Training in progress, epoch 1
Browse files- all_results.json +12 -0
- eval_results.json +7 -0
- pytorch_model.bin +1 -1
- runs/Oct18_20-30-39_EE4E077/events.out.tfevents.1697633256.EE4E077.354999.2 +3 -0
- runs/Oct18_20-56-45_EE4E077/events.out.tfevents.1697634289.EE4E077.354999.3 +3 -0
- runs/Oct20_23-31-07_EE4E077/1697815939.6338866/events.out.tfevents.1697815939.EE4E077.825718.1 +3 -0
- runs/Oct20_23-31-07_EE4E077/events.out.tfevents.1697815939.EE4E077.825718.0 +3 -0
- runs/Oct20_23-48-31_EE4E077/1697816917.4046345/events.out.tfevents.1697816917.EE4E077.825718.3 +3 -0
- runs/Oct20_23-48-31_EE4E077/events.out.tfevents.1697816917.EE4E077.825718.2 +3 -0
- runs/Oct20_23-58-02_EE4E077/1697817487.9031978/events.out.tfevents.1697817487.EE4E077.825718.5 +3 -0
- runs/Oct20_23-58-02_EE4E077/events.out.tfevents.1697817487.EE4E077.825718.4 +3 -0
- runs/Oct21_00-07-32_EE4E077/1697818095.455782/events.out.tfevents.1697818095.EE4E077.825718.7 +3 -0
- runs/Oct21_00-07-32_EE4E077/1697818577.9317267/events.out.tfevents.1697818577.EE4E077.825718.9 +3 -0
- runs/Oct21_00-07-32_EE4E077/1697819545.3757477/events.out.tfevents.1697819545.EE4E077.825718.11 +3 -0
- runs/Oct21_00-07-32_EE4E077/1697819819.5327485/events.out.tfevents.1697819819.EE4E077.825718.13 +3 -0
- runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697818095.EE4E077.825718.6 +3 -0
- runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697818577.EE4E077.825718.8 +3 -0
- runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697819545.EE4E077.825718.10 +3 -0
- runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697819819.EE4E077.825718.12 +3 -0
- train_results.json +7 -0
- trainer_state.json +1561 -0
- training_args.bin +1 -1
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.78505291005291,
|
| 4 |
+
"eval_loss": 0.6116965413093567,
|
| 5 |
+
"eval_runtime": 8.6246,
|
| 6 |
+
"eval_samples_per_second": 175.312,
|
| 7 |
+
"eval_steps_per_second": 11.015,
|
| 8 |
+
"train_loss": 0.3521563596071336,
|
| 9 |
+
"train_runtime": 855.1502,
|
| 10 |
+
"train_samples_per_second": 46.846,
|
| 11 |
+
"train_steps_per_second": 2.928
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"eval_accuracy": 0.78505291005291,
|
| 3 |
+
"eval_loss": 0.6116965413093567,
|
| 4 |
+
"eval_runtime": 8.6246,
|
| 5 |
+
"eval_samples_per_second": 175.312,
|
| 6 |
+
"eval_steps_per_second": 11.015
|
| 7 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343284077
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36ae3a3be54b486513c0545ac296116960571e64b6f7784e2e3128aa656043bb
|
| 3 |
size 343284077
|
runs/Oct18_20-30-39_EE4E077/events.out.tfevents.1697633256.EE4E077.354999.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d346423ed74a16b8fa996486a5a39d6ea21893989d58e22d2f4b0d21635a0f2c
|
| 3 |
+
size 411
|
runs/Oct18_20-56-45_EE4E077/events.out.tfevents.1697634289.EE4E077.354999.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2e29c66141cf622394f315b4e7eefb3d97cd03450cb0e342ac4c41f49d96912
|
| 3 |
+
size 346
|
runs/Oct20_23-31-07_EE4E077/1697815939.6338866/events.out.tfevents.1697815939.EE4E077.825718.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3daa5d5db05bc7a6556a94230a71db89cc5ca4bd798af8fa39e6731129a15a94
|
| 3 |
+
size 5913
|
runs/Oct20_23-31-07_EE4E077/events.out.tfevents.1697815939.EE4E077.825718.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:16fc0af7c53a01e596e32b940c348e7409f8cc9724d6e53fde7e22ddccce4f94
|
| 3 |
+
size 13851
|
runs/Oct20_23-48-31_EE4E077/1697816917.4046345/events.out.tfevents.1697816917.EE4E077.825718.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a755426dff7a1693683fab156a075fa74477c2a491a5722b5f34658f35c310ac
|
| 3 |
+
size 5913
|
runs/Oct20_23-48-31_EE4E077/events.out.tfevents.1697816917.EE4E077.825718.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35b4a81562d3b195121c18e363f2d81ce9b4318b451786609448c21b821175e6
|
| 3 |
+
size 13900
|
runs/Oct20_23-58-02_EE4E077/1697817487.9031978/events.out.tfevents.1697817487.EE4E077.825718.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef31a96b5996626b75cb688189cbc5f6cbe55303bc181348dfd0c0ea8e470454
|
| 3 |
+
size 5913
|
runs/Oct20_23-58-02_EE4E077/events.out.tfevents.1697817487.EE4E077.825718.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4492e50de039b059739140ead2e03d2069e2fba97caed4574b2891b13c15e6c
|
| 3 |
+
size 13900
|
runs/Oct21_00-07-32_EE4E077/1697818095.455782/events.out.tfevents.1697818095.EE4E077.825718.7
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:485a30fa90c781b0375bd2b14c2d2caef0e6697771a7e5972711033f32cfa1ca
|
| 3 |
+
size 5913
|
runs/Oct21_00-07-32_EE4E077/1697818577.9317267/events.out.tfevents.1697818577.EE4E077.825718.9
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e38185b41a5c814df7c65e2b4bf7eba2d47511da36851e703f743631d81e4647
|
| 3 |
+
size 5913
|
runs/Oct21_00-07-32_EE4E077/1697819545.3757477/events.out.tfevents.1697819545.EE4E077.825718.11
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9017676182dad99635cd62905ffba2abb0ac52d5bc2e5cc8c3326aeff3366484
|
| 3 |
+
size 5913
|
runs/Oct21_00-07-32_EE4E077/1697819819.5327485/events.out.tfevents.1697819819.EE4E077.825718.13
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79f2b0f4d4d3ebafc342de48424ef9bb535cfe02a41389094f8ba52d6ea1f2a6
|
| 3 |
+
size 5913
|
runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697818095.EE4E077.825718.6
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34ef466aba58b5e1d8300d6fe10dd2a2fe05868a86494079313dedfedcbdc713
|
| 3 |
+
size 13851
|
runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697818577.EE4E077.825718.8
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18abb38353de18110f7ee6516b4128ef06b12b50f8815a55a60ab7c380b67a65
|
| 3 |
+
size 14122
|
runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697819545.EE4E077.825718.10
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fedc3d79ea5188d65a913dc6b9fda990f4b6e737fb5ab797f36bd30734323b3d
|
| 3 |
+
size 13945
|
runs/Oct21_00-07-32_EE4E077/events.out.tfevents.1697819819.EE4E077.825718.12
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d09e8f5c65814a6dacbde927ac35d9cccbffb5a6d8953995546bc4e5c737d772
|
| 3 |
+
size 14216
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"train_loss": 0.3521563596071336,
|
| 4 |
+
"train_runtime": 855.1502,
|
| 5 |
+
"train_samples_per_second": 46.846,
|
| 6 |
+
"train_steps_per_second": 2.928
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1561 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8911917098445595,
|
| 3 |
+
"best_model_checkpoint": "./vit-base-skin/checkpoint-1252",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 2504,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 0.00019920127795527157,
|
| 13 |
+
"loss": 1.3607,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.03,
|
| 18 |
+
"learning_rate": 0.00019840255591054313,
|
| 19 |
+
"loss": 1.0863,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.05,
|
| 24 |
+
"learning_rate": 0.00019760383386581472,
|
| 25 |
+
"loss": 0.8928,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.06,
|
| 30 |
+
"learning_rate": 0.00019680511182108628,
|
| 31 |
+
"loss": 1.0954,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.08,
|
| 36 |
+
"learning_rate": 0.00019600638977635784,
|
| 37 |
+
"loss": 0.7844,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.1,
|
| 42 |
+
"learning_rate": 0.0001952076677316294,
|
| 43 |
+
"loss": 0.8456,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.11,
|
| 48 |
+
"learning_rate": 0.00019440894568690097,
|
| 49 |
+
"loss": 0.8827,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.13,
|
| 54 |
+
"learning_rate": 0.00019361022364217253,
|
| 55 |
+
"loss": 0.8214,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.14,
|
| 60 |
+
"learning_rate": 0.0001928115015974441,
|
| 61 |
+
"loss": 0.9063,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.16,
|
| 66 |
+
"learning_rate": 0.00019201277955271565,
|
| 67 |
+
"loss": 0.8956,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.18,
|
| 72 |
+
"learning_rate": 0.0001912140575079872,
|
| 73 |
+
"loss": 0.9014,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.19,
|
| 78 |
+
"learning_rate": 0.0001904153354632588,
|
| 79 |
+
"loss": 0.83,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.21,
|
| 84 |
+
"learning_rate": 0.00018961661341853036,
|
| 85 |
+
"loss": 0.7877,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.22,
|
| 90 |
+
"learning_rate": 0.00018881789137380192,
|
| 91 |
+
"loss": 0.6501,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.24,
|
| 96 |
+
"learning_rate": 0.00018801916932907348,
|
| 97 |
+
"loss": 0.8166,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.26,
|
| 102 |
+
"learning_rate": 0.00018722044728434505,
|
| 103 |
+
"loss": 0.8288,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.27,
|
| 108 |
+
"learning_rate": 0.00018642172523961663,
|
| 109 |
+
"loss": 0.662,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.29,
|
| 114 |
+
"learning_rate": 0.0001856230031948882,
|
| 115 |
+
"loss": 0.7948,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.3,
|
| 120 |
+
"learning_rate": 0.00018482428115015976,
|
| 121 |
+
"loss": 0.6333,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.32,
|
| 126 |
+
"learning_rate": 0.00018402555910543132,
|
| 127 |
+
"loss": 0.7313,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.34,
|
| 132 |
+
"learning_rate": 0.00018322683706070288,
|
| 133 |
+
"loss": 0.7396,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.35,
|
| 138 |
+
"learning_rate": 0.00018242811501597444,
|
| 139 |
+
"loss": 0.683,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.37,
|
| 144 |
+
"learning_rate": 0.000181629392971246,
|
| 145 |
+
"loss": 0.6597,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.38,
|
| 150 |
+
"learning_rate": 0.00018083067092651756,
|
| 151 |
+
"loss": 0.6867,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.4,
|
| 156 |
+
"learning_rate": 0.00018003194888178913,
|
| 157 |
+
"loss": 0.5824,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.42,
|
| 162 |
+
"learning_rate": 0.00017923322683706071,
|
| 163 |
+
"loss": 0.7605,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.43,
|
| 168 |
+
"learning_rate": 0.00017843450479233228,
|
| 169 |
+
"loss": 0.7816,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.45,
|
| 174 |
+
"learning_rate": 0.00017763578274760384,
|
| 175 |
+
"loss": 0.7451,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.46,
|
| 180 |
+
"learning_rate": 0.0001768370607028754,
|
| 181 |
+
"loss": 0.7161,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.48,
|
| 186 |
+
"learning_rate": 0.000176038338658147,
|
| 187 |
+
"loss": 0.6315,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.5,
|
| 192 |
+
"learning_rate": 0.00017523961661341855,
|
| 193 |
+
"loss": 0.6246,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.51,
|
| 198 |
+
"learning_rate": 0.0001744408945686901,
|
| 199 |
+
"loss": 0.5847,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.53,
|
| 204 |
+
"learning_rate": 0.00017364217252396167,
|
| 205 |
+
"loss": 0.632,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.54,
|
| 210 |
+
"learning_rate": 0.00017284345047923323,
|
| 211 |
+
"loss": 0.7901,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.56,
|
| 216 |
+
"learning_rate": 0.00017204472843450482,
|
| 217 |
+
"loss": 0.6926,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.58,
|
| 222 |
+
"learning_rate": 0.00017124600638977638,
|
| 223 |
+
"loss": 0.6079,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.59,
|
| 228 |
+
"learning_rate": 0.00017044728434504792,
|
| 229 |
+
"loss": 0.6323,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.61,
|
| 234 |
+
"learning_rate": 0.00016964856230031948,
|
| 235 |
+
"loss": 0.6733,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.62,
|
| 240 |
+
"learning_rate": 0.00016884984025559104,
|
| 241 |
+
"loss": 0.6248,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.64,
|
| 246 |
+
"learning_rate": 0.00016805111821086263,
|
| 247 |
+
"loss": 0.7995,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.65,
|
| 252 |
+
"learning_rate": 0.0001672523961661342,
|
| 253 |
+
"loss": 0.7554,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.67,
|
| 258 |
+
"learning_rate": 0.00016645367412140575,
|
| 259 |
+
"loss": 0.6262,
|
| 260 |
+
"step": 420
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.69,
|
| 264 |
+
"learning_rate": 0.0001656549520766773,
|
| 265 |
+
"loss": 0.71,
|
| 266 |
+
"step": 430
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.7,
|
| 270 |
+
"learning_rate": 0.0001648562300319489,
|
| 271 |
+
"loss": 0.6281,
|
| 272 |
+
"step": 440
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.72,
|
| 276 |
+
"learning_rate": 0.00016405750798722046,
|
| 277 |
+
"loss": 0.6005,
|
| 278 |
+
"step": 450
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.73,
|
| 282 |
+
"learning_rate": 0.00016325878594249202,
|
| 283 |
+
"loss": 0.5122,
|
| 284 |
+
"step": 460
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.75,
|
| 288 |
+
"learning_rate": 0.00016246006389776359,
|
| 289 |
+
"loss": 0.5836,
|
| 290 |
+
"step": 470
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.77,
|
| 294 |
+
"learning_rate": 0.00016166134185303515,
|
| 295 |
+
"loss": 0.5401,
|
| 296 |
+
"step": 480
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.78,
|
| 300 |
+
"learning_rate": 0.00016086261980830674,
|
| 301 |
+
"loss": 0.8159,
|
| 302 |
+
"step": 490
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.8,
|
| 306 |
+
"learning_rate": 0.0001600638977635783,
|
| 307 |
+
"loss": 0.632,
|
| 308 |
+
"step": 500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.81,
|
| 312 |
+
"learning_rate": 0.00015926517571884986,
|
| 313 |
+
"loss": 0.6293,
|
| 314 |
+
"step": 510
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.83,
|
| 318 |
+
"learning_rate": 0.00015846645367412142,
|
| 319 |
+
"loss": 0.5831,
|
| 320 |
+
"step": 520
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.85,
|
| 324 |
+
"learning_rate": 0.00015766773162939298,
|
| 325 |
+
"loss": 0.4642,
|
| 326 |
+
"step": 530
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.86,
|
| 330 |
+
"learning_rate": 0.00015686900958466454,
|
| 331 |
+
"loss": 0.6678,
|
| 332 |
+
"step": 540
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.88,
|
| 336 |
+
"learning_rate": 0.0001560702875399361,
|
| 337 |
+
"loss": 0.6439,
|
| 338 |
+
"step": 550
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.89,
|
| 342 |
+
"learning_rate": 0.00015527156549520767,
|
| 343 |
+
"loss": 0.6224,
|
| 344 |
+
"step": 560
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.91,
|
| 348 |
+
"learning_rate": 0.00015447284345047923,
|
| 349 |
+
"loss": 0.5306,
|
| 350 |
+
"step": 570
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.93,
|
| 354 |
+
"learning_rate": 0.00015367412140575082,
|
| 355 |
+
"loss": 0.4909,
|
| 356 |
+
"step": 580
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.94,
|
| 360 |
+
"learning_rate": 0.00015287539936102238,
|
| 361 |
+
"loss": 0.4913,
|
| 362 |
+
"step": 590
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.96,
|
| 366 |
+
"learning_rate": 0.00015207667731629394,
|
| 367 |
+
"loss": 0.5959,
|
| 368 |
+
"step": 600
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.97,
|
| 372 |
+
"learning_rate": 0.0001512779552715655,
|
| 373 |
+
"loss": 0.7232,
|
| 374 |
+
"step": 610
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.99,
|
| 378 |
+
"learning_rate": 0.00015047923322683706,
|
| 379 |
+
"loss": 0.4987,
|
| 380 |
+
"step": 620
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.0,
|
| 384 |
+
"eval_accuracy": 0.8031088082901554,
|
| 385 |
+
"eval_loss": 0.5480978488922119,
|
| 386 |
+
"eval_runtime": 1.2438,
|
| 387 |
+
"eval_samples_per_second": 155.169,
|
| 388 |
+
"eval_steps_per_second": 10.452,
|
| 389 |
+
"step": 626
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.01,
|
| 393 |
+
"learning_rate": 0.00014968051118210865,
|
| 394 |
+
"loss": 0.5424,
|
| 395 |
+
"step": 630
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.02,
|
| 399 |
+
"learning_rate": 0.0001488817891373802,
|
| 400 |
+
"loss": 0.3795,
|
| 401 |
+
"step": 640
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.04,
|
| 405 |
+
"learning_rate": 0.00014808306709265177,
|
| 406 |
+
"loss": 0.5057,
|
| 407 |
+
"step": 650
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.05,
|
| 411 |
+
"learning_rate": 0.00014728434504792333,
|
| 412 |
+
"loss": 0.5375,
|
| 413 |
+
"step": 660
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.07,
|
| 417 |
+
"learning_rate": 0.0001464856230031949,
|
| 418 |
+
"loss": 0.5068,
|
| 419 |
+
"step": 670
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.09,
|
| 423 |
+
"learning_rate": 0.00014568690095846646,
|
| 424 |
+
"loss": 0.5207,
|
| 425 |
+
"step": 680
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.1,
|
| 429 |
+
"learning_rate": 0.00014488817891373802,
|
| 430 |
+
"loss": 0.4394,
|
| 431 |
+
"step": 690
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.12,
|
| 435 |
+
"learning_rate": 0.00014408945686900958,
|
| 436 |
+
"loss": 0.3614,
|
| 437 |
+
"step": 700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.13,
|
| 441 |
+
"learning_rate": 0.00014329073482428114,
|
| 442 |
+
"loss": 0.5282,
|
| 443 |
+
"step": 710
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.15,
|
| 447 |
+
"learning_rate": 0.00014249201277955273,
|
| 448 |
+
"loss": 0.4466,
|
| 449 |
+
"step": 720
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.17,
|
| 453 |
+
"learning_rate": 0.0001416932907348243,
|
| 454 |
+
"loss": 0.4081,
|
| 455 |
+
"step": 730
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.18,
|
| 459 |
+
"learning_rate": 0.00014089456869009585,
|
| 460 |
+
"loss": 0.4235,
|
| 461 |
+
"step": 740
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.2,
|
| 465 |
+
"learning_rate": 0.0001400958466453674,
|
| 466 |
+
"loss": 0.3791,
|
| 467 |
+
"step": 750
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 1.21,
|
| 471 |
+
"learning_rate": 0.000139297124600639,
|
| 472 |
+
"loss": 0.4024,
|
| 473 |
+
"step": 760
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.23,
|
| 477 |
+
"learning_rate": 0.00013849840255591056,
|
| 478 |
+
"loss": 0.3729,
|
| 479 |
+
"step": 770
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.25,
|
| 483 |
+
"learning_rate": 0.00013769968051118212,
|
| 484 |
+
"loss": 0.4715,
|
| 485 |
+
"step": 780
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.26,
|
| 489 |
+
"learning_rate": 0.00013690095846645369,
|
| 490 |
+
"loss": 0.4664,
|
| 491 |
+
"step": 790
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 1.28,
|
| 495 |
+
"learning_rate": 0.00013610223642172525,
|
| 496 |
+
"loss": 0.4945,
|
| 497 |
+
"step": 800
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.29,
|
| 501 |
+
"learning_rate": 0.0001353035143769968,
|
| 502 |
+
"loss": 0.4936,
|
| 503 |
+
"step": 810
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.31,
|
| 507 |
+
"learning_rate": 0.00013450479233226837,
|
| 508 |
+
"loss": 0.5814,
|
| 509 |
+
"step": 820
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 1.33,
|
| 513 |
+
"learning_rate": 0.00013370607028753993,
|
| 514 |
+
"loss": 0.3679,
|
| 515 |
+
"step": 830
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.34,
|
| 519 |
+
"learning_rate": 0.0001329073482428115,
|
| 520 |
+
"loss": 0.4897,
|
| 521 |
+
"step": 840
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.36,
|
| 525 |
+
"learning_rate": 0.00013210862619808305,
|
| 526 |
+
"loss": 0.4264,
|
| 527 |
+
"step": 850
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 1.37,
|
| 531 |
+
"learning_rate": 0.00013130990415335464,
|
| 532 |
+
"loss": 0.6188,
|
| 533 |
+
"step": 860
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 1.39,
|
| 537 |
+
"learning_rate": 0.0001305111821086262,
|
| 538 |
+
"loss": 0.4897,
|
| 539 |
+
"step": 870
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 1.41,
|
| 543 |
+
"learning_rate": 0.00012971246006389777,
|
| 544 |
+
"loss": 0.3392,
|
| 545 |
+
"step": 880
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.42,
|
| 549 |
+
"learning_rate": 0.00012891373801916933,
|
| 550 |
+
"loss": 0.5327,
|
| 551 |
+
"step": 890
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 1.44,
|
| 555 |
+
"learning_rate": 0.00012811501597444092,
|
| 556 |
+
"loss": 0.3651,
|
| 557 |
+
"step": 900
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.45,
|
| 561 |
+
"learning_rate": 0.00012731629392971248,
|
| 562 |
+
"loss": 0.4061,
|
| 563 |
+
"step": 910
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.47,
|
| 567 |
+
"learning_rate": 0.00012651757188498404,
|
| 568 |
+
"loss": 0.4701,
|
| 569 |
+
"step": 920
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.49,
|
| 573 |
+
"learning_rate": 0.0001257188498402556,
|
| 574 |
+
"loss": 0.3794,
|
| 575 |
+
"step": 930
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 1.5,
|
| 579 |
+
"learning_rate": 0.00012492012779552716,
|
| 580 |
+
"loss": 0.5627,
|
| 581 |
+
"step": 940
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.52,
|
| 585 |
+
"learning_rate": 0.00012412140575079872,
|
| 586 |
+
"loss": 0.514,
|
| 587 |
+
"step": 950
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.53,
|
| 591 |
+
"learning_rate": 0.00012332268370607028,
|
| 592 |
+
"loss": 0.424,
|
| 593 |
+
"step": 960
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.55,
|
| 597 |
+
"learning_rate": 0.00012252396166134185,
|
| 598 |
+
"loss": 0.4829,
|
| 599 |
+
"step": 970
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.57,
|
| 603 |
+
"learning_rate": 0.00012172523961661342,
|
| 604 |
+
"loss": 0.387,
|
| 605 |
+
"step": 980
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.58,
|
| 609 |
+
"learning_rate": 0.000120926517571885,
|
| 610 |
+
"loss": 0.472,
|
| 611 |
+
"step": 990
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.6,
|
| 615 |
+
"learning_rate": 0.00012012779552715656,
|
| 616 |
+
"loss": 0.3891,
|
| 617 |
+
"step": 1000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.61,
|
| 621 |
+
"learning_rate": 0.00011932907348242812,
|
| 622 |
+
"loss": 0.4143,
|
| 623 |
+
"step": 1010
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.63,
|
| 627 |
+
"learning_rate": 0.00011853035143769968,
|
| 628 |
+
"loss": 0.4191,
|
| 629 |
+
"step": 1020
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.65,
|
| 633 |
+
"learning_rate": 0.00011773162939297124,
|
| 634 |
+
"loss": 0.44,
|
| 635 |
+
"step": 1030
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 1.66,
|
| 639 |
+
"learning_rate": 0.00011693290734824283,
|
| 640 |
+
"loss": 0.4942,
|
| 641 |
+
"step": 1040
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.68,
|
| 645 |
+
"learning_rate": 0.00011613418530351439,
|
| 646 |
+
"loss": 0.4446,
|
| 647 |
+
"step": 1050
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.69,
|
| 651 |
+
"learning_rate": 0.00011533546325878595,
|
| 652 |
+
"loss": 0.3974,
|
| 653 |
+
"step": 1060
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 1.71,
|
| 657 |
+
"learning_rate": 0.00011453674121405751,
|
| 658 |
+
"loss": 0.4112,
|
| 659 |
+
"step": 1070
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 1.73,
|
| 663 |
+
"learning_rate": 0.00011373801916932908,
|
| 664 |
+
"loss": 0.4987,
|
| 665 |
+
"step": 1080
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.74,
|
| 669 |
+
"learning_rate": 0.00011293929712460065,
|
| 670 |
+
"loss": 0.3738,
|
| 671 |
+
"step": 1090
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.76,
|
| 675 |
+
"learning_rate": 0.00011214057507987221,
|
| 676 |
+
"loss": 0.3729,
|
| 677 |
+
"step": 1100
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.77,
|
| 681 |
+
"learning_rate": 0.00011134185303514377,
|
| 682 |
+
"loss": 0.582,
|
| 683 |
+
"step": 1110
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.79,
|
| 687 |
+
"learning_rate": 0.00011054313099041533,
|
| 688 |
+
"loss": 0.3539,
|
| 689 |
+
"step": 1120
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.81,
|
| 693 |
+
"learning_rate": 0.00010974440894568691,
|
| 694 |
+
"loss": 0.222,
|
| 695 |
+
"step": 1130
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 1.82,
|
| 699 |
+
"learning_rate": 0.00010894568690095847,
|
| 700 |
+
"loss": 0.2987,
|
| 701 |
+
"step": 1140
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.84,
|
| 705 |
+
"learning_rate": 0.00010814696485623003,
|
| 706 |
+
"loss": 0.5448,
|
| 707 |
+
"step": 1150
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.85,
|
| 711 |
+
"learning_rate": 0.0001073482428115016,
|
| 712 |
+
"loss": 0.377,
|
| 713 |
+
"step": 1160
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.87,
|
| 717 |
+
"learning_rate": 0.00010654952076677316,
|
| 718 |
+
"loss": 0.3518,
|
| 719 |
+
"step": 1170
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 1.88,
|
| 723 |
+
"learning_rate": 0.00010575079872204474,
|
| 724 |
+
"loss": 0.5725,
|
| 725 |
+
"step": 1180
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.9,
|
| 729 |
+
"learning_rate": 0.0001049520766773163,
|
| 730 |
+
"loss": 0.2406,
|
| 731 |
+
"step": 1190
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.92,
|
| 735 |
+
"learning_rate": 0.00010415335463258787,
|
| 736 |
+
"loss": 0.3681,
|
| 737 |
+
"step": 1200
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.93,
|
| 741 |
+
"learning_rate": 0.00010335463258785943,
|
| 742 |
+
"loss": 0.4454,
|
| 743 |
+
"step": 1210
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.95,
|
| 747 |
+
"learning_rate": 0.000102555910543131,
|
| 748 |
+
"loss": 0.2933,
|
| 749 |
+
"step": 1220
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.96,
|
| 753 |
+
"learning_rate": 0.00010175718849840256,
|
| 754 |
+
"loss": 0.3093,
|
| 755 |
+
"step": 1230
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.98,
|
| 759 |
+
"learning_rate": 0.00010095846645367413,
|
| 760 |
+
"loss": 0.4952,
|
| 761 |
+
"step": 1240
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 2.0,
|
| 765 |
+
"learning_rate": 0.00010015974440894569,
|
| 766 |
+
"loss": 0.3814,
|
| 767 |
+
"step": 1250
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 2.0,
|
| 771 |
+
"eval_accuracy": 0.8911917098445595,
|
| 772 |
+
"eval_loss": 0.38037967681884766,
|
| 773 |
+
"eval_runtime": 1.2431,
|
| 774 |
+
"eval_samples_per_second": 155.258,
|
| 775 |
+
"eval_steps_per_second": 10.458,
|
| 776 |
+
"step": 1252
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 2.01,
|
| 780 |
+
"learning_rate": 9.936102236421726e-05,
|
| 781 |
+
"loss": 0.347,
|
| 782 |
+
"step": 1260
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.03,
|
| 786 |
+
"learning_rate": 9.856230031948882e-05,
|
| 787 |
+
"loss": 0.1718,
|
| 788 |
+
"step": 1270
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.04,
|
| 792 |
+
"learning_rate": 9.77635782747604e-05,
|
| 793 |
+
"loss": 0.2688,
|
| 794 |
+
"step": 1280
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 2.06,
|
| 798 |
+
"learning_rate": 9.696485623003195e-05,
|
| 799 |
+
"loss": 0.2019,
|
| 800 |
+
"step": 1290
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 2.08,
|
| 804 |
+
"learning_rate": 9.616613418530351e-05,
|
| 805 |
+
"loss": 0.2318,
|
| 806 |
+
"step": 1300
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.09,
|
| 810 |
+
"learning_rate": 9.536741214057508e-05,
|
| 811 |
+
"loss": 0.2364,
|
| 812 |
+
"step": 1310
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.11,
|
| 816 |
+
"learning_rate": 9.456869009584664e-05,
|
| 817 |
+
"loss": 0.2729,
|
| 818 |
+
"step": 1320
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.12,
|
| 822 |
+
"learning_rate": 9.376996805111822e-05,
|
| 823 |
+
"loss": 0.2046,
|
| 824 |
+
"step": 1330
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.14,
|
| 828 |
+
"learning_rate": 9.297124600638978e-05,
|
| 829 |
+
"loss": 0.2405,
|
| 830 |
+
"step": 1340
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.16,
|
| 834 |
+
"learning_rate": 9.217252396166136e-05,
|
| 835 |
+
"loss": 0.1687,
|
| 836 |
+
"step": 1350
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 2.17,
|
| 840 |
+
"learning_rate": 9.137380191693292e-05,
|
| 841 |
+
"loss": 0.2252,
|
| 842 |
+
"step": 1360
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.19,
|
| 846 |
+
"learning_rate": 9.057507987220448e-05,
|
| 847 |
+
"loss": 0.1544,
|
| 848 |
+
"step": 1370
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.2,
|
| 852 |
+
"learning_rate": 8.977635782747604e-05,
|
| 853 |
+
"loss": 0.2436,
|
| 854 |
+
"step": 1380
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 2.22,
|
| 858 |
+
"learning_rate": 8.89776357827476e-05,
|
| 859 |
+
"loss": 0.2051,
|
| 860 |
+
"step": 1390
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.24,
|
| 864 |
+
"learning_rate": 8.817891373801918e-05,
|
| 865 |
+
"loss": 0.163,
|
| 866 |
+
"step": 1400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.25,
|
| 870 |
+
"learning_rate": 8.738019169329074e-05,
|
| 871 |
+
"loss": 0.2489,
|
| 872 |
+
"step": 1410
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.27,
|
| 876 |
+
"learning_rate": 8.658146964856231e-05,
|
| 877 |
+
"loss": 0.2738,
|
| 878 |
+
"step": 1420
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.28,
|
| 882 |
+
"learning_rate": 8.578274760383387e-05,
|
| 883 |
+
"loss": 0.1508,
|
| 884 |
+
"step": 1430
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.3,
|
| 888 |
+
"learning_rate": 8.498402555910544e-05,
|
| 889 |
+
"loss": 0.1544,
|
| 890 |
+
"step": 1440
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 2.32,
|
| 894 |
+
"learning_rate": 8.4185303514377e-05,
|
| 895 |
+
"loss": 0.2727,
|
| 896 |
+
"step": 1450
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 2.33,
|
| 900 |
+
"learning_rate": 8.338658146964856e-05,
|
| 901 |
+
"loss": 0.1686,
|
| 902 |
+
"step": 1460
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.35,
|
| 906 |
+
"learning_rate": 8.258785942492013e-05,
|
| 907 |
+
"loss": 0.1416,
|
| 908 |
+
"step": 1470
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 2.36,
|
| 912 |
+
"learning_rate": 8.17891373801917e-05,
|
| 913 |
+
"loss": 0.4586,
|
| 914 |
+
"step": 1480
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 2.38,
|
| 918 |
+
"learning_rate": 8.099041533546327e-05,
|
| 919 |
+
"loss": 0.1554,
|
| 920 |
+
"step": 1490
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.4,
|
| 924 |
+
"learning_rate": 8.019169329073483e-05,
|
| 925 |
+
"loss": 0.2235,
|
| 926 |
+
"step": 1500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 2.41,
|
| 930 |
+
"learning_rate": 7.939297124600639e-05,
|
| 931 |
+
"loss": 0.1812,
|
| 932 |
+
"step": 1510
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.43,
|
| 936 |
+
"learning_rate": 7.859424920127795e-05,
|
| 937 |
+
"loss": 0.1648,
|
| 938 |
+
"step": 1520
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 2.44,
|
| 942 |
+
"learning_rate": 7.779552715654951e-05,
|
| 943 |
+
"loss": 0.1314,
|
| 944 |
+
"step": 1530
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.46,
|
| 948 |
+
"learning_rate": 7.699680511182109e-05,
|
| 949 |
+
"loss": 0.2276,
|
| 950 |
+
"step": 1540
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 2.48,
|
| 954 |
+
"learning_rate": 7.619808306709265e-05,
|
| 955 |
+
"loss": 0.1624,
|
| 956 |
+
"step": 1550
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.49,
|
| 960 |
+
"learning_rate": 7.539936102236423e-05,
|
| 961 |
+
"loss": 0.1603,
|
| 962 |
+
"step": 1560
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.51,
|
| 966 |
+
"learning_rate": 7.460063897763579e-05,
|
| 967 |
+
"loss": 0.2533,
|
| 968 |
+
"step": 1570
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 2.52,
|
| 972 |
+
"learning_rate": 7.380191693290735e-05,
|
| 973 |
+
"loss": 0.1245,
|
| 974 |
+
"step": 1580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.54,
|
| 978 |
+
"learning_rate": 7.300319488817891e-05,
|
| 979 |
+
"loss": 0.1848,
|
| 980 |
+
"step": 1590
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 2.56,
|
| 984 |
+
"learning_rate": 7.220447284345049e-05,
|
| 985 |
+
"loss": 0.223,
|
| 986 |
+
"step": 1600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.57,
|
| 990 |
+
"learning_rate": 7.140575079872205e-05,
|
| 991 |
+
"loss": 0.1223,
|
| 992 |
+
"step": 1610
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 2.59,
|
| 996 |
+
"learning_rate": 7.060702875399361e-05,
|
| 997 |
+
"loss": 0.309,
|
| 998 |
+
"step": 1620
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.6,
|
| 1002 |
+
"learning_rate": 6.980830670926518e-05,
|
| 1003 |
+
"loss": 0.354,
|
| 1004 |
+
"step": 1630
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.62,
|
| 1008 |
+
"learning_rate": 6.900958466453674e-05,
|
| 1009 |
+
"loss": 0.2051,
|
| 1010 |
+
"step": 1640
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 2.64,
|
| 1014 |
+
"learning_rate": 6.821086261980832e-05,
|
| 1015 |
+
"loss": 0.2932,
|
| 1016 |
+
"step": 1650
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.65,
|
| 1020 |
+
"learning_rate": 6.741214057507987e-05,
|
| 1021 |
+
"loss": 0.1401,
|
| 1022 |
+
"step": 1660
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 2.67,
|
| 1026 |
+
"learning_rate": 6.661341853035144e-05,
|
| 1027 |
+
"loss": 0.3012,
|
| 1028 |
+
"step": 1670
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.68,
|
| 1032 |
+
"learning_rate": 6.5814696485623e-05,
|
| 1033 |
+
"loss": 0.1481,
|
| 1034 |
+
"step": 1680
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.7,
|
| 1038 |
+
"learning_rate": 6.501597444089457e-05,
|
| 1039 |
+
"loss": 0.3556,
|
| 1040 |
+
"step": 1690
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 2.72,
|
| 1044 |
+
"learning_rate": 6.421725239616614e-05,
|
| 1045 |
+
"loss": 0.1193,
|
| 1046 |
+
"step": 1700
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.73,
|
| 1050 |
+
"learning_rate": 6.34185303514377e-05,
|
| 1051 |
+
"loss": 0.2956,
|
| 1052 |
+
"step": 1710
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 2.75,
|
| 1056 |
+
"learning_rate": 6.261980830670928e-05,
|
| 1057 |
+
"loss": 0.164,
|
| 1058 |
+
"step": 1720
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.76,
|
| 1062 |
+
"learning_rate": 6.182108626198084e-05,
|
| 1063 |
+
"loss": 0.1218,
|
| 1064 |
+
"step": 1730
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 2.78,
|
| 1068 |
+
"learning_rate": 6.1022364217252406e-05,
|
| 1069 |
+
"loss": 0.094,
|
| 1070 |
+
"step": 1740
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.8,
|
| 1074 |
+
"learning_rate": 6.022364217252396e-05,
|
| 1075 |
+
"loss": 0.1787,
|
| 1076 |
+
"step": 1750
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.81,
|
| 1080 |
+
"learning_rate": 5.942492012779552e-05,
|
| 1081 |
+
"loss": 0.1833,
|
| 1082 |
+
"step": 1760
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 2.83,
|
| 1086 |
+
"learning_rate": 5.86261980830671e-05,
|
| 1087 |
+
"loss": 0.1389,
|
| 1088 |
+
"step": 1770
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.84,
|
| 1092 |
+
"learning_rate": 5.782747603833866e-05,
|
| 1093 |
+
"loss": 0.2109,
|
| 1094 |
+
"step": 1780
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 2.86,
|
| 1098 |
+
"learning_rate": 5.702875399361023e-05,
|
| 1099 |
+
"loss": 0.1668,
|
| 1100 |
+
"step": 1790
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 2.88,
|
| 1104 |
+
"learning_rate": 5.623003194888179e-05,
|
| 1105 |
+
"loss": 0.1651,
|
| 1106 |
+
"step": 1800
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 2.89,
|
| 1110 |
+
"learning_rate": 5.543130990415336e-05,
|
| 1111 |
+
"loss": 0.2081,
|
| 1112 |
+
"step": 1810
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.91,
|
| 1116 |
+
"learning_rate": 5.4632587859424925e-05,
|
| 1117 |
+
"loss": 0.2506,
|
| 1118 |
+
"step": 1820
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 2.92,
|
| 1122 |
+
"learning_rate": 5.383386581469649e-05,
|
| 1123 |
+
"loss": 0.1272,
|
| 1124 |
+
"step": 1830
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 2.94,
|
| 1128 |
+
"learning_rate": 5.3035143769968054e-05,
|
| 1129 |
+
"loss": 0.1332,
|
| 1130 |
+
"step": 1840
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.96,
|
| 1134 |
+
"learning_rate": 5.2236421725239616e-05,
|
| 1135 |
+
"loss": 0.0827,
|
| 1136 |
+
"step": 1850
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 2.97,
|
| 1140 |
+
"learning_rate": 5.1437699680511184e-05,
|
| 1141 |
+
"loss": 0.2688,
|
| 1142 |
+
"step": 1860
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.99,
|
| 1146 |
+
"learning_rate": 5.0638977635782745e-05,
|
| 1147 |
+
"loss": 0.3182,
|
| 1148 |
+
"step": 1870
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 3.0,
|
| 1152 |
+
"eval_accuracy": 0.8652849740932642,
|
| 1153 |
+
"eval_loss": 0.3915177285671234,
|
| 1154 |
+
"eval_runtime": 1.2215,
|
| 1155 |
+
"eval_samples_per_second": 158.006,
|
| 1156 |
+
"eval_steps_per_second": 10.643,
|
| 1157 |
+
"step": 1878
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 3.0,
|
| 1161 |
+
"learning_rate": 4.984025559105431e-05,
|
| 1162 |
+
"loss": 0.1101,
|
| 1163 |
+
"step": 1880
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 3.02,
|
| 1167 |
+
"learning_rate": 4.904153354632588e-05,
|
| 1168 |
+
"loss": 0.08,
|
| 1169 |
+
"step": 1890
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 3.04,
|
| 1173 |
+
"learning_rate": 4.824281150159744e-05,
|
| 1174 |
+
"loss": 0.0661,
|
| 1175 |
+
"step": 1900
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 3.05,
|
| 1179 |
+
"learning_rate": 4.744408945686901e-05,
|
| 1180 |
+
"loss": 0.088,
|
| 1181 |
+
"step": 1910
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 3.07,
|
| 1185 |
+
"learning_rate": 4.664536741214058e-05,
|
| 1186 |
+
"loss": 0.069,
|
| 1187 |
+
"step": 1920
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 3.08,
|
| 1191 |
+
"learning_rate": 4.584664536741215e-05,
|
| 1192 |
+
"loss": 0.0646,
|
| 1193 |
+
"step": 1930
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 3.1,
|
| 1197 |
+
"learning_rate": 4.504792332268371e-05,
|
| 1198 |
+
"loss": 0.0334,
|
| 1199 |
+
"step": 1940
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 3.12,
|
| 1203 |
+
"learning_rate": 4.424920127795527e-05,
|
| 1204 |
+
"loss": 0.0727,
|
| 1205 |
+
"step": 1950
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 3.13,
|
| 1209 |
+
"learning_rate": 4.345047923322684e-05,
|
| 1210 |
+
"loss": 0.0762,
|
| 1211 |
+
"step": 1960
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 3.15,
|
| 1215 |
+
"learning_rate": 4.265175718849841e-05,
|
| 1216 |
+
"loss": 0.0199,
|
| 1217 |
+
"step": 1970
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 3.16,
|
| 1221 |
+
"learning_rate": 4.185303514376997e-05,
|
| 1222 |
+
"loss": 0.0673,
|
| 1223 |
+
"step": 1980
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 3.18,
|
| 1227 |
+
"learning_rate": 4.1054313099041536e-05,
|
| 1228 |
+
"loss": 0.0639,
|
| 1229 |
+
"step": 1990
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 3.19,
|
| 1233 |
+
"learning_rate": 4.0255591054313104e-05,
|
| 1234 |
+
"loss": 0.068,
|
| 1235 |
+
"step": 2000
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 3.21,
|
| 1239 |
+
"learning_rate": 3.9456869009584666e-05,
|
| 1240 |
+
"loss": 0.034,
|
| 1241 |
+
"step": 2010
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 3.23,
|
| 1245 |
+
"learning_rate": 3.8658146964856234e-05,
|
| 1246 |
+
"loss": 0.0384,
|
| 1247 |
+
"step": 2020
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 3.24,
|
| 1251 |
+
"learning_rate": 3.7859424920127795e-05,
|
| 1252 |
+
"loss": 0.0797,
|
| 1253 |
+
"step": 2030
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 3.26,
|
| 1257 |
+
"learning_rate": 3.7060702875399364e-05,
|
| 1258 |
+
"loss": 0.0842,
|
| 1259 |
+
"step": 2040
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 3.27,
|
| 1263 |
+
"learning_rate": 3.6261980830670925e-05,
|
| 1264 |
+
"loss": 0.0483,
|
| 1265 |
+
"step": 2050
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 3.29,
|
| 1269 |
+
"learning_rate": 3.546325878594249e-05,
|
| 1270 |
+
"loss": 0.0688,
|
| 1271 |
+
"step": 2060
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 3.31,
|
| 1275 |
+
"learning_rate": 3.466453674121406e-05,
|
| 1276 |
+
"loss": 0.0651,
|
| 1277 |
+
"step": 2070
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.32,
|
| 1281 |
+
"learning_rate": 3.386581469648562e-05,
|
| 1282 |
+
"loss": 0.0215,
|
| 1283 |
+
"step": 2080
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 3.34,
|
| 1287 |
+
"learning_rate": 3.306709265175719e-05,
|
| 1288 |
+
"loss": 0.0203,
|
| 1289 |
+
"step": 2090
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 3.35,
|
| 1293 |
+
"learning_rate": 3.226837060702875e-05,
|
| 1294 |
+
"loss": 0.035,
|
| 1295 |
+
"step": 2100
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 3.37,
|
| 1299 |
+
"learning_rate": 3.146964856230032e-05,
|
| 1300 |
+
"loss": 0.1068,
|
| 1301 |
+
"step": 2110
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 3.39,
|
| 1305 |
+
"learning_rate": 3.067092651757188e-05,
|
| 1306 |
+
"loss": 0.0426,
|
| 1307 |
+
"step": 2120
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 3.4,
|
| 1311 |
+
"learning_rate": 2.987220447284345e-05,
|
| 1312 |
+
"loss": 0.0234,
|
| 1313 |
+
"step": 2130
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 3.42,
|
| 1317 |
+
"learning_rate": 2.907348242811502e-05,
|
| 1318 |
+
"loss": 0.0345,
|
| 1319 |
+
"step": 2140
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 3.43,
|
| 1323 |
+
"learning_rate": 2.8274760383386583e-05,
|
| 1324 |
+
"loss": 0.1023,
|
| 1325 |
+
"step": 2150
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 3.45,
|
| 1329 |
+
"learning_rate": 2.747603833865815e-05,
|
| 1330 |
+
"loss": 0.0391,
|
| 1331 |
+
"step": 2160
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 3.47,
|
| 1335 |
+
"learning_rate": 2.6677316293929716e-05,
|
| 1336 |
+
"loss": 0.0536,
|
| 1337 |
+
"step": 2170
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 3.48,
|
| 1341 |
+
"learning_rate": 2.5878594249201278e-05,
|
| 1342 |
+
"loss": 0.0318,
|
| 1343 |
+
"step": 2180
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 3.5,
|
| 1347 |
+
"learning_rate": 2.5079872204472842e-05,
|
| 1348 |
+
"loss": 0.0285,
|
| 1349 |
+
"step": 2190
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 3.51,
|
| 1353 |
+
"learning_rate": 2.428115015974441e-05,
|
| 1354 |
+
"loss": 0.062,
|
| 1355 |
+
"step": 2200
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 3.53,
|
| 1359 |
+
"learning_rate": 2.3482428115015975e-05,
|
| 1360 |
+
"loss": 0.0315,
|
| 1361 |
+
"step": 2210
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.55,
|
| 1365 |
+
"learning_rate": 2.268370607028754e-05,
|
| 1366 |
+
"loss": 0.0656,
|
| 1367 |
+
"step": 2220
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 3.56,
|
| 1371 |
+
"learning_rate": 2.188498402555911e-05,
|
| 1372 |
+
"loss": 0.0152,
|
| 1373 |
+
"step": 2230
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 3.58,
|
| 1377 |
+
"learning_rate": 2.108626198083067e-05,
|
| 1378 |
+
"loss": 0.1108,
|
| 1379 |
+
"step": 2240
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 3.59,
|
| 1383 |
+
"learning_rate": 2.0287539936102238e-05,
|
| 1384 |
+
"loss": 0.0951,
|
| 1385 |
+
"step": 2250
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 3.61,
|
| 1389 |
+
"learning_rate": 1.9488817891373803e-05,
|
| 1390 |
+
"loss": 0.1139,
|
| 1391 |
+
"step": 2260
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 3.63,
|
| 1395 |
+
"learning_rate": 1.869009584664537e-05,
|
| 1396 |
+
"loss": 0.027,
|
| 1397 |
+
"step": 2270
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 3.64,
|
| 1401 |
+
"learning_rate": 1.7891373801916932e-05,
|
| 1402 |
+
"loss": 0.0577,
|
| 1403 |
+
"step": 2280
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.66,
|
| 1407 |
+
"learning_rate": 1.7172523961661345e-05,
|
| 1408 |
+
"loss": 0.0411,
|
| 1409 |
+
"step": 2290
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 3.67,
|
| 1413 |
+
"learning_rate": 1.6373801916932906e-05,
|
| 1414 |
+
"loss": 0.0706,
|
| 1415 |
+
"step": 2300
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 3.69,
|
| 1419 |
+
"learning_rate": 1.5575079872204475e-05,
|
| 1420 |
+
"loss": 0.1357,
|
| 1421 |
+
"step": 2310
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 3.71,
|
| 1425 |
+
"learning_rate": 1.477635782747604e-05,
|
| 1426 |
+
"loss": 0.0638,
|
| 1427 |
+
"step": 2320
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 3.72,
|
| 1431 |
+
"learning_rate": 1.3977635782747606e-05,
|
| 1432 |
+
"loss": 0.0508,
|
| 1433 |
+
"step": 2330
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 3.74,
|
| 1437 |
+
"learning_rate": 1.3178913738019169e-05,
|
| 1438 |
+
"loss": 0.0224,
|
| 1439 |
+
"step": 2340
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 3.75,
|
| 1443 |
+
"learning_rate": 1.2380191693290735e-05,
|
| 1444 |
+
"loss": 0.0108,
|
| 1445 |
+
"step": 2350
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.77,
|
| 1449 |
+
"learning_rate": 1.1581469648562302e-05,
|
| 1450 |
+
"loss": 0.0652,
|
| 1451 |
+
"step": 2360
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 3.79,
|
| 1455 |
+
"learning_rate": 1.0782747603833867e-05,
|
| 1456 |
+
"loss": 0.0548,
|
| 1457 |
+
"step": 2370
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 3.8,
|
| 1461 |
+
"learning_rate": 9.984025559105432e-06,
|
| 1462 |
+
"loss": 0.0335,
|
| 1463 |
+
"step": 2380
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 3.82,
|
| 1467 |
+
"learning_rate": 9.185303514376996e-06,
|
| 1468 |
+
"loss": 0.0372,
|
| 1469 |
+
"step": 2390
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 3.83,
|
| 1473 |
+
"learning_rate": 8.386581469648563e-06,
|
| 1474 |
+
"loss": 0.0994,
|
| 1475 |
+
"step": 2400
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 3.85,
|
| 1479 |
+
"learning_rate": 7.5878594249201285e-06,
|
| 1480 |
+
"loss": 0.0135,
|
| 1481 |
+
"step": 2410
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 3.87,
|
| 1485 |
+
"learning_rate": 6.789137380191693e-06,
|
| 1486 |
+
"loss": 0.0583,
|
| 1487 |
+
"step": 2420
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.88,
|
| 1491 |
+
"learning_rate": 5.990415335463259e-06,
|
| 1492 |
+
"loss": 0.0961,
|
| 1493 |
+
"step": 2430
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 3.9,
|
| 1497 |
+
"learning_rate": 5.191693290734825e-06,
|
| 1498 |
+
"loss": 0.0292,
|
| 1499 |
+
"step": 2440
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 3.91,
|
| 1503 |
+
"learning_rate": 4.39297124600639e-06,
|
| 1504 |
+
"loss": 0.0288,
|
| 1505 |
+
"step": 2450
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 3.93,
|
| 1509 |
+
"learning_rate": 3.5942492012779555e-06,
|
| 1510 |
+
"loss": 0.0565,
|
| 1511 |
+
"step": 2460
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 3.95,
|
| 1515 |
+
"learning_rate": 2.7955271565495207e-06,
|
| 1516 |
+
"loss": 0.0607,
|
| 1517 |
+
"step": 2470
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 3.96,
|
| 1521 |
+
"learning_rate": 1.9968051118210863e-06,
|
| 1522 |
+
"loss": 0.0529,
|
| 1523 |
+
"step": 2480
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 3.98,
|
| 1527 |
+
"learning_rate": 1.1980830670926517e-06,
|
| 1528 |
+
"loss": 0.0449,
|
| 1529 |
+
"step": 2490
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 3.99,
|
| 1533 |
+
"learning_rate": 3.9936102236421723e-07,
|
| 1534 |
+
"loss": 0.067,
|
| 1535 |
+
"step": 2500
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 4.0,
|
| 1539 |
+
"eval_accuracy": 0.8704663212435233,
|
| 1540 |
+
"eval_loss": 0.4523600935935974,
|
| 1541 |
+
"eval_runtime": 1.2991,
|
| 1542 |
+
"eval_samples_per_second": 148.567,
|
| 1543 |
+
"eval_steps_per_second": 10.007,
|
| 1544 |
+
"step": 2504
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 4.0,
|
| 1548 |
+
"step": 2504,
|
| 1549 |
+
"total_flos": 3.104468219559813e+18,
|
| 1550 |
+
"train_loss": 0.3521563596071336,
|
| 1551 |
+
"train_runtime": 855.1502,
|
| 1552 |
+
"train_samples_per_second": 46.846,
|
| 1553 |
+
"train_steps_per_second": 2.928
|
| 1554 |
+
}
|
| 1555 |
+
],
|
| 1556 |
+
"max_steps": 2504,
|
| 1557 |
+
"num_train_epochs": 4,
|
| 1558 |
+
"total_flos": 3.104468219559813e+18,
|
| 1559 |
+
"trial_name": null,
|
| 1560 |
+
"trial_params": null
|
| 1561 |
+
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3899
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b9327532a3b8055a3cc3c9f84ab6a6bea92d89b653599fe55ef75bfa90283be
|
| 3 |
size 3899
|