Upload folder using huggingface_hub
Browse files- resnet50/checkpoint-2000/model.safetensors +3 -0
- resnet50/checkpoint-2000/optimizer.pt +3 -0
- resnet50/checkpoint-2000/rng_state.pth +3 -0
- resnet50/checkpoint-2000/scaler.pt +3 -0
- resnet50/checkpoint-2000/scheduler.pt +3 -0
- resnet50/checkpoint-2000/trainer_state.json +1470 -0
- resnet50/checkpoint-2000/training_args.bin +3 -0
resnet50/checkpoint-2000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e3f7b532a57f21c97627fdb3e3291a2bd4a66d6b027d9aa32a4ed8ffdb946321
|
| 3 |
+
size 97505840
|
resnet50/checkpoint-2000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d63ceddbd203e7fb0ed077c811815589c7914673053b05abedb305ce016e97c8
|
| 3 |
+
size 194658106
|
resnet50/checkpoint-2000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ecefbb3f17bb76b6655eb0157c98b5287c17fa4b4c72a6b9068b0823ce9fd18d
|
| 3 |
+
size 14244
|
resnet50/checkpoint-2000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:22c1484d2a97be7e296b43912794fafa4f457a68e8b18b070e09de5f428749f7
|
| 3 |
+
size 988
|
resnet50/checkpoint-2000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:57e0cd6e9aa45ad663e07fd86f016f49e40d06d6569fdff4a5dafc3b2c15a017
|
| 3 |
+
size 1064
|
resnet50/checkpoint-2000/trainer_state.json
ADDED
|
@@ -0,0 +1,1470 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 2000,
|
| 3 |
+
"best_metric": 0.6596935206968907,
|
| 4 |
+
"best_model_checkpoint": "/workspace/output/resnet50/checkpoint-2000",
|
| 5 |
+
"epoch": 0.28388928317955997,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0014194464158978,
|
| 14 |
+
"grad_norm": 3.2342276573181152,
|
| 15 |
+
"learning_rate": 9.999872249822569e-05,
|
| 16 |
+
"loss": 5.98863525390625,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0028388928317956,
|
| 21 |
+
"grad_norm": 3.3994972705841064,
|
| 22 |
+
"learning_rate": 9.99973030518098e-05,
|
| 23 |
+
"loss": 5.97633056640625,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0042583392476933995,
|
| 28 |
+
"grad_norm": 3.3180341720581055,
|
| 29 |
+
"learning_rate": 9.99958836053939e-05,
|
| 30 |
+
"loss": 5.97711181640625,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.0056777856635912,
|
| 35 |
+
"grad_norm": 2.9379143714904785,
|
| 36 |
+
"learning_rate": 9.999446415897801e-05,
|
| 37 |
+
"loss": 5.9991455078125,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.007097232079488999,
|
| 42 |
+
"grad_norm": 2.2698018550872803,
|
| 43 |
+
"learning_rate": 9.99930447125621e-05,
|
| 44 |
+
"loss": 5.96363525390625,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.008516678495386799,
|
| 49 |
+
"grad_norm": 2.0626659393310547,
|
| 50 |
+
"learning_rate": 9.99916252661462e-05,
|
| 51 |
+
"loss": 5.96995849609375,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.0099361249112846,
|
| 56 |
+
"grad_norm": 2.814460277557373,
|
| 57 |
+
"learning_rate": 9.999020581973031e-05,
|
| 58 |
+
"loss": 5.9493408203125,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.0113555713271824,
|
| 63 |
+
"grad_norm": 2.871051788330078,
|
| 64 |
+
"learning_rate": 9.998878637331441e-05,
|
| 65 |
+
"loss": 5.9510498046875,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0127750177430802,
|
| 70 |
+
"grad_norm": 2.3897151947021484,
|
| 71 |
+
"learning_rate": 9.998736692689852e-05,
|
| 72 |
+
"loss": 5.94254150390625,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.014194464158977998,
|
| 77 |
+
"grad_norm": 2.9910531044006348,
|
| 78 |
+
"learning_rate": 9.99859474804826e-05,
|
| 79 |
+
"loss": 5.9062255859375,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.015613910574875798,
|
| 84 |
+
"grad_norm": 3.137518882751465,
|
| 85 |
+
"learning_rate": 9.998452803406672e-05,
|
| 86 |
+
"loss": 5.9070068359375,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.017033356990773598,
|
| 91 |
+
"grad_norm": 3.021024703979492,
|
| 92 |
+
"learning_rate": 9.998310858765082e-05,
|
| 93 |
+
"loss": 5.87197265625,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.018452803406671398,
|
| 98 |
+
"grad_norm": 3.499450445175171,
|
| 99 |
+
"learning_rate": 9.998168914123493e-05,
|
| 100 |
+
"loss": 5.8237548828125,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.0198722498225692,
|
| 105 |
+
"grad_norm": 3.87576961517334,
|
| 106 |
+
"learning_rate": 9.998026969481902e-05,
|
| 107 |
+
"loss": 5.754150390625,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.021291696238467,
|
| 112 |
+
"grad_norm": 3.9846458435058594,
|
| 113 |
+
"learning_rate": 9.997885024840313e-05,
|
| 114 |
+
"loss": 5.697198486328125,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.0227111426543648,
|
| 119 |
+
"grad_norm": 4.339130878448486,
|
| 120 |
+
"learning_rate": 9.997743080198723e-05,
|
| 121 |
+
"loss": 5.63760986328125,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0241305890702626,
|
| 126 |
+
"grad_norm": 4.891483783721924,
|
| 127 |
+
"learning_rate": 9.997601135557133e-05,
|
| 128 |
+
"loss": 5.5271728515625,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0255500354861604,
|
| 133 |
+
"grad_norm": 5.147222995758057,
|
| 134 |
+
"learning_rate": 9.997459190915544e-05,
|
| 135 |
+
"loss": 5.45938720703125,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.0269694819020582,
|
| 140 |
+
"grad_norm": 5.365755558013916,
|
| 141 |
+
"learning_rate": 9.997317246273954e-05,
|
| 142 |
+
"loss": 5.355255126953125,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.028388928317955996,
|
| 147 |
+
"grad_norm": 5.888001918792725,
|
| 148 |
+
"learning_rate": 9.997175301632365e-05,
|
| 149 |
+
"loss": 5.1554931640625,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.029808374733853796,
|
| 154 |
+
"grad_norm": 6.100172996520996,
|
| 155 |
+
"learning_rate": 9.997033356990773e-05,
|
| 156 |
+
"loss": 5.035284423828125,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.031227821149751596,
|
| 161 |
+
"grad_norm": 6.491486549377441,
|
| 162 |
+
"learning_rate": 9.996891412349184e-05,
|
| 163 |
+
"loss": 4.899530029296875,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.032647267565649396,
|
| 168 |
+
"grad_norm": 6.916806697845459,
|
| 169 |
+
"learning_rate": 9.996749467707594e-05,
|
| 170 |
+
"loss": 4.851350402832031,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.034066713981547196,
|
| 175 |
+
"grad_norm": 6.837950706481934,
|
| 176 |
+
"learning_rate": 9.996607523066005e-05,
|
| 177 |
+
"loss": 4.726431274414063,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.035486160397444996,
|
| 182 |
+
"grad_norm": 7.554074287414551,
|
| 183 |
+
"learning_rate": 9.996465578424415e-05,
|
| 184 |
+
"loss": 4.4839630126953125,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.036905606813342796,
|
| 189 |
+
"grad_norm": 7.574995994567871,
|
| 190 |
+
"learning_rate": 9.996323633782825e-05,
|
| 191 |
+
"loss": 4.506732177734375,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.0383250532292406,
|
| 196 |
+
"grad_norm": 7.498238563537598,
|
| 197 |
+
"learning_rate": 9.996181689141236e-05,
|
| 198 |
+
"loss": 4.319998168945313,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.0397444996451384,
|
| 203 |
+
"grad_norm": 7.978142261505127,
|
| 204 |
+
"learning_rate": 9.996039744499645e-05,
|
| 205 |
+
"loss": 4.214613342285157,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.0411639460610362,
|
| 210 |
+
"grad_norm": 8.194511413574219,
|
| 211 |
+
"learning_rate": 9.995897799858057e-05,
|
| 212 |
+
"loss": 4.212762451171875,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.042583392476934,
|
| 217 |
+
"grad_norm": 8.136639595031738,
|
| 218 |
+
"learning_rate": 9.995755855216466e-05,
|
| 219 |
+
"loss": 4.009028625488281,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.0440028388928318,
|
| 224 |
+
"grad_norm": 8.684012413024902,
|
| 225 |
+
"learning_rate": 9.995613910574876e-05,
|
| 226 |
+
"loss": 3.9817459106445314,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0454222853087296,
|
| 231 |
+
"grad_norm": 8.888952255249023,
|
| 232 |
+
"learning_rate": 9.995471965933286e-05,
|
| 233 |
+
"loss": 3.94019775390625,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.0468417317246274,
|
| 238 |
+
"grad_norm": 8.79919719696045,
|
| 239 |
+
"learning_rate": 9.995330021291697e-05,
|
| 240 |
+
"loss": 3.9265777587890627,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0482611781405252,
|
| 245 |
+
"grad_norm": 8.571785926818848,
|
| 246 |
+
"learning_rate": 9.995188076650107e-05,
|
| 247 |
+
"loss": 3.7262115478515625,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.049680624556423,
|
| 252 |
+
"grad_norm": 8.640142440795898,
|
| 253 |
+
"learning_rate": 9.995046132008518e-05,
|
| 254 |
+
"loss": 3.644915771484375,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.0511000709723208,
|
| 259 |
+
"grad_norm": 9.322779655456543,
|
| 260 |
+
"learning_rate": 9.994904187366927e-05,
|
| 261 |
+
"loss": 3.644049072265625,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.0525195173882186,
|
| 266 |
+
"grad_norm": 8.790424346923828,
|
| 267 |
+
"learning_rate": 9.994762242725337e-05,
|
| 268 |
+
"loss": 3.4869285583496095,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.0539389638041164,
|
| 273 |
+
"grad_norm": 9.344154357910156,
|
| 274 |
+
"learning_rate": 9.994620298083748e-05,
|
| 275 |
+
"loss": 3.55142822265625,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.05535841022001419,
|
| 280 |
+
"grad_norm": 8.807840347290039,
|
| 281 |
+
"learning_rate": 9.994478353442158e-05,
|
| 282 |
+
"loss": 3.4293190002441407,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.05677785663591199,
|
| 287 |
+
"grad_norm": 9.36971378326416,
|
| 288 |
+
"learning_rate": 9.994336408800569e-05,
|
| 289 |
+
"loss": 3.429082489013672,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.05819730305180979,
|
| 294 |
+
"grad_norm": 9.73521900177002,
|
| 295 |
+
"learning_rate": 9.994194464158977e-05,
|
| 296 |
+
"loss": 3.408639907836914,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.05961674946770759,
|
| 301 |
+
"grad_norm": 9.646844863891602,
|
| 302 |
+
"learning_rate": 9.994052519517389e-05,
|
| 303 |
+
"loss": 3.1950119018554686,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.06103619588360539,
|
| 308 |
+
"grad_norm": 9.722207069396973,
|
| 309 |
+
"learning_rate": 9.993910574875798e-05,
|
| 310 |
+
"loss": 3.4140243530273438,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.06245564229950319,
|
| 315 |
+
"grad_norm": 10.609601020812988,
|
| 316 |
+
"learning_rate": 9.99376863023421e-05,
|
| 317 |
+
"loss": 3.320109558105469,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.063875088715401,
|
| 322 |
+
"grad_norm": 10.271575927734375,
|
| 323 |
+
"learning_rate": 9.993626685592619e-05,
|
| 324 |
+
"loss": 3.232251739501953,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.06529453513129879,
|
| 329 |
+
"grad_norm": 9.766585350036621,
|
| 330 |
+
"learning_rate": 9.993484740951029e-05,
|
| 331 |
+
"loss": 3.149517059326172,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.0667139815471966,
|
| 336 |
+
"grad_norm": 10.358244895935059,
|
| 337 |
+
"learning_rate": 9.99334279630944e-05,
|
| 338 |
+
"loss": 3.1863967895507814,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.06813342796309439,
|
| 343 |
+
"grad_norm": 10.473136901855469,
|
| 344 |
+
"learning_rate": 9.99320085166785e-05,
|
| 345 |
+
"loss": 3.222390365600586,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.0695528743789922,
|
| 350 |
+
"grad_norm": 9.905110359191895,
|
| 351 |
+
"learning_rate": 9.993058907026261e-05,
|
| 352 |
+
"loss": 3.1823768615722656,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.07097232079488999,
|
| 357 |
+
"grad_norm": 9.858973503112793,
|
| 358 |
+
"learning_rate": 9.99291696238467e-05,
|
| 359 |
+
"loss": 2.9202560424804687,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.07097232079488999,
|
| 364 |
+
"eval_accuracy": 0.1867489031601704,
|
| 365 |
+
"eval_loss": 3.0744524002075195,
|
| 366 |
+
"eval_runtime": 31.2289,
|
| 367 |
+
"eval_samples_per_second": 503.605,
|
| 368 |
+
"eval_steps_per_second": 15.755,
|
| 369 |
+
"step": 500
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.0723917672107878,
|
| 373 |
+
"grad_norm": 10.224215507507324,
|
| 374 |
+
"learning_rate": 9.992775017743082e-05,
|
| 375 |
+
"loss": 3.0410499572753906,
|
| 376 |
+
"step": 510
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.07381121362668559,
|
| 380 |
+
"grad_norm": 9.867650032043457,
|
| 381 |
+
"learning_rate": 9.99263307310149e-05,
|
| 382 |
+
"loss": 3.116912078857422,
|
| 383 |
+
"step": 520
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 0.07523066004258339,
|
| 387 |
+
"grad_norm": 10.343064308166504,
|
| 388 |
+
"learning_rate": 9.992491128459901e-05,
|
| 389 |
+
"loss": 3.06390266418457,
|
| 390 |
+
"step": 530
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 0.0766501064584812,
|
| 394 |
+
"grad_norm": 10.38116455078125,
|
| 395 |
+
"learning_rate": 9.992349183818311e-05,
|
| 396 |
+
"loss": 2.973680114746094,
|
| 397 |
+
"step": 540
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 0.07806955287437899,
|
| 401 |
+
"grad_norm": 10.979643821716309,
|
| 402 |
+
"learning_rate": 9.992207239176722e-05,
|
| 403 |
+
"loss": 3.0906436920166014,
|
| 404 |
+
"step": 550
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.0794889992902768,
|
| 408 |
+
"grad_norm": 10.06657886505127,
|
| 409 |
+
"learning_rate": 9.992065294535132e-05,
|
| 410 |
+
"loss": 3.0091484069824217,
|
| 411 |
+
"step": 560
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.08090844570617459,
|
| 415 |
+
"grad_norm": 10.663322448730469,
|
| 416 |
+
"learning_rate": 9.991923349893541e-05,
|
| 417 |
+
"loss": 2.862255859375,
|
| 418 |
+
"step": 570
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.0823278921220724,
|
| 422 |
+
"grad_norm": 9.277785301208496,
|
| 423 |
+
"learning_rate": 9.991781405251952e-05,
|
| 424 |
+
"loss": 2.8638259887695314,
|
| 425 |
+
"step": 580
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.08374733853797019,
|
| 429 |
+
"grad_norm": 10.807332038879395,
|
| 430 |
+
"learning_rate": 9.991639460610362e-05,
|
| 431 |
+
"loss": 2.732352066040039,
|
| 432 |
+
"step": 590
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.085166784953868,
|
| 436 |
+
"grad_norm": 9.970373153686523,
|
| 437 |
+
"learning_rate": 9.991497515968773e-05,
|
| 438 |
+
"loss": 2.736968231201172,
|
| 439 |
+
"step": 600
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.08658623136976579,
|
| 443 |
+
"grad_norm": 11.008269309997559,
|
| 444 |
+
"learning_rate": 9.991355571327183e-05,
|
| 445 |
+
"loss": 2.7735246658325194,
|
| 446 |
+
"step": 610
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.0880056777856636,
|
| 450 |
+
"grad_norm": 8.758193969726562,
|
| 451 |
+
"learning_rate": 9.991213626685593e-05,
|
| 452 |
+
"loss": 2.5436214447021483,
|
| 453 |
+
"step": 620
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.08942512420156139,
|
| 457 |
+
"grad_norm": 11.253259658813477,
|
| 458 |
+
"learning_rate": 9.991071682044003e-05,
|
| 459 |
+
"loss": 2.748835563659668,
|
| 460 |
+
"step": 630
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.0908445706174592,
|
| 464 |
+
"grad_norm": 10.979547500610352,
|
| 465 |
+
"learning_rate": 9.990929737402414e-05,
|
| 466 |
+
"loss": 2.7314834594726562,
|
| 467 |
+
"step": 640
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.09226401703335699,
|
| 471 |
+
"grad_norm": 11.182887077331543,
|
| 472 |
+
"learning_rate": 9.990787792760823e-05,
|
| 473 |
+
"loss": 2.645678901672363,
|
| 474 |
+
"step": 650
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.0936834634492548,
|
| 478 |
+
"grad_norm": 10.636208534240723,
|
| 479 |
+
"learning_rate": 9.990645848119234e-05,
|
| 480 |
+
"loss": 2.5704013824462892,
|
| 481 |
+
"step": 660
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 0.09510290986515259,
|
| 485 |
+
"grad_norm": 10.351170539855957,
|
| 486 |
+
"learning_rate": 9.990503903477644e-05,
|
| 487 |
+
"loss": 2.5628406524658205,
|
| 488 |
+
"step": 670
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.0965223562810504,
|
| 492 |
+
"grad_norm": 9.914809226989746,
|
| 493 |
+
"learning_rate": 9.990361958836054e-05,
|
| 494 |
+
"loss": 2.5872230529785156,
|
| 495 |
+
"step": 680
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.09794180269694819,
|
| 499 |
+
"grad_norm": 10.839837074279785,
|
| 500 |
+
"learning_rate": 9.990220014194465e-05,
|
| 501 |
+
"loss": 2.490940475463867,
|
| 502 |
+
"step": 690
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 0.099361249112846,
|
| 506 |
+
"grad_norm": 11.259613990783691,
|
| 507 |
+
"learning_rate": 9.990078069552875e-05,
|
| 508 |
+
"loss": 2.64483585357666,
|
| 509 |
+
"step": 700
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.10078069552874379,
|
| 513 |
+
"grad_norm": 11.213078498840332,
|
| 514 |
+
"learning_rate": 9.989936124911286e-05,
|
| 515 |
+
"loss": 2.5397150039672853,
|
| 516 |
+
"step": 710
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.1022001419446416,
|
| 520 |
+
"grad_norm": 10.366206169128418,
|
| 521 |
+
"learning_rate": 9.989794180269694e-05,
|
| 522 |
+
"loss": 2.457781219482422,
|
| 523 |
+
"step": 720
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 0.10361958836053939,
|
| 527 |
+
"grad_norm": 11.44458293914795,
|
| 528 |
+
"learning_rate": 9.989652235628105e-05,
|
| 529 |
+
"loss": 2.5090484619140625,
|
| 530 |
+
"step": 730
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.1050390347764372,
|
| 534 |
+
"grad_norm": 11.689805030822754,
|
| 535 |
+
"learning_rate": 9.989510290986515e-05,
|
| 536 |
+
"loss": 2.409171485900879,
|
| 537 |
+
"step": 740
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.10645848119233499,
|
| 541 |
+
"grad_norm": 10.568279266357422,
|
| 542 |
+
"learning_rate": 9.989368346344926e-05,
|
| 543 |
+
"loss": 2.3308380126953123,
|
| 544 |
+
"step": 750
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 0.1078779276082328,
|
| 548 |
+
"grad_norm": 11.917696952819824,
|
| 549 |
+
"learning_rate": 9.989226401703337e-05,
|
| 550 |
+
"loss": 2.3733493804931642,
|
| 551 |
+
"step": 760
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.10929737402413059,
|
| 555 |
+
"grad_norm": 9.960722923278809,
|
| 556 |
+
"learning_rate": 9.989098651525906e-05,
|
| 557 |
+
"loss": 2.4058095932006838,
|
| 558 |
+
"step": 770
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 0.11071682044002838,
|
| 562 |
+
"grad_norm": 11.068999290466309,
|
| 563 |
+
"learning_rate": 9.988956706884315e-05,
|
| 564 |
+
"loss": 2.4371658325195313,
|
| 565 |
+
"step": 780
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.11213626685592619,
|
| 569 |
+
"grad_norm": 10.340009689331055,
|
| 570 |
+
"learning_rate": 9.988814762242725e-05,
|
| 571 |
+
"loss": 2.2587520599365236,
|
| 572 |
+
"step": 790
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.11355571327182398,
|
| 576 |
+
"grad_norm": 9.941303253173828,
|
| 577 |
+
"learning_rate": 9.988672817601136e-05,
|
| 578 |
+
"loss": 2.268446350097656,
|
| 579 |
+
"step": 800
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.11497515968772179,
|
| 583 |
+
"grad_norm": 11.490272521972656,
|
| 584 |
+
"learning_rate": 9.988530872959546e-05,
|
| 585 |
+
"loss": 2.471067428588867,
|
| 586 |
+
"step": 810
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.11639460610361958,
|
| 590 |
+
"grad_norm": 10.67241382598877,
|
| 591 |
+
"learning_rate": 9.988388928317957e-05,
|
| 592 |
+
"loss": 2.3497791290283203,
|
| 593 |
+
"step": 820
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.11781405251951739,
|
| 597 |
+
"grad_norm": 10.710894584655762,
|
| 598 |
+
"learning_rate": 9.988246983676367e-05,
|
| 599 |
+
"loss": 2.1724626541137697,
|
| 600 |
+
"step": 830
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.11923349893541518,
|
| 604 |
+
"grad_norm": 10.985452651977539,
|
| 605 |
+
"learning_rate": 9.988105039034778e-05,
|
| 606 |
+
"loss": 2.1848114013671873,
|
| 607 |
+
"step": 840
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.12065294535131299,
|
| 611 |
+
"grad_norm": 10.063145637512207,
|
| 612 |
+
"learning_rate": 9.987963094393186e-05,
|
| 613 |
+
"loss": 2.180558776855469,
|
| 614 |
+
"step": 850
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.12207239176721078,
|
| 618 |
+
"grad_norm": 11.236614227294922,
|
| 619 |
+
"learning_rate": 9.987821149751597e-05,
|
| 620 |
+
"loss": 2.282668876647949,
|
| 621 |
+
"step": 860
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.12349183818310859,
|
| 625 |
+
"grad_norm": 10.98898983001709,
|
| 626 |
+
"learning_rate": 9.987679205110007e-05,
|
| 627 |
+
"loss": 2.235186767578125,
|
| 628 |
+
"step": 870
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 0.12491128459900638,
|
| 632 |
+
"grad_norm": 11.805492401123047,
|
| 633 |
+
"learning_rate": 9.987537260468418e-05,
|
| 634 |
+
"loss": 2.2264921188354494,
|
| 635 |
+
"step": 880
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.1263307310149042,
|
| 639 |
+
"grad_norm": 10.717041015625,
|
| 640 |
+
"learning_rate": 9.987395315826828e-05,
|
| 641 |
+
"loss": 2.1385255813598634,
|
| 642 |
+
"step": 890
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.127750177430802,
|
| 646 |
+
"grad_norm": 9.613192558288574,
|
| 647 |
+
"learning_rate": 9.987253371185238e-05,
|
| 648 |
+
"loss": 2.1964336395263673,
|
| 649 |
+
"step": 900
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 0.12916962384669978,
|
| 653 |
+
"grad_norm": 10.594833374023438,
|
| 654 |
+
"learning_rate": 9.987111426543649e-05,
|
| 655 |
+
"loss": 2.050688362121582,
|
| 656 |
+
"step": 910
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.13058907026259758,
|
| 660 |
+
"grad_norm": 11.596671104431152,
|
| 661 |
+
"learning_rate": 9.986969481902059e-05,
|
| 662 |
+
"loss": 2.077385139465332,
|
| 663 |
+
"step": 920
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 0.1320085166784954,
|
| 667 |
+
"grad_norm": 10.779032707214355,
|
| 668 |
+
"learning_rate": 9.98682753726047e-05,
|
| 669 |
+
"loss": 2.0280479431152343,
|
| 670 |
+
"step": 930
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.1334279630943932,
|
| 674 |
+
"grad_norm": 10.522924423217773,
|
| 675 |
+
"learning_rate": 9.98668559261888e-05,
|
| 676 |
+
"loss": 1.9384689331054688,
|
| 677 |
+
"step": 940
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.13484740951029098,
|
| 681 |
+
"grad_norm": 9.86844539642334,
|
| 682 |
+
"learning_rate": 9.986543647977289e-05,
|
| 683 |
+
"loss": 2.0612548828125,
|
| 684 |
+
"step": 950
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.13626685592618878,
|
| 688 |
+
"grad_norm": 12.521405220031738,
|
| 689 |
+
"learning_rate": 9.986401703335699e-05,
|
| 690 |
+
"loss": 2.139466094970703,
|
| 691 |
+
"step": 960
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.1376863023420866,
|
| 695 |
+
"grad_norm": 11.292656898498535,
|
| 696 |
+
"learning_rate": 9.98625975869411e-05,
|
| 697 |
+
"loss": 2.077956199645996,
|
| 698 |
+
"step": 970
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.1391057487579844,
|
| 702 |
+
"grad_norm": 11.186986923217773,
|
| 703 |
+
"learning_rate": 9.98611781405252e-05,
|
| 704 |
+
"loss": 2.028730010986328,
|
| 705 |
+
"step": 980
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.14052519517388218,
|
| 709 |
+
"grad_norm": 10.553022384643555,
|
| 710 |
+
"learning_rate": 9.985975869410931e-05,
|
| 711 |
+
"loss": 1.9375551223754883,
|
| 712 |
+
"step": 990
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.14194464158977999,
|
| 716 |
+
"grad_norm": 11.089204788208008,
|
| 717 |
+
"learning_rate": 9.98583392476934e-05,
|
| 718 |
+
"loss": 2.0689823150634767,
|
| 719 |
+
"step": 1000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 0.14194464158977999,
|
| 723 |
+
"eval_accuracy": 0.42239460799898265,
|
| 724 |
+
"eval_loss": 1.9010688066482544,
|
| 725 |
+
"eval_runtime": 31.4593,
|
| 726 |
+
"eval_samples_per_second": 499.916,
|
| 727 |
+
"eval_steps_per_second": 15.639,
|
| 728 |
+
"step": 1000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.1433640880056778,
|
| 732 |
+
"grad_norm": 10.988676071166992,
|
| 733 |
+
"learning_rate": 9.98569198012775e-05,
|
| 734 |
+
"loss": 1.9830604553222657,
|
| 735 |
+
"step": 1010
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 0.1447835344215756,
|
| 739 |
+
"grad_norm": 11.2459077835083,
|
| 740 |
+
"learning_rate": 9.985550035486161e-05,
|
| 741 |
+
"loss": 1.9190074920654296,
|
| 742 |
+
"step": 1020
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 0.14620298083747338,
|
| 746 |
+
"grad_norm": 10.437894821166992,
|
| 747 |
+
"learning_rate": 9.985408090844571e-05,
|
| 748 |
+
"loss": 1.8999460220336915,
|
| 749 |
+
"step": 1030
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 0.14762242725337119,
|
| 753 |
+
"grad_norm": 10.94793701171875,
|
| 754 |
+
"learning_rate": 9.985266146202982e-05,
|
| 755 |
+
"loss": 1.8579456329345703,
|
| 756 |
+
"step": 1040
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.149041873669269,
|
| 760 |
+
"grad_norm": 11.168233871459961,
|
| 761 |
+
"learning_rate": 9.98512420156139e-05,
|
| 762 |
+
"loss": 1.8979732513427734,
|
| 763 |
+
"step": 1050
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.15046132008516677,
|
| 767 |
+
"grad_norm": 10.14195728302002,
|
| 768 |
+
"learning_rate": 9.984982256919802e-05,
|
| 769 |
+
"loss": 1.7833553314208985,
|
| 770 |
+
"step": 1060
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.15188076650106458,
|
| 774 |
+
"grad_norm": 9.160737991333008,
|
| 775 |
+
"learning_rate": 9.984840312278211e-05,
|
| 776 |
+
"loss": 1.8624576568603515,
|
| 777 |
+
"step": 1070
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.1533002129169624,
|
| 781 |
+
"grad_norm": 11.151049613952637,
|
| 782 |
+
"learning_rate": 9.984698367636623e-05,
|
| 783 |
+
"loss": 1.8210905075073243,
|
| 784 |
+
"step": 1080
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 0.1547196593328602,
|
| 788 |
+
"grad_norm": 10.053725242614746,
|
| 789 |
+
"learning_rate": 9.984556422995032e-05,
|
| 790 |
+
"loss": 1.7738643646240235,
|
| 791 |
+
"step": 1090
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 0.15613910574875797,
|
| 795 |
+
"grad_norm": 10.97727108001709,
|
| 796 |
+
"learning_rate": 9.984414478353442e-05,
|
| 797 |
+
"loss": 1.866429328918457,
|
| 798 |
+
"step": 1100
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 0.15755855216465578,
|
| 802 |
+
"grad_norm": 12.384384155273438,
|
| 803 |
+
"learning_rate": 9.984272533711853e-05,
|
| 804 |
+
"loss": 1.8680984497070312,
|
| 805 |
+
"step": 1110
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 0.1589779985805536,
|
| 809 |
+
"grad_norm": 11.387879371643066,
|
| 810 |
+
"learning_rate": 9.984130589070263e-05,
|
| 811 |
+
"loss": 1.8034194946289062,
|
| 812 |
+
"step": 1120
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.1603974449964514,
|
| 816 |
+
"grad_norm": 10.6587495803833,
|
| 817 |
+
"learning_rate": 9.983988644428674e-05,
|
| 818 |
+
"loss": 1.772690773010254,
|
| 819 |
+
"step": 1130
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.16181689141234917,
|
| 823 |
+
"grad_norm": 12.721858024597168,
|
| 824 |
+
"learning_rate": 9.983846699787084e-05,
|
| 825 |
+
"loss": 1.7724496841430664,
|
| 826 |
+
"step": 1140
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.16323633782824698,
|
| 830 |
+
"grad_norm": 11.116838455200195,
|
| 831 |
+
"learning_rate": 9.983704755145493e-05,
|
| 832 |
+
"loss": 1.7527042388916017,
|
| 833 |
+
"step": 1150
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 0.1646557842441448,
|
| 837 |
+
"grad_norm": 10.033406257629395,
|
| 838 |
+
"learning_rate": 9.983562810503903e-05,
|
| 839 |
+
"loss": 1.674898338317871,
|
| 840 |
+
"step": 1160
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 0.1660752306600426,
|
| 844 |
+
"grad_norm": 11.121773719787598,
|
| 845 |
+
"learning_rate": 9.983420865862314e-05,
|
| 846 |
+
"loss": 1.741505241394043,
|
| 847 |
+
"step": 1170
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 0.16749467707594037,
|
| 851 |
+
"grad_norm": 11.052094459533691,
|
| 852 |
+
"learning_rate": 9.983278921220724e-05,
|
| 853 |
+
"loss": 1.7749841690063477,
|
| 854 |
+
"step": 1180
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.16891412349183818,
|
| 858 |
+
"grad_norm": 10.183452606201172,
|
| 859 |
+
"learning_rate": 9.983136976579135e-05,
|
| 860 |
+
"loss": 1.6881484985351562,
|
| 861 |
+
"step": 1190
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.170333569907736,
|
| 865 |
+
"grad_norm": 11.106999397277832,
|
| 866 |
+
"learning_rate": 9.982995031937545e-05,
|
| 867 |
+
"loss": 1.814961051940918,
|
| 868 |
+
"step": 1200
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 0.1717530163236338,
|
| 872 |
+
"grad_norm": 12.08647632598877,
|
| 873 |
+
"learning_rate": 9.982853087295955e-05,
|
| 874 |
+
"loss": 1.682515525817871,
|
| 875 |
+
"step": 1210
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 0.17317246273953157,
|
| 879 |
+
"grad_norm": 13.744584083557129,
|
| 880 |
+
"learning_rate": 9.982711142654366e-05,
|
| 881 |
+
"loss": 1.6713733673095703,
|
| 882 |
+
"step": 1220
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 0.17459190915542938,
|
| 886 |
+
"grad_norm": 9.970173835754395,
|
| 887 |
+
"learning_rate": 9.982569198012775e-05,
|
| 888 |
+
"loss": 1.711156463623047,
|
| 889 |
+
"step": 1230
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 0.1760113555713272,
|
| 893 |
+
"grad_norm": 11.027495384216309,
|
| 894 |
+
"learning_rate": 9.982427253371186e-05,
|
| 895 |
+
"loss": 1.759619140625,
|
| 896 |
+
"step": 1240
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.177430801987225,
|
| 900 |
+
"grad_norm": 10.876315116882324,
|
| 901 |
+
"learning_rate": 9.982285308729596e-05,
|
| 902 |
+
"loss": 1.618482780456543,
|
| 903 |
+
"step": 1250
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 0.17885024840312277,
|
| 907 |
+
"grad_norm": 10.26490592956543,
|
| 908 |
+
"learning_rate": 9.982143364088006e-05,
|
| 909 |
+
"loss": 1.6674427032470702,
|
| 910 |
+
"step": 1260
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.18026969481902058,
|
| 914 |
+
"grad_norm": 11.872292518615723,
|
| 915 |
+
"learning_rate": 9.982001419446416e-05,
|
| 916 |
+
"loss": 1.6325908660888673,
|
| 917 |
+
"step": 1270
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.1816891412349184,
|
| 921 |
+
"grad_norm": 9.946234703063965,
|
| 922 |
+
"learning_rate": 9.981859474804827e-05,
|
| 923 |
+
"loss": 1.5453743934631348,
|
| 924 |
+
"step": 1280
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.18310858765081617,
|
| 928 |
+
"grad_norm": 11.03128719329834,
|
| 929 |
+
"learning_rate": 9.981717530163236e-05,
|
| 930 |
+
"loss": 1.658684539794922,
|
| 931 |
+
"step": 1290
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 0.18452803406671398,
|
| 935 |
+
"grad_norm": 12.145915031433105,
|
| 936 |
+
"learning_rate": 9.981575585521648e-05,
|
| 937 |
+
"loss": 1.5792274475097656,
|
| 938 |
+
"step": 1300
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.18594748048261178,
|
| 942 |
+
"grad_norm": 11.820379257202148,
|
| 943 |
+
"learning_rate": 9.981433640880057e-05,
|
| 944 |
+
"loss": 1.5301803588867187,
|
| 945 |
+
"step": 1310
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.1873669268985096,
|
| 949 |
+
"grad_norm": 11.046746253967285,
|
| 950 |
+
"learning_rate": 9.981291696238467e-05,
|
| 951 |
+
"loss": 1.6124080657958983,
|
| 952 |
+
"step": 1320
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 0.18878637331440737,
|
| 956 |
+
"grad_norm": 9.545868873596191,
|
| 957 |
+
"learning_rate": 9.981149751596878e-05,
|
| 958 |
+
"loss": 1.5502593994140625,
|
| 959 |
+
"step": 1330
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.19020581973030518,
|
| 963 |
+
"grad_norm": 11.999979019165039,
|
| 964 |
+
"learning_rate": 9.981007806955288e-05,
|
| 965 |
+
"loss": 1.5360203742980958,
|
| 966 |
+
"step": 1340
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 0.19162526614620298,
|
| 970 |
+
"grad_norm": 9.949675559997559,
|
| 971 |
+
"learning_rate": 9.980865862313699e-05,
|
| 972 |
+
"loss": 1.353858470916748,
|
| 973 |
+
"step": 1350
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 0.1930447125621008,
|
| 977 |
+
"grad_norm": 11.573400497436523,
|
| 978 |
+
"learning_rate": 9.980723917672107e-05,
|
| 979 |
+
"loss": 1.3946660995483398,
|
| 980 |
+
"step": 1360
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.19446415897799857,
|
| 984 |
+
"grad_norm": 10.249485969543457,
|
| 985 |
+
"learning_rate": 9.980581973030518e-05,
|
| 986 |
+
"loss": 1.518262004852295,
|
| 987 |
+
"step": 1370
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.19588360539389638,
|
| 991 |
+
"grad_norm": 10.011629104614258,
|
| 992 |
+
"learning_rate": 9.980440028388928e-05,
|
| 993 |
+
"loss": 1.5000194549560546,
|
| 994 |
+
"step": 1380
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.19730305180979418,
|
| 998 |
+
"grad_norm": 12.186440467834473,
|
| 999 |
+
"learning_rate": 9.980298083747339e-05,
|
| 1000 |
+
"loss": 1.554741382598877,
|
| 1001 |
+
"step": 1390
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.198722498225692,
|
| 1005 |
+
"grad_norm": 11.845844268798828,
|
| 1006 |
+
"learning_rate": 9.980156139105749e-05,
|
| 1007 |
+
"loss": 1.4599843978881837,
|
| 1008 |
+
"step": 1400
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 0.20014194464158977,
|
| 1012 |
+
"grad_norm": 10.98592472076416,
|
| 1013 |
+
"learning_rate": 9.980014194464159e-05,
|
| 1014 |
+
"loss": 1.4062080383300781,
|
| 1015 |
+
"step": 1410
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.20156139105748758,
|
| 1019 |
+
"grad_norm": 11.54171371459961,
|
| 1020 |
+
"learning_rate": 9.97987224982257e-05,
|
| 1021 |
+
"loss": 1.5128003120422364,
|
| 1022 |
+
"step": 1420
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.20298083747338538,
|
| 1026 |
+
"grad_norm": 10.248682022094727,
|
| 1027 |
+
"learning_rate": 9.97973030518098e-05,
|
| 1028 |
+
"loss": 1.5022719383239747,
|
| 1029 |
+
"step": 1430
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.2044002838892832,
|
| 1033 |
+
"grad_norm": 8.78536319732666,
|
| 1034 |
+
"learning_rate": 9.97958836053939e-05,
|
| 1035 |
+
"loss": 1.4118841171264649,
|
| 1036 |
+
"step": 1440
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.20581973030518097,
|
| 1040 |
+
"grad_norm": 9.993626594543457,
|
| 1041 |
+
"learning_rate": 9.9794464158978e-05,
|
| 1042 |
+
"loss": 1.3945957183837892,
|
| 1043 |
+
"step": 1450
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 0.20723917672107878,
|
| 1047 |
+
"grad_norm": 11.31412124633789,
|
| 1048 |
+
"learning_rate": 9.97930447125621e-05,
|
| 1049 |
+
"loss": 1.26229887008667,
|
| 1050 |
+
"step": 1460
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.20865862313697658,
|
| 1054 |
+
"grad_norm": 11.182840347290039,
|
| 1055 |
+
"learning_rate": 9.97916252661462e-05,
|
| 1056 |
+
"loss": 1.3171740531921388,
|
| 1057 |
+
"step": 1470
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.2100780695528744,
|
| 1061 |
+
"grad_norm": 12.25224781036377,
|
| 1062 |
+
"learning_rate": 9.979020581973031e-05,
|
| 1063 |
+
"loss": 1.3310781478881837,
|
| 1064 |
+
"step": 1480
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.21149751596877217,
|
| 1068 |
+
"grad_norm": 11.81201457977295,
|
| 1069 |
+
"learning_rate": 9.978878637331441e-05,
|
| 1070 |
+
"loss": 1.3043070793151856,
|
| 1071 |
+
"step": 1490
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.21291696238466998,
|
| 1075 |
+
"grad_norm": 10.484480857849121,
|
| 1076 |
+
"learning_rate": 9.978736692689852e-05,
|
| 1077 |
+
"loss": 1.2629288673400878,
|
| 1078 |
+
"step": 1500
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 0.21291696238466998,
|
| 1082 |
+
"eval_accuracy": 0.5395180263241559,
|
| 1083 |
+
"eval_loss": 1.438815712928772,
|
| 1084 |
+
"eval_runtime": 32.1456,
|
| 1085 |
+
"eval_samples_per_second": 489.242,
|
| 1086 |
+
"eval_steps_per_second": 15.305,
|
| 1087 |
+
"step": 1500
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.21433640880056778,
|
| 1091 |
+
"grad_norm": 10.796157836914062,
|
| 1092 |
+
"learning_rate": 9.978594748048262e-05,
|
| 1093 |
+
"loss": 1.3752121925354004,
|
| 1094 |
+
"step": 1510
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.2157558552164656,
|
| 1098 |
+
"grad_norm": 10.1256742477417,
|
| 1099 |
+
"learning_rate": 9.978452803406671e-05,
|
| 1100 |
+
"loss": 1.3005435943603516,
|
| 1101 |
+
"step": 1520
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.21717530163236337,
|
| 1105 |
+
"grad_norm": 11.182530403137207,
|
| 1106 |
+
"learning_rate": 9.978310858765082e-05,
|
| 1107 |
+
"loss": 1.3048934936523438,
|
| 1108 |
+
"step": 1530
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.21859474804826118,
|
| 1112 |
+
"grad_norm": 10.190278053283691,
|
| 1113 |
+
"learning_rate": 9.978168914123492e-05,
|
| 1114 |
+
"loss": 1.3993605613708495,
|
| 1115 |
+
"step": 1540
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.22001419446415899,
|
| 1119 |
+
"grad_norm": 10.497735977172852,
|
| 1120 |
+
"learning_rate": 9.978026969481903e-05,
|
| 1121 |
+
"loss": 1.303945541381836,
|
| 1122 |
+
"step": 1550
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.22143364088005676,
|
| 1126 |
+
"grad_norm": 10.535606384277344,
|
| 1127 |
+
"learning_rate": 9.977885024840313e-05,
|
| 1128 |
+
"loss": 1.2210904121398927,
|
| 1129 |
+
"step": 1560
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.22285308729595457,
|
| 1133 |
+
"grad_norm": 11.385029792785645,
|
| 1134 |
+
"learning_rate": 9.977743080198723e-05,
|
| 1135 |
+
"loss": 1.3508376121520995,
|
| 1136 |
+
"step": 1570
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.22427253371185238,
|
| 1140 |
+
"grad_norm": 9.528643608093262,
|
| 1141 |
+
"learning_rate": 9.977601135557132e-05,
|
| 1142 |
+
"loss": 1.2278815269470216,
|
| 1143 |
+
"step": 1580
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.22569198012775019,
|
| 1147 |
+
"grad_norm": 13.161009788513184,
|
| 1148 |
+
"learning_rate": 9.977459190915544e-05,
|
| 1149 |
+
"loss": 1.254448413848877,
|
| 1150 |
+
"step": 1590
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.22711142654364797,
|
| 1154 |
+
"grad_norm": 11.288809776306152,
|
| 1155 |
+
"learning_rate": 9.977317246273953e-05,
|
| 1156 |
+
"loss": 1.271047878265381,
|
| 1157 |
+
"step": 1600
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.22853087295954577,
|
| 1161 |
+
"grad_norm": 11.30105209350586,
|
| 1162 |
+
"learning_rate": 9.977175301632364e-05,
|
| 1163 |
+
"loss": 1.3242988586425781,
|
| 1164 |
+
"step": 1610
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.22995031937544358,
|
| 1168 |
+
"grad_norm": 10.600774765014648,
|
| 1169 |
+
"learning_rate": 9.977033356990774e-05,
|
| 1170 |
+
"loss": 1.3170942306518554,
|
| 1171 |
+
"step": 1620
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.2313697657913414,
|
| 1175 |
+
"grad_norm": 10.652543067932129,
|
| 1176 |
+
"learning_rate": 9.976891412349184e-05,
|
| 1177 |
+
"loss": 1.3998719215393067,
|
| 1178 |
+
"step": 1630
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.23278921220723917,
|
| 1182 |
+
"grad_norm": 11.354793548583984,
|
| 1183 |
+
"learning_rate": 9.976749467707595e-05,
|
| 1184 |
+
"loss": 1.270443820953369,
|
| 1185 |
+
"step": 1640
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.23420865862313697,
|
| 1189 |
+
"grad_norm": 9.926568031311035,
|
| 1190 |
+
"learning_rate": 9.976607523066005e-05,
|
| 1191 |
+
"loss": 1.117215347290039,
|
| 1192 |
+
"step": 1650
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.23562810503903478,
|
| 1196 |
+
"grad_norm": 11.167335510253906,
|
| 1197 |
+
"learning_rate": 9.976465578424416e-05,
|
| 1198 |
+
"loss": 1.348717212677002,
|
| 1199 |
+
"step": 1660
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.2370475514549326,
|
| 1203 |
+
"grad_norm": 11.364425659179688,
|
| 1204 |
+
"learning_rate": 9.976323633782824e-05,
|
| 1205 |
+
"loss": 1.2113998413085938,
|
| 1206 |
+
"step": 1670
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.23846699787083037,
|
| 1210 |
+
"grad_norm": 10.315034866333008,
|
| 1211 |
+
"learning_rate": 9.976181689141235e-05,
|
| 1212 |
+
"loss": 1.2621678352355956,
|
| 1213 |
+
"step": 1680
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.23988644428672817,
|
| 1217 |
+
"grad_norm": 11.332146644592285,
|
| 1218 |
+
"learning_rate": 9.976039744499645e-05,
|
| 1219 |
+
"loss": 1.2919418334960937,
|
| 1220 |
+
"step": 1690
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.24130589070262598,
|
| 1224 |
+
"grad_norm": 9.863037109375,
|
| 1225 |
+
"learning_rate": 9.975897799858056e-05,
|
| 1226 |
+
"loss": 1.262222957611084,
|
| 1227 |
+
"step": 1700
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.2427253371185238,
|
| 1231 |
+
"grad_norm": 13.898163795471191,
|
| 1232 |
+
"learning_rate": 9.975755855216467e-05,
|
| 1233 |
+
"loss": 1.349098300933838,
|
| 1234 |
+
"step": 1710
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.24414478353442157,
|
| 1238 |
+
"grad_norm": 9.008386611938477,
|
| 1239 |
+
"learning_rate": 9.975613910574876e-05,
|
| 1240 |
+
"loss": 1.1653017044067382,
|
| 1241 |
+
"step": 1720
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.24556422995031937,
|
| 1245 |
+
"grad_norm": 9.755669593811035,
|
| 1246 |
+
"learning_rate": 9.975471965933287e-05,
|
| 1247 |
+
"loss": 1.304057788848877,
|
| 1248 |
+
"step": 1730
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.24698367636621718,
|
| 1252 |
+
"grad_norm": 10.742278099060059,
|
| 1253 |
+
"learning_rate": 9.975330021291696e-05,
|
| 1254 |
+
"loss": 1.1656038284301757,
|
| 1255 |
+
"step": 1740
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.248403122782115,
|
| 1259 |
+
"grad_norm": 11.937880516052246,
|
| 1260 |
+
"learning_rate": 9.975188076650107e-05,
|
| 1261 |
+
"loss": 1.2565963745117188,
|
| 1262 |
+
"step": 1750
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.24982256919801277,
|
| 1266 |
+
"grad_norm": 9.80545711517334,
|
| 1267 |
+
"learning_rate": 9.975046132008517e-05,
|
| 1268 |
+
"loss": 1.1316876411437988,
|
| 1269 |
+
"step": 1760
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.2512420156139106,
|
| 1273 |
+
"grad_norm": 11.162557601928711,
|
| 1274 |
+
"learning_rate": 9.974904187366927e-05,
|
| 1275 |
+
"loss": 1.2094581604003907,
|
| 1276 |
+
"step": 1770
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.2526614620298084,
|
| 1280 |
+
"grad_norm": 12.278450965881348,
|
| 1281 |
+
"learning_rate": 9.974762242725337e-05,
|
| 1282 |
+
"loss": 1.2499947547912598,
|
| 1283 |
+
"step": 1780
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.2540809084457062,
|
| 1287 |
+
"grad_norm": 10.95953369140625,
|
| 1288 |
+
"learning_rate": 9.974620298083748e-05,
|
| 1289 |
+
"loss": 1.1540046691894532,
|
| 1290 |
+
"step": 1790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.255500354861604,
|
| 1294 |
+
"grad_norm": 7.865696430206299,
|
| 1295 |
+
"learning_rate": 9.974478353442159e-05,
|
| 1296 |
+
"loss": 1.1665989875793457,
|
| 1297 |
+
"step": 1800
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.25691980127750175,
|
| 1301 |
+
"grad_norm": 12.1609468460083,
|
| 1302 |
+
"learning_rate": 9.974336408800569e-05,
|
| 1303 |
+
"loss": 1.120746898651123,
|
| 1304 |
+
"step": 1810
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.25833924769339955,
|
| 1308 |
+
"grad_norm": 9.554359436035156,
|
| 1309 |
+
"learning_rate": 9.974194464158978e-05,
|
| 1310 |
+
"loss": 1.3381189346313476,
|
| 1311 |
+
"step": 1820
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.25975869410929736,
|
| 1315 |
+
"grad_norm": 9.497129440307617,
|
| 1316 |
+
"learning_rate": 9.974052519517388e-05,
|
| 1317 |
+
"loss": 1.1758546829223633,
|
| 1318 |
+
"step": 1830
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.26117814052519517,
|
| 1322 |
+
"grad_norm": 10.584992408752441,
|
| 1323 |
+
"learning_rate": 9.973910574875799e-05,
|
| 1324 |
+
"loss": 1.0787659645080567,
|
| 1325 |
+
"step": 1840
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.262597586941093,
|
| 1329 |
+
"grad_norm": 9.558980941772461,
|
| 1330 |
+
"learning_rate": 9.973768630234209e-05,
|
| 1331 |
+
"loss": 0.9334567070007325,
|
| 1332 |
+
"step": 1850
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.2640170333569908,
|
| 1336 |
+
"grad_norm": 9.41112995147705,
|
| 1337 |
+
"learning_rate": 9.97362668559262e-05,
|
| 1338 |
+
"loss": 1.1376053810119628,
|
| 1339 |
+
"step": 1860
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.2654364797728886,
|
| 1343 |
+
"grad_norm": 11.666831970214844,
|
| 1344 |
+
"learning_rate": 9.973484740951028e-05,
|
| 1345 |
+
"loss": 1.207914447784424,
|
| 1346 |
+
"step": 1870
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.2668559261887864,
|
| 1350 |
+
"grad_norm": 11.217955589294434,
|
| 1351 |
+
"learning_rate": 9.97334279630944e-05,
|
| 1352 |
+
"loss": 1.052849578857422,
|
| 1353 |
+
"step": 1880
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.26827537260468415,
|
| 1357 |
+
"grad_norm": 8.3615083694458,
|
| 1358 |
+
"learning_rate": 9.97320085166785e-05,
|
| 1359 |
+
"loss": 0.9782976150512696,
|
| 1360 |
+
"step": 1890
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.26969481902058196,
|
| 1364 |
+
"grad_norm": 10.69944953918457,
|
| 1365 |
+
"learning_rate": 9.97305890702626e-05,
|
| 1366 |
+
"loss": 0.9639101982116699,
|
| 1367 |
+
"step": 1900
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.27111426543647976,
|
| 1371 |
+
"grad_norm": 11.15194034576416,
|
| 1372 |
+
"learning_rate": 9.972916962384671e-05,
|
| 1373 |
+
"loss": 1.0744239807128906,
|
| 1374 |
+
"step": 1910
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.27253371185237757,
|
| 1378 |
+
"grad_norm": 10.363690376281738,
|
| 1379 |
+
"learning_rate": 9.972775017743081e-05,
|
| 1380 |
+
"loss": 1.1180108070373536,
|
| 1381 |
+
"step": 1920
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.2739531582682754,
|
| 1385 |
+
"grad_norm": 10.816513061523438,
|
| 1386 |
+
"learning_rate": 9.972633073101491e-05,
|
| 1387 |
+
"loss": 1.118791103363037,
|
| 1388 |
+
"step": 1930
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.2753726046841732,
|
| 1392 |
+
"grad_norm": 8.64388656616211,
|
| 1393 |
+
"learning_rate": 9.9724911284599e-05,
|
| 1394 |
+
"loss": 1.1368459701538085,
|
| 1395 |
+
"step": 1940
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.276792051100071,
|
| 1399 |
+
"grad_norm": 9.002252578735352,
|
| 1400 |
+
"learning_rate": 9.972349183818312e-05,
|
| 1401 |
+
"loss": 1.1344121932983398,
|
| 1402 |
+
"step": 1950
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.2782114975159688,
|
| 1406 |
+
"grad_norm": 11.083386421203613,
|
| 1407 |
+
"learning_rate": 9.972207239176721e-05,
|
| 1408 |
+
"loss": 1.1827295303344727,
|
| 1409 |
+
"step": 1960
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.27963094393186655,
|
| 1413 |
+
"grad_norm": 8.360145568847656,
|
| 1414 |
+
"learning_rate": 9.972065294535133e-05,
|
| 1415 |
+
"loss": 0.9954969406127929,
|
| 1416 |
+
"step": 1970
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.28105039034776436,
|
| 1420 |
+
"grad_norm": 12.982026100158691,
|
| 1421 |
+
"learning_rate": 9.971923349893542e-05,
|
| 1422 |
+
"loss": 0.9865982055664062,
|
| 1423 |
+
"step": 1980
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.28246983676366216,
|
| 1427 |
+
"grad_norm": 9.3854341506958,
|
| 1428 |
+
"learning_rate": 9.971781405251952e-05,
|
| 1429 |
+
"loss": 0.9238475799560547,
|
| 1430 |
+
"step": 1990
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.28388928317955997,
|
| 1434 |
+
"grad_norm": 10.693597793579102,
|
| 1435 |
+
"learning_rate": 9.971639460610363e-05,
|
| 1436 |
+
"loss": 0.9660484313964843,
|
| 1437 |
+
"step": 2000
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.28388928317955997,
|
| 1441 |
+
"eval_accuracy": 0.6596935206968907,
|
| 1442 |
+
"eval_loss": 1.0827350616455078,
|
| 1443 |
+
"eval_runtime": 31.44,
|
| 1444 |
+
"eval_samples_per_second": 500.222,
|
| 1445 |
+
"eval_steps_per_second": 15.649,
|
| 1446 |
+
"step": 2000
|
| 1447 |
+
}
|
| 1448 |
+
],
|
| 1449 |
+
"logging_steps": 10,
|
| 1450 |
+
"max_steps": 704500,
|
| 1451 |
+
"num_input_tokens_seen": 0,
|
| 1452 |
+
"num_train_epochs": 100,
|
| 1453 |
+
"save_steps": 500,
|
| 1454 |
+
"stateful_callbacks": {
|
| 1455 |
+
"TrainerControl": {
|
| 1456 |
+
"args": {
|
| 1457 |
+
"should_epoch_stop": false,
|
| 1458 |
+
"should_evaluate": false,
|
| 1459 |
+
"should_log": false,
|
| 1460 |
+
"should_save": true,
|
| 1461 |
+
"should_training_stop": false
|
| 1462 |
+
},
|
| 1463 |
+
"attributes": {}
|
| 1464 |
+
}
|
| 1465 |
+
},
|
| 1466 |
+
"total_flos": 0.0,
|
| 1467 |
+
"train_batch_size": 32,
|
| 1468 |
+
"trial_name": null,
|
| 1469 |
+
"trial_params": null
|
| 1470 |
+
}
|
resnet50/checkpoint-2000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28e9fe3174d0b803b98c252bc2c29b64559a6dab448804062e04756be96263c8
|
| 3 |
+
size 4856
|