": 0.6324385596537514,
|
| 10505 |
+
"eval_pearson_euclidean": 0.7407851614356957,
|
| 10506 |
+
"eval_pearson_manhattan": 0.7403546898217854,
|
| 10507 |
+
"eval_runtime": 39.3844,
|
| 10508 |
+
"eval_samples_per_second": 38.086,
|
| 10509 |
+
"eval_spearman_cosine": 0.7841122028697166,
|
| 10510 |
+
"eval_spearman_dot": 0.6478511175419069,
|
| 10511 |
+
"eval_spearman_euclidean": 0.7589082543335365,
|
| 10512 |
+
"eval_spearman_manhattan": 0.7586951153918645,
|
| 10513 |
+
"eval_steps_per_second": 38.086,
|
| 10514 |
+
"step": 13750
|
| 10515 |
+
},
|
| 10516 |
+
{
|
| 10517 |
+
"epoch": 6.447985004686036,
|
| 10518 |
+
"grad_norm": 2.0003013610839844,
|
| 10519 |
+
"learning_rate": 9.194001874414246e-06,
|
| 10520 |
+
"loss": 0.0752,
|
| 10521 |
+
"step": 13760
|
| 10522 |
+
},
|
| 10523 |
+
{
|
| 10524 |
+
"epoch": 6.452671040299906,
|
| 10525 |
+
"grad_norm": 1.2026764154434204,
|
| 10526 |
+
"learning_rate": 9.193416119962513e-06,
|
| 10527 |
+
"loss": 0.0879,
|
| 10528 |
+
"step": 13770
|
| 10529 |
+
},
|
| 10530 |
+
{
|
| 10531 |
+
"epoch": 6.457357075913777,
|
| 10532 |
+
"grad_norm": 1.3370740413665771,
|
| 10533 |
+
"learning_rate": 9.192830365510778e-06,
|
| 10534 |
+
"loss": 0.1001,
|
| 10535 |
+
"step": 13780
|
| 10536 |
+
},
|
| 10537 |
+
{
|
| 10538 |
+
"epoch": 6.462043111527648,
|
| 10539 |
+
"grad_norm": 2.915313482284546,
|
| 10540 |
+
"learning_rate": 9.192244611059045e-06,
|
| 10541 |
+
"loss": 0.0893,
|
| 10542 |
+
"step": 13790
|
| 10543 |
+
},
|
| 10544 |
+
{
|
| 10545 |
+
"epoch": 6.4667291471415185,
|
| 10546 |
+
"grad_norm": 2.472095012664795,
|
| 10547 |
+
"learning_rate": 9.19165885660731e-06,
|
| 10548 |
+
"loss": 0.0833,
|
| 10549 |
+
"step": 13800
|
| 10550 |
+
},
|
| 10551 |
+
{
|
| 10552 |
+
"epoch": 6.471415182755389,
|
| 10553 |
+
"grad_norm": 2.4989047050476074,
|
| 10554 |
+
"learning_rate": 9.191073102155577e-06,
|
| 10555 |
+
"loss": 0.0808,
|
| 10556 |
+
"step": 13810
|
| 10557 |
+
},
|
| 10558 |
+
{
|
| 10559 |
+
"epoch": 6.4761012183692594,
|
| 10560 |
+
"grad_norm": 1.9011194705963135,
|
| 10561 |
+
"learning_rate": 9.190487347703844e-06,
|
| 10562 |
+
"loss": 0.0845,
|
| 10563 |
+
"step": 13820
|
| 10564 |
+
},
|
| 10565 |
+
{
|
| 10566 |
+
"epoch": 6.48078725398313,
|
| 10567 |
+
"grad_norm": 1.4347310066223145,
|
| 10568 |
+
"learning_rate": 9.18990159325211e-06,
|
| 10569 |
+
"loss": 0.1029,
|
| 10570 |
+
"step": 13830
|
| 10571 |
+
},
|
| 10572 |
+
{
|
| 10573 |
+
"epoch": 6.485473289597001,
|
| 10574 |
+
"grad_norm": 2.54189395904541,
|
| 10575 |
+
"learning_rate": 9.189315838800376e-06,
|
| 10576 |
+
"loss": 0.0779,
|
| 10577 |
+
"step": 13840
|
| 10578 |
+
},
|
| 10579 |
+
{
|
| 10580 |
+
"epoch": 6.490159325210872,
|
| 10581 |
+
"grad_norm": 1.2057007551193237,
|
| 10582 |
+
"learning_rate": 9.188730084348641e-06,
|
| 10583 |
+
"loss": 0.0894,
|
| 10584 |
+
"step": 13850
|
| 10585 |
+
},
|
| 10586 |
+
{
|
| 10587 |
+
"epoch": 6.494845360824742,
|
| 10588 |
+
"grad_norm": 1.7278800010681152,
|
| 10589 |
+
"learning_rate": 9.188144329896908e-06,
|
| 10590 |
+
"loss": 0.0779,
|
| 10591 |
+
"step": 13860
|
| 10592 |
+
},
|
| 10593 |
+
{
|
| 10594 |
+
"epoch": 6.499531396438613,
|
| 10595 |
+
"grad_norm": 1.5743318796157837,
|
| 10596 |
+
"learning_rate": 9.187558575445175e-06,
|
| 10597 |
+
"loss": 0.0927,
|
| 10598 |
+
"step": 13870
|
| 10599 |
+
},
|
| 10600 |
+
{
|
| 10601 |
+
"epoch": 6.504217432052483,
|
| 10602 |
+
"grad_norm": 1.5857350826263428,
|
| 10603 |
+
"learning_rate": 9.18697282099344e-06,
|
| 10604 |
+
"loss": 0.0947,
|
| 10605 |
+
"step": 13880
|
| 10606 |
+
},
|
| 10607 |
+
{
|
| 10608 |
+
"epoch": 6.508903467666355,
|
| 10609 |
+
"grad_norm": 1.6598036289215088,
|
| 10610 |
+
"learning_rate": 9.186387066541707e-06,
|
| 10611 |
+
"loss": 0.0686,
|
| 10612 |
+
"step": 13890
|
| 10613 |
+
},
|
| 10614 |
+
{
|
| 10615 |
+
"epoch": 6.513589503280225,
|
| 10616 |
+
"grad_norm": 1.861210823059082,
|
| 10617 |
+
"learning_rate": 9.185801312089973e-06,
|
| 10618 |
+
"loss": 0.0785,
|
| 10619 |
+
"step": 13900
|
| 10620 |
+
},
|
| 10621 |
+
{
|
| 10622 |
+
"epoch": 6.518275538894096,
|
| 10623 |
+
"grad_norm": 2.236298084259033,
|
| 10624 |
+
"learning_rate": 9.185215557638238e-06,
|
| 10625 |
+
"loss": 0.0786,
|
| 10626 |
+
"step": 13910
|
| 10627 |
+
},
|
| 10628 |
+
{
|
| 10629 |
+
"epoch": 6.522961574507966,
|
| 10630 |
+
"grad_norm": 2.7919623851776123,
|
| 10631 |
+
"learning_rate": 9.184629803186505e-06,
|
| 10632 |
+
"loss": 0.091,
|
| 10633 |
+
"step": 13920
|
| 10634 |
+
},
|
| 10635 |
+
{
|
| 10636 |
+
"epoch": 6.527647610121837,
|
| 10637 |
+
"grad_norm": 2.071950674057007,
|
| 10638 |
+
"learning_rate": 9.184044048734772e-06,
|
| 10639 |
+
"loss": 0.0815,
|
| 10640 |
+
"step": 13930
|
| 10641 |
+
},
|
| 10642 |
+
{
|
| 10643 |
+
"epoch": 6.532333645735708,
|
| 10644 |
+
"grad_norm": 2.533116102218628,
|
| 10645 |
+
"learning_rate": 9.183458294283037e-06,
|
| 10646 |
+
"loss": 0.0787,
|
| 10647 |
+
"step": 13940
|
| 10648 |
+
},
|
| 10649 |
+
{
|
| 10650 |
+
"epoch": 6.5370196813495784,
|
| 10651 |
+
"grad_norm": 2.1527774333953857,
|
| 10652 |
+
"learning_rate": 9.182872539831304e-06,
|
| 10653 |
+
"loss": 0.0749,
|
| 10654 |
+
"step": 13950
|
| 10655 |
+
},
|
| 10656 |
+
{
|
| 10657 |
+
"epoch": 6.541705716963449,
|
| 10658 |
+
"grad_norm": 1.961341142654419,
|
| 10659 |
+
"learning_rate": 9.182286785379569e-06,
|
| 10660 |
+
"loss": 0.0953,
|
| 10661 |
+
"step": 13960
|
| 10662 |
+
},
|
| 10663 |
+
{
|
| 10664 |
+
"epoch": 6.546391752577319,
|
| 10665 |
+
"grad_norm": 2.28048038482666,
|
| 10666 |
+
"learning_rate": 9.181701030927836e-06,
|
| 10667 |
+
"loss": 0.0915,
|
| 10668 |
+
"step": 13970
|
| 10669 |
+
},
|
| 10670 |
+
{
|
| 10671 |
+
"epoch": 6.55107778819119,
|
| 10672 |
+
"grad_norm": 2.0521295070648193,
|
| 10673 |
+
"learning_rate": 9.181115276476103e-06,
|
| 10674 |
+
"loss": 0.0895,
|
| 10675 |
+
"step": 13980
|
| 10676 |
+
},
|
| 10677 |
+
{
|
| 10678 |
+
"epoch": 6.555763823805061,
|
| 10679 |
+
"grad_norm": 1.6671199798583984,
|
| 10680 |
+
"learning_rate": 9.180529522024368e-06,
|
| 10681 |
+
"loss": 0.0728,
|
| 10682 |
+
"step": 13990
|
| 10683 |
+
},
|
| 10684 |
+
{
|
| 10685 |
+
"epoch": 6.560449859418932,
|
| 10686 |
+
"grad_norm": 1.6179335117340088,
|
| 10687 |
+
"learning_rate": 9.179943767572635e-06,
|
| 10688 |
+
"loss": 0.0948,
|
| 10689 |
+
"step": 14000
|
| 10690 |
+
},
|
| 10691 |
+
{
|
| 10692 |
+
"epoch": 6.560449859418932,
|
| 10693 |
+
"eval_loss": 0.03971678018569946,
|
| 10694 |
+
"eval_pearson_cosine": 0.7830490563978572,
|
| 10695 |
+
"eval_pearson_dot": 0.6307782966974682,
|
| 10696 |
+
"eval_pearson_euclidean": 0.7414858554074186,
|
| 10697 |
+
"eval_pearson_manhattan": 0.7410214751653541,
|
| 10698 |
+
"eval_runtime": 40.1664,
|
| 10699 |
+
"eval_samples_per_second": 37.345,
|
| 10700 |
+
"eval_spearman_cosine": 0.7866220854382757,
|
| 10701 |
+
"eval_spearman_dot": 0.6460234579111318,
|
| 10702 |
+
"eval_spearman_euclidean": 0.7578877607249501,
|
| 10703 |
+
"eval_spearman_manhattan": 0.757795808702236,
|
| 10704 |
+
"eval_steps_per_second": 37.345,
|
| 10705 |
+
"step": 14000
|
| 10706 |
}
|
| 10707 |
],
|
| 10708 |
"logging_steps": 10,
|