Upload folder using huggingface_hub
Browse files- checkpoints-v2.6-c/checkpoint-71680/ema.safetensors +3 -0
- checkpoints-v2.6-c/checkpoint-71680/model.safetensors +3 -0
- checkpoints-v2.6-c/checkpoint-71680/optimizer.pt +3 -0
- checkpoints-v2.6-c/checkpoint-71680/rng_state.pth +3 -0
- checkpoints-v2.6-c/checkpoint-71680/scaler.pt +3 -0
- checkpoints-v2.6-c/checkpoint-71680/scheduler.pt +3 -0
- checkpoints-v2.6-c/checkpoint-71680/trainer_state.json +2274 -0
- checkpoints-v2.6-c/checkpoint-71680/training_args.bin +3 -0
checkpoints-v2.6-c/checkpoint-71680/ema.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7fa901624a198745648ce7c19a21bae50cae4a030919b7e265f0a20625ec1355
|
| 3 |
+
size 550088
|
checkpoints-v2.6-c/checkpoint-71680/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:518b2a43b5e79e4a6436295916a47b3ca17ece5b79c24340ed4ad7b929197d53
|
| 3 |
+
size 19331448
|
checkpoints-v2.6-c/checkpoint-71680/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:10b99b6602b1c42cdfb13ea1caf8d5d91065c2f48d8cdb101806ebe56d268ab8
|
| 3 |
+
size 1175115
|
checkpoints-v2.6-c/checkpoint-71680/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5eec900aba99141679433683f7821f1560dd0ab767380e4157dee35db2b863a8
|
| 3 |
+
size 14645
|
checkpoints-v2.6-c/checkpoint-71680/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1da24dc4171c201e81c1763b65f1284392cb5bd2be3be942c424c72244961094
|
| 3 |
+
size 1383
|
checkpoints-v2.6-c/checkpoint-71680/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:64f499961cb1867b9b55bfc1f504651538bc763adf3fba23857186038ee5104b
|
| 3 |
+
size 1465
|
checkpoints-v2.6-c/checkpoint-71680/trainer_state.json
ADDED
|
@@ -0,0 +1,2274 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.7450446423931233,
|
| 6 |
+
"eval_steps": 1024,
|
| 7 |
+
"global_step": 71680,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.010643494891330332,
|
| 14 |
+
"grad_norm": 0.13342437148094177,
|
| 15 |
+
"learning_rate": 0.0003330078125,
|
| 16 |
+
"loss": 2.2998437881469727,
|
| 17 |
+
"step": 1024
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.010643494891330332,
|
| 21 |
+
"eval_cos_loss": 0.5988449528813362,
|
| 22 |
+
"eval_loss": 1.9600126259028912,
|
| 23 |
+
"eval_mse_loss": 1.6605901420116425,
|
| 24 |
+
"flow/cos_sim": 0.4011551085859537,
|
| 25 |
+
"flow/improvement_ratio": 0.942170824855566,
|
| 26 |
+
"flow/mag_ratio_mean": 0.37856073677539825,
|
| 27 |
+
"flow/mag_ratio_std": 0.14085532305762172,
|
| 28 |
+
"step": 1024
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.010643494891330332,
|
| 32 |
+
"eval_cos_loss": 0.5988449528813362,
|
| 33 |
+
"eval_loss": 1.9600126259028912,
|
| 34 |
+
"eval_mse_loss": 1.6605901420116425,
|
| 35 |
+
"eval_runtime": 2.6584,
|
| 36 |
+
"eval_samples_per_second": 752.329,
|
| 37 |
+
"eval_steps_per_second": 12.037,
|
| 38 |
+
"flow/cos_sim": 0.4011551085859537,
|
| 39 |
+
"flow/improvement_ratio": 0.942170824855566,
|
| 40 |
+
"flow/mag_ratio_mean": 0.37856073677539825,
|
| 41 |
+
"flow/mag_ratio_std": 0.14085532305762172,
|
| 42 |
+
"step": 1024
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.021286989782660665,
|
| 46 |
+
"grad_norm": 0.25054192543029785,
|
| 47 |
+
"learning_rate": 0.0006663411458333333,
|
| 48 |
+
"loss": 1.8492329120635986,
|
| 49 |
+
"step": 2048
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.021286989782660665,
|
| 53 |
+
"eval_cos_loss": 0.5117531130090356,
|
| 54 |
+
"eval_loss": 1.7429817728698254,
|
| 55 |
+
"eval_mse_loss": 1.4871052131056786,
|
| 56 |
+
"flow/cos_sim": 0.4882468534633517,
|
| 57 |
+
"flow/improvement_ratio": 0.9563530795276165,
|
| 58 |
+
"flow/mag_ratio_mean": 0.47669631242752075,
|
| 59 |
+
"flow/mag_ratio_std": 0.17675806442275643,
|
| 60 |
+
"step": 2048
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.021286989782660665,
|
| 64 |
+
"eval_cos_loss": 0.5117531130090356,
|
| 65 |
+
"eval_loss": 1.7429817728698254,
|
| 66 |
+
"eval_mse_loss": 1.4871052131056786,
|
| 67 |
+
"eval_runtime": 2.511,
|
| 68 |
+
"eval_samples_per_second": 796.509,
|
| 69 |
+
"eval_steps_per_second": 12.744,
|
| 70 |
+
"flow/cos_sim": 0.4882468534633517,
|
| 71 |
+
"flow/improvement_ratio": 0.9563530795276165,
|
| 72 |
+
"flow/mag_ratio_mean": 0.47669631242752075,
|
| 73 |
+
"flow/mag_ratio_std": 0.17675806442275643,
|
| 74 |
+
"step": 2048
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.031930484673991,
|
| 78 |
+
"grad_norm": 0.30941224098205566,
|
| 79 |
+
"learning_rate": 0.0009996744791666667,
|
| 80 |
+
"loss": 1.730944037437439,
|
| 81 |
+
"step": 3072
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.031930484673991,
|
| 85 |
+
"eval_cos_loss": 0.4815286351367831,
|
| 86 |
+
"eval_loss": 1.6586528308689594,
|
| 87 |
+
"eval_mse_loss": 1.4178885221481323,
|
| 88 |
+
"flow/cos_sim": 0.5184714393690228,
|
| 89 |
+
"flow/improvement_ratio": 0.9605911839753389,
|
| 90 |
+
"flow/mag_ratio_mean": 0.49818364903330803,
|
| 91 |
+
"flow/mag_ratio_std": 0.1928270636126399,
|
| 92 |
+
"step": 3072
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.031930484673991,
|
| 96 |
+
"eval_cos_loss": 0.4815286351367831,
|
| 97 |
+
"eval_loss": 1.6586528308689594,
|
| 98 |
+
"eval_mse_loss": 1.4178885221481323,
|
| 99 |
+
"eval_runtime": 3.1033,
|
| 100 |
+
"eval_samples_per_second": 644.485,
|
| 101 |
+
"eval_steps_per_second": 10.312,
|
| 102 |
+
"flow/cos_sim": 0.5184714393690228,
|
| 103 |
+
"flow/improvement_ratio": 0.9605911839753389,
|
| 104 |
+
"flow/mag_ratio_mean": 0.49818364903330803,
|
| 105 |
+
"flow/mag_ratio_std": 0.1928270636126399,
|
| 106 |
+
"step": 3072
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.04257397956532133,
|
| 110 |
+
"grad_norm": 0.22964967787265778,
|
| 111 |
+
"learning_rate": 0.0009997023516784352,
|
| 112 |
+
"loss": 1.6850833892822266,
|
| 113 |
+
"step": 4096
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.04257397956532133,
|
| 117 |
+
"eval_cos_loss": 0.476364528760314,
|
| 118 |
+
"eval_loss": 1.6391540355980396,
|
| 119 |
+
"eval_mse_loss": 1.4009717665612698,
|
| 120 |
+
"flow/cos_sim": 0.5236354488879442,
|
| 121 |
+
"flow/improvement_ratio": 0.9618693646043539,
|
| 122 |
+
"flow/mag_ratio_mean": 0.5105963433161378,
|
| 123 |
+
"flow/mag_ratio_std": 0.20592432795092463,
|
| 124 |
+
"step": 4096
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.04257397956532133,
|
| 128 |
+
"eval_cos_loss": 0.476364528760314,
|
| 129 |
+
"eval_loss": 1.6391540355980396,
|
| 130 |
+
"eval_mse_loss": 1.4009717665612698,
|
| 131 |
+
"eval_runtime": 2.5129,
|
| 132 |
+
"eval_samples_per_second": 795.895,
|
| 133 |
+
"eval_steps_per_second": 12.734,
|
| 134 |
+
"flow/cos_sim": 0.5236354488879442,
|
| 135 |
+
"flow/improvement_ratio": 0.9618693646043539,
|
| 136 |
+
"flow/mag_ratio_mean": 0.5105963433161378,
|
| 137 |
+
"flow/mag_ratio_std": 0.20592432795092463,
|
| 138 |
+
"step": 4096
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.05321747445665166,
|
| 142 |
+
"grad_norm": 0.2645546495914459,
|
| 143 |
+
"learning_rate": 0.0009988085977910004,
|
| 144 |
+
"loss": 1.6617510318756104,
|
| 145 |
+
"step": 5120
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.05321747445665166,
|
| 149 |
+
"eval_cos_loss": 0.4789119102060795,
|
| 150 |
+
"eval_loss": 1.645260013639927,
|
| 151 |
+
"eval_mse_loss": 1.405804067850113,
|
| 152 |
+
"flow/cos_sim": 0.5210880534723401,
|
| 153 |
+
"flow/improvement_ratio": 0.9588682930916548,
|
| 154 |
+
"flow/mag_ratio_mean": 0.504288200289011,
|
| 155 |
+
"flow/mag_ratio_std": 0.20718340016901493,
|
| 156 |
+
"step": 5120
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.05321747445665166,
|
| 160 |
+
"eval_cos_loss": 0.4789119102060795,
|
| 161 |
+
"eval_loss": 1.645260013639927,
|
| 162 |
+
"eval_mse_loss": 1.405804067850113,
|
| 163 |
+
"eval_runtime": 3.0976,
|
| 164 |
+
"eval_samples_per_second": 645.654,
|
| 165 |
+
"eval_steps_per_second": 10.33,
|
| 166 |
+
"flow/cos_sim": 0.5210880534723401,
|
| 167 |
+
"flow/improvement_ratio": 0.9588682930916548,
|
| 168 |
+
"flow/mag_ratio_mean": 0.504288200289011,
|
| 169 |
+
"flow/mag_ratio_std": 0.20718340016901493,
|
| 170 |
+
"step": 5120
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.063860969347982,
|
| 174 |
+
"grad_norm": 0.2762889862060547,
|
| 175 |
+
"learning_rate": 0.0009973198042317873,
|
| 176 |
+
"loss": 1.645796775817871,
|
| 177 |
+
"step": 6144
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.063860969347982,
|
| 181 |
+
"eval_cos_loss": 0.4598818449303508,
|
| 182 |
+
"eval_loss": 1.5948525853455067,
|
| 183 |
+
"eval_mse_loss": 1.3649116680026054,
|
| 184 |
+
"flow/cos_sim": 0.5401182025671005,
|
| 185 |
+
"flow/improvement_ratio": 0.9647715575993061,
|
| 186 |
+
"flow/mag_ratio_mean": 0.5178880272433162,
|
| 187 |
+
"flow/mag_ratio_std": 0.21153279254212976,
|
| 188 |
+
"step": 6144
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.063860969347982,
|
| 192 |
+
"eval_cos_loss": 0.4598818449303508,
|
| 193 |
+
"eval_loss": 1.5948525853455067,
|
| 194 |
+
"eval_mse_loss": 1.3649116680026054,
|
| 195 |
+
"eval_runtime": 3.0831,
|
| 196 |
+
"eval_samples_per_second": 648.695,
|
| 197 |
+
"eval_steps_per_second": 10.379,
|
| 198 |
+
"flow/cos_sim": 0.5401182025671005,
|
| 199 |
+
"flow/improvement_ratio": 0.9647715575993061,
|
| 200 |
+
"flow/mag_ratio_mean": 0.5178880272433162,
|
| 201 |
+
"flow/mag_ratio_std": 0.21153279254212976,
|
| 202 |
+
"step": 6144
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.07450446423931233,
|
| 206 |
+
"grad_norm": 0.17679959535598755,
|
| 207 |
+
"learning_rate": 0.0009952377470151526,
|
| 208 |
+
"loss": 1.6353809833526611,
|
| 209 |
+
"step": 7168
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.07450446423931233,
|
| 213 |
+
"eval_cos_loss": 0.4634226718917489,
|
| 214 |
+
"eval_loss": 1.6022505089640617,
|
| 215 |
+
"eval_mse_loss": 1.3705391697585583,
|
| 216 |
+
"flow/cos_sim": 0.5365773290395737,
|
| 217 |
+
"flow/improvement_ratio": 0.9635819494724274,
|
| 218 |
+
"flow/mag_ratio_mean": 0.5194354858249426,
|
| 219 |
+
"flow/mag_ratio_std": 0.21515046246349812,
|
| 220 |
+
"step": 7168
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.07450446423931233,
|
| 224 |
+
"eval_cos_loss": 0.4634226718917489,
|
| 225 |
+
"eval_loss": 1.6022505089640617,
|
| 226 |
+
"eval_mse_loss": 1.3705391697585583,
|
| 227 |
+
"eval_runtime": 2.8419,
|
| 228 |
+
"eval_samples_per_second": 703.759,
|
| 229 |
+
"eval_steps_per_second": 11.26,
|
| 230 |
+
"flow/cos_sim": 0.5365773290395737,
|
| 231 |
+
"flow/improvement_ratio": 0.9635819494724274,
|
| 232 |
+
"flow/mag_ratio_mean": 0.5194354858249426,
|
| 233 |
+
"flow/mag_ratio_std": 0.21515046246349812,
|
| 234 |
+
"step": 7168
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.08514795913064266,
|
| 238 |
+
"grad_norm": 0.14975515007972717,
|
| 239 |
+
"learning_rate": 0.000992564909872628,
|
| 240 |
+
"loss": 1.6262034177780151,
|
| 241 |
+
"step": 8192
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.08514795913064266,
|
| 245 |
+
"eval_cos_loss": 0.45912545546889305,
|
| 246 |
+
"eval_loss": 1.5890175811946392,
|
| 247 |
+
"eval_mse_loss": 1.3594548553228378,
|
| 248 |
+
"flow/cos_sim": 0.5408745482563972,
|
| 249 |
+
"flow/improvement_ratio": 0.9590303134173155,
|
| 250 |
+
"flow/mag_ratio_mean": 0.5143361240625381,
|
| 251 |
+
"flow/mag_ratio_std": 0.21537457825616002,
|
| 252 |
+
"step": 8192
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.08514795913064266,
|
| 256 |
+
"eval_cos_loss": 0.45912545546889305,
|
| 257 |
+
"eval_loss": 1.5890175811946392,
|
| 258 |
+
"eval_mse_loss": 1.3594548553228378,
|
| 259 |
+
"eval_runtime": 2.9302,
|
| 260 |
+
"eval_samples_per_second": 682.537,
|
| 261 |
+
"eval_steps_per_second": 10.921,
|
| 262 |
+
"flow/cos_sim": 0.5408745482563972,
|
| 263 |
+
"flow/improvement_ratio": 0.9590303134173155,
|
| 264 |
+
"flow/mag_ratio_mean": 0.5143361240625381,
|
| 265 |
+
"flow/mag_ratio_std": 0.21537457825616002,
|
| 266 |
+
"step": 8192
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.09579145402197299,
|
| 270 |
+
"grad_norm": 0.19106586277484894,
|
| 271 |
+
"learning_rate": 0.000989307950724573,
|
| 272 |
+
"loss": 1.6214015483856201,
|
| 273 |
+
"step": 9216
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.09579145402197299,
|
| 277 |
+
"eval_cos_loss": 0.4567577252164483,
|
| 278 |
+
"eval_loss": 1.5844898335635662,
|
| 279 |
+
"eval_mse_loss": 1.356110967695713,
|
| 280 |
+
"flow/cos_sim": 0.5432424321770668,
|
| 281 |
+
"flow/improvement_ratio": 0.9650511220097542,
|
| 282 |
+
"flow/mag_ratio_mean": 0.5244949720799923,
|
| 283 |
+
"flow/mag_ratio_std": 0.21130397450178862,
|
| 284 |
+
"step": 9216
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.09579145402197299,
|
| 288 |
+
"eval_cos_loss": 0.4567577252164483,
|
| 289 |
+
"eval_loss": 1.5844898335635662,
|
| 290 |
+
"eval_mse_loss": 1.356110967695713,
|
| 291 |
+
"eval_runtime": 3.0508,
|
| 292 |
+
"eval_samples_per_second": 655.562,
|
| 293 |
+
"eval_steps_per_second": 10.489,
|
| 294 |
+
"flow/cos_sim": 0.5432424321770668,
|
| 295 |
+
"flow/improvement_ratio": 0.9650511220097542,
|
| 296 |
+
"flow/mag_ratio_mean": 0.5244949720799923,
|
| 297 |
+
"flow/mag_ratio_std": 0.21130397450178862,
|
| 298 |
+
"step": 9216
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.10643494891330332,
|
| 302 |
+
"grad_norm": 0.22245089709758759,
|
| 303 |
+
"learning_rate": 0.000985464388035817,
|
| 304 |
+
"loss": 1.6132733821868896,
|
| 305 |
+
"step": 10240
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.10643494891330332,
|
| 309 |
+
"eval_cos_loss": 0.4598613306879997,
|
| 310 |
+
"eval_loss": 1.589576181024313,
|
| 311 |
+
"eval_mse_loss": 1.359645515680313,
|
| 312 |
+
"flow/cos_sim": 0.5401386898010969,
|
| 313 |
+
"flow/improvement_ratio": 0.9610863700509071,
|
| 314 |
+
"flow/mag_ratio_mean": 0.5160716716200113,
|
| 315 |
+
"flow/mag_ratio_std": 0.21545762522146106,
|
| 316 |
+
"step": 10240
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.10643494891330332,
|
| 320 |
+
"eval_cos_loss": 0.4598613306879997,
|
| 321 |
+
"eval_loss": 1.589576181024313,
|
| 322 |
+
"eval_mse_loss": 1.359645515680313,
|
| 323 |
+
"eval_runtime": 3.1847,
|
| 324 |
+
"eval_samples_per_second": 627.993,
|
| 325 |
+
"eval_steps_per_second": 10.048,
|
| 326 |
+
"flow/cos_sim": 0.5401386898010969,
|
| 327 |
+
"flow/improvement_ratio": 0.9610863700509071,
|
| 328 |
+
"flow/mag_ratio_mean": 0.5160716716200113,
|
| 329 |
+
"flow/mag_ratio_std": 0.21545762522146106,
|
| 330 |
+
"step": 10240
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.11707844380463366,
|
| 334 |
+
"grad_norm": 0.1567550003528595,
|
| 335 |
+
"learning_rate": 0.0009810417042745768,
|
| 336 |
+
"loss": 1.6070518493652344,
|
| 337 |
+
"step": 11264
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.11707844380463366,
|
| 341 |
+
"eval_cos_loss": 0.4550258554518223,
|
| 342 |
+
"eval_loss": 1.577816877514124,
|
| 343 |
+
"eval_mse_loss": 1.350303951650858,
|
| 344 |
+
"flow/cos_sim": 0.5449741557240486,
|
| 345 |
+
"flow/improvement_ratio": 0.9648044053465128,
|
| 346 |
+
"flow/mag_ratio_mean": 0.5290831215679646,
|
| 347 |
+
"flow/mag_ratio_std": 0.21279342425987124,
|
| 348 |
+
"step": 11264
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.11707844380463366,
|
| 352 |
+
"eval_cos_loss": 0.4550258554518223,
|
| 353 |
+
"eval_loss": 1.577816877514124,
|
| 354 |
+
"eval_mse_loss": 1.350303951650858,
|
| 355 |
+
"eval_runtime": 2.6352,
|
| 356 |
+
"eval_samples_per_second": 758.945,
|
| 357 |
+
"eval_steps_per_second": 12.143,
|
| 358 |
+
"flow/cos_sim": 0.5449741557240486,
|
| 359 |
+
"flow/improvement_ratio": 0.9648044053465128,
|
| 360 |
+
"flow/mag_ratio_mean": 0.5290831215679646,
|
| 361 |
+
"flow/mag_ratio_std": 0.21279342425987124,
|
| 362 |
+
"step": 11264
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.127721938695964,
|
| 366 |
+
"grad_norm": 0.19316641986370087,
|
| 367 |
+
"learning_rate": 0.0009760451753569162,
|
| 368 |
+
"loss": 1.6028146743774414,
|
| 369 |
+
"step": 12288
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.127721938695964,
|
| 373 |
+
"eval_cos_loss": 0.45873888209462166,
|
| 374 |
+
"eval_loss": 1.5860362015664577,
|
| 375 |
+
"eval_mse_loss": 1.3566667586565018,
|
| 376 |
+
"flow/cos_sim": 0.5412612538784742,
|
| 377 |
+
"flow/improvement_ratio": 0.9610528890043497,
|
| 378 |
+
"flow/mag_ratio_mean": 0.5225661229342222,
|
| 379 |
+
"flow/mag_ratio_std": 0.2149493475444615,
|
| 380 |
+
"step": 12288
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.127721938695964,
|
| 384 |
+
"eval_cos_loss": 0.45873888209462166,
|
| 385 |
+
"eval_loss": 1.5860362015664577,
|
| 386 |
+
"eval_mse_loss": 1.3566667586565018,
|
| 387 |
+
"eval_runtime": 2.6091,
|
| 388 |
+
"eval_samples_per_second": 766.55,
|
| 389 |
+
"eval_steps_per_second": 12.265,
|
| 390 |
+
"flow/cos_sim": 0.5412612538784742,
|
| 391 |
+
"flow/improvement_ratio": 0.9610528890043497,
|
| 392 |
+
"flow/mag_ratio_mean": 0.5225661229342222,
|
| 393 |
+
"flow/mag_ratio_std": 0.2149493475444615,
|
| 394 |
+
"step": 12288
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.13836543358729433,
|
| 398 |
+
"grad_norm": 0.17066629230976105,
|
| 399 |
+
"learning_rate": 0.000970486470662755,
|
| 400 |
+
"loss": 1.5989067554473877,
|
| 401 |
+
"step": 13312
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.13836543358729433,
|
| 405 |
+
"eval_cos_loss": 0.4526587063446641,
|
| 406 |
+
"eval_loss": 1.5703520886600018,
|
| 407 |
+
"eval_mse_loss": 1.3440227322280407,
|
| 408 |
+
"flow/cos_sim": 0.547341376543045,
|
| 409 |
+
"flow/improvement_ratio": 0.9634687285870314,
|
| 410 |
+
"flow/mag_ratio_mean": 0.5251006819307804,
|
| 411 |
+
"flow/mag_ratio_std": 0.2169443154707551,
|
| 412 |
+
"step": 13312
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 0.13836543358729433,
|
| 416 |
+
"eval_cos_loss": 0.4526587063446641,
|
| 417 |
+
"eval_loss": 1.5703520886600018,
|
| 418 |
+
"eval_mse_loss": 1.3440227322280407,
|
| 419 |
+
"eval_runtime": 2.6502,
|
| 420 |
+
"eval_samples_per_second": 754.652,
|
| 421 |
+
"eval_steps_per_second": 12.074,
|
| 422 |
+
"flow/cos_sim": 0.547341376543045,
|
| 423 |
+
"flow/improvement_ratio": 0.9634687285870314,
|
| 424 |
+
"flow/mag_ratio_mean": 0.5251006819307804,
|
| 425 |
+
"flow/mag_ratio_std": 0.2169443154707551,
|
| 426 |
+
"step": 13312
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.14900892847862465,
|
| 430 |
+
"grad_norm": 0.19086262583732605,
|
| 431 |
+
"learning_rate": 0.0009643613549160033,
|
| 432 |
+
"loss": 1.5941526889801025,
|
| 433 |
+
"step": 14336
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.14900892847862465,
|
| 437 |
+
"eval_cos_loss": 0.45674111880362034,
|
| 438 |
+
"eval_loss": 1.5803881026804447,
|
| 439 |
+
"eval_mse_loss": 1.3520175516605377,
|
| 440 |
+
"flow/cos_sim": 0.5432589612901211,
|
| 441 |
+
"flow/improvement_ratio": 0.9569191709160805,
|
| 442 |
+
"flow/mag_ratio_mean": 0.5241195531561971,
|
| 443 |
+
"flow/mag_ratio_std": 0.2207528604194522,
|
| 444 |
+
"step": 14336
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.14900892847862465,
|
| 448 |
+
"eval_cos_loss": 0.45674111880362034,
|
| 449 |
+
"eval_loss": 1.5803881026804447,
|
| 450 |
+
"eval_mse_loss": 1.3520175516605377,
|
| 451 |
+
"eval_runtime": 2.84,
|
| 452 |
+
"eval_samples_per_second": 704.229,
|
| 453 |
+
"eval_steps_per_second": 11.268,
|
| 454 |
+
"flow/cos_sim": 0.5432589612901211,
|
| 455 |
+
"flow/improvement_ratio": 0.9569191709160805,
|
| 456 |
+
"flow/mag_ratio_mean": 0.5241195531561971,
|
| 457 |
+
"flow/mag_ratio_std": 0.2207528604194522,
|
| 458 |
+
"step": 14336
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.159652423369955,
|
| 462 |
+
"grad_norm": 0.20660291612148285,
|
| 463 |
+
"learning_rate": 0.0009576890825691249,
|
| 464 |
+
"loss": 1.5903245210647583,
|
| 465 |
+
"step": 15360
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.159652423369955,
|
| 469 |
+
"eval_cos_loss": 0.4470532648265362,
|
| 470 |
+
"eval_loss": 1.5533855073153973,
|
| 471 |
+
"eval_mse_loss": 1.329858873039484,
|
| 472 |
+
"flow/cos_sim": 0.5529466420412064,
|
| 473 |
+
"flow/improvement_ratio": 0.9680595081299543,
|
| 474 |
+
"flow/mag_ratio_mean": 0.5352848172187805,
|
| 475 |
+
"flow/mag_ratio_std": 0.22097993176430464,
|
| 476 |
+
"step": 15360
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.159652423369955,
|
| 480 |
+
"eval_cos_loss": 0.4470532648265362,
|
| 481 |
+
"eval_loss": 1.5533855073153973,
|
| 482 |
+
"eval_mse_loss": 1.329858873039484,
|
| 483 |
+
"eval_runtime": 2.6578,
|
| 484 |
+
"eval_samples_per_second": 752.494,
|
| 485 |
+
"eval_steps_per_second": 12.04,
|
| 486 |
+
"flow/cos_sim": 0.5529466420412064,
|
| 487 |
+
"flow/improvement_ratio": 0.9680595081299543,
|
| 488 |
+
"flow/mag_ratio_mean": 0.5352848172187805,
|
| 489 |
+
"flow/mag_ratio_std": 0.22097993176430464,
|
| 490 |
+
"step": 15360
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.17029591826128532,
|
| 494 |
+
"grad_norm": 0.23885692656040192,
|
| 495 |
+
"learning_rate": 0.0009504645698990064,
|
| 496 |
+
"loss": 1.589218020439148,
|
| 497 |
+
"step": 16384
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.17029591826128532,
|
| 501 |
+
"eval_cos_loss": 0.44447089545428753,
|
| 502 |
+
"eval_loss": 1.5484142042696476,
|
| 503 |
+
"eval_mse_loss": 1.3261787556111813,
|
| 504 |
+
"flow/cos_sim": 0.5555290877819061,
|
| 505 |
+
"flow/improvement_ratio": 0.9635521955788136,
|
| 506 |
+
"flow/mag_ratio_mean": 0.5299641713500023,
|
| 507 |
+
"flow/mag_ratio_std": 0.215805409476161,
|
| 508 |
+
"step": 16384
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.17029591826128532,
|
| 512 |
+
"eval_cos_loss": 0.44447089545428753,
|
| 513 |
+
"eval_loss": 1.5484142042696476,
|
| 514 |
+
"eval_mse_loss": 1.3261787556111813,
|
| 515 |
+
"eval_runtime": 2.6371,
|
| 516 |
+
"eval_samples_per_second": 758.402,
|
| 517 |
+
"eval_steps_per_second": 12.134,
|
| 518 |
+
"flow/cos_sim": 0.5555290877819061,
|
| 519 |
+
"flow/improvement_ratio": 0.9635521955788136,
|
| 520 |
+
"flow/mag_ratio_mean": 0.5299641713500023,
|
| 521 |
+
"flow/mag_ratio_std": 0.215805409476161,
|
| 522 |
+
"step": 16384
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.18093941315261566,
|
| 526 |
+
"grad_norm": 0.2062983363866806,
|
| 527 |
+
"learning_rate": 0.0009427105273394636,
|
| 528 |
+
"loss": 1.585401177406311,
|
| 529 |
+
"step": 17408
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.18093941315261566,
|
| 533 |
+
"eval_cos_loss": 0.45094432309269905,
|
| 534 |
+
"eval_loss": 1.5633347816765308,
|
| 535 |
+
"eval_mse_loss": 1.3378626182675362,
|
| 536 |
+
"flow/cos_sim": 0.5490557141602039,
|
| 537 |
+
"flow/improvement_ratio": 0.9632246606051922,
|
| 538 |
+
"flow/mag_ratio_mean": 0.5234426287934184,
|
| 539 |
+
"flow/mag_ratio_std": 0.22044725203886628,
|
| 540 |
+
"step": 17408
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.18093941315261566,
|
| 544 |
+
"eval_cos_loss": 0.45094432309269905,
|
| 545 |
+
"eval_loss": 1.5633347816765308,
|
| 546 |
+
"eval_mse_loss": 1.3378626182675362,
|
| 547 |
+
"eval_runtime": 2.5802,
|
| 548 |
+
"eval_samples_per_second": 775.125,
|
| 549 |
+
"eval_steps_per_second": 12.402,
|
| 550 |
+
"flow/cos_sim": 0.5490557141602039,
|
| 551 |
+
"flow/improvement_ratio": 0.9632246606051922,
|
| 552 |
+
"flow/mag_ratio_mean": 0.5234426287934184,
|
| 553 |
+
"flow/mag_ratio_std": 0.22044725203886628,
|
| 554 |
+
"step": 17408
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.19158290804394598,
|
| 558 |
+
"grad_norm": 0.15858766436576843,
|
| 559 |
+
"learning_rate": 0.0009344210469473947,
|
| 560 |
+
"loss": 1.5826770067214966,
|
| 561 |
+
"step": 18432
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.19158290804394598,
|
| 565 |
+
"eval_cos_loss": 0.44898632261902094,
|
| 566 |
+
"eval_loss": 1.5564597770571709,
|
| 567 |
+
"eval_mse_loss": 1.331966608762741,
|
| 568 |
+
"flow/cos_sim": 0.5510137844830751,
|
| 569 |
+
"flow/improvement_ratio": 0.9625816307961941,
|
| 570 |
+
"flow/mag_ratio_mean": 0.5298811597749591,
|
| 571 |
+
"flow/mag_ratio_std": 0.22253544814884663,
|
| 572 |
+
"step": 18432
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.19158290804394598,
|
| 576 |
+
"eval_cos_loss": 0.44898632261902094,
|
| 577 |
+
"eval_loss": 1.5564597770571709,
|
| 578 |
+
"eval_mse_loss": 1.331966608762741,
|
| 579 |
+
"eval_runtime": 2.5531,
|
| 580 |
+
"eval_samples_per_second": 783.347,
|
| 581 |
+
"eval_steps_per_second": 12.534,
|
| 582 |
+
"flow/cos_sim": 0.5510137844830751,
|
| 583 |
+
"flow/improvement_ratio": 0.9625816307961941,
|
| 584 |
+
"flow/mag_ratio_mean": 0.5298811597749591,
|
| 585 |
+
"flow/mag_ratio_std": 0.22253544814884663,
|
| 586 |
+
"step": 18432
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.20222640293527633,
|
| 590 |
+
"grad_norm": 0.2525703012943268,
|
| 591 |
+
"learning_rate": 0.0009256133361993658,
|
| 592 |
+
"loss": 1.5798900127410889,
|
| 593 |
+
"step": 19456
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.20222640293527633,
|
| 597 |
+
"eval_cos_loss": 0.45141084399074316,
|
| 598 |
+
"eval_loss": 1.567859135568142,
|
| 599 |
+
"eval_mse_loss": 1.3421537093818188,
|
| 600 |
+
"flow/cos_sim": 0.5485891196876764,
|
| 601 |
+
"flow/improvement_ratio": 0.96523248963058,
|
| 602 |
+
"flow/mag_ratio_mean": 0.5181732634082437,
|
| 603 |
+
"flow/mag_ratio_std": 0.22030179109424353,
|
| 604 |
+
"step": 19456
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.20222640293527633,
|
| 608 |
+
"eval_cos_loss": 0.45141084399074316,
|
| 609 |
+
"eval_loss": 1.567859135568142,
|
| 610 |
+
"eval_mse_loss": 1.3421537093818188,
|
| 611 |
+
"eval_runtime": 2.5625,
|
| 612 |
+
"eval_samples_per_second": 780.491,
|
| 613 |
+
"eval_steps_per_second": 12.488,
|
| 614 |
+
"flow/cos_sim": 0.5485891196876764,
|
| 615 |
+
"flow/improvement_ratio": 0.96523248963058,
|
| 616 |
+
"flow/mag_ratio_mean": 0.5181732634082437,
|
| 617 |
+
"flow/mag_ratio_std": 0.22030179109424353,
|
| 618 |
+
"step": 19456
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.21286989782660665,
|
| 622 |
+
"grad_norm": 0.18312996625900269,
|
| 623 |
+
"learning_rate": 0.0009163072432159066,
|
| 624 |
+
"loss": 1.579535961151123,
|
| 625 |
+
"step": 20480
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.21286989782660665,
|
| 629 |
+
"eval_cos_loss": 0.45327545143663883,
|
| 630 |
+
"eval_loss": 1.5679056644439697,
|
| 631 |
+
"eval_mse_loss": 1.3412679433822632,
|
| 632 |
+
"flow/cos_sim": 0.5467245355248451,
|
| 633 |
+
"flow/improvement_ratio": 0.9616729654371738,
|
| 634 |
+
"flow/mag_ratio_mean": 0.5273217614740133,
|
| 635 |
+
"flow/mag_ratio_std": 0.2235504975542426,
|
| 636 |
+
"step": 20480
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.21286989782660665,
|
| 640 |
+
"eval_cos_loss": 0.45327545143663883,
|
| 641 |
+
"eval_loss": 1.5679056644439697,
|
| 642 |
+
"eval_mse_loss": 1.3412679433822632,
|
| 643 |
+
"eval_runtime": 2.5697,
|
| 644 |
+
"eval_samples_per_second": 778.293,
|
| 645 |
+
"eval_steps_per_second": 12.453,
|
| 646 |
+
"flow/cos_sim": 0.5467245355248451,
|
| 647 |
+
"flow/improvement_ratio": 0.9616729654371738,
|
| 648 |
+
"flow/mag_ratio_mean": 0.5273217614740133,
|
| 649 |
+
"flow/mag_ratio_std": 0.2235504975542426,
|
| 650 |
+
"step": 20480
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.223513392717937,
|
| 654 |
+
"grad_norm": 0.21262691915035248,
|
| 655 |
+
"learning_rate": 0.0009064956775190607,
|
| 656 |
+
"loss": 1.577104926109314,
|
| 657 |
+
"step": 21504
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.223513392717937,
|
| 661 |
+
"eval_cos_loss": 0.4483450762927532,
|
| 662 |
+
"eval_loss": 1.553330171853304,
|
| 663 |
+
"eval_mse_loss": 1.329157643020153,
|
| 664 |
+
"flow/cos_sim": 0.5516549795866013,
|
| 665 |
+
"flow/improvement_ratio": 0.9627660047262907,
|
| 666 |
+
"flow/mag_ratio_mean": 0.5387043142691255,
|
| 667 |
+
"flow/mag_ratio_std": 0.2251730626448989,
|
| 668 |
+
"step": 21504
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.223513392717937,
|
| 672 |
+
"eval_cos_loss": 0.4483450762927532,
|
| 673 |
+
"eval_loss": 1.553330171853304,
|
| 674 |
+
"eval_mse_loss": 1.329157643020153,
|
| 675 |
+
"eval_runtime": 2.5071,
|
| 676 |
+
"eval_samples_per_second": 797.724,
|
| 677 |
+
"eval_steps_per_second": 12.764,
|
| 678 |
+
"flow/cos_sim": 0.5516549795866013,
|
| 679 |
+
"flow/improvement_ratio": 0.9627660047262907,
|
| 680 |
+
"flow/mag_ratio_mean": 0.5387043142691255,
|
| 681 |
+
"flow/mag_ratio_std": 0.2251730626448989,
|
| 682 |
+
"step": 21504
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.2341568876092673,
|
| 686 |
+
"grad_norm": 0.17988671362400055,
|
| 687 |
+
"learning_rate": 0.0008961991942494195,
|
| 688 |
+
"loss": 1.574266791343689,
|
| 689 |
+
"step": 22528
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.2341568876092673,
|
| 693 |
+
"eval_cos_loss": 0.44411917496472597,
|
| 694 |
+
"eval_loss": 1.543789055198431,
|
| 695 |
+
"eval_mse_loss": 1.3217294700443745,
|
| 696 |
+
"flow/cos_sim": 0.5558808352798223,
|
| 697 |
+
"flow/improvement_ratio": 0.9671246875077486,
|
| 698 |
+
"flow/mag_ratio_mean": 0.5348946927115321,
|
| 699 |
+
"flow/mag_ratio_std": 0.22368196118623018,
|
| 700 |
+
"step": 22528
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 0.2341568876092673,
|
| 704 |
+
"eval_cos_loss": 0.44411917496472597,
|
| 705 |
+
"eval_loss": 1.543789055198431,
|
| 706 |
+
"eval_mse_loss": 1.3217294700443745,
|
| 707 |
+
"eval_runtime": 2.5967,
|
| 708 |
+
"eval_samples_per_second": 770.215,
|
| 709 |
+
"eval_steps_per_second": 12.323,
|
| 710 |
+
"flow/cos_sim": 0.5558808352798223,
|
| 711 |
+
"flow/improvement_ratio": 0.9671246875077486,
|
| 712 |
+
"flow/mag_ratio_mean": 0.5348946927115321,
|
| 713 |
+
"flow/mag_ratio_std": 0.22368196118623018,
|
| 714 |
+
"step": 22528
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.24480038250059766,
|
| 718 |
+
"grad_norm": 0.22547593712806702,
|
| 719 |
+
"learning_rate": 0.0008854408194461756,
|
| 720 |
+
"loss": 1.5733323097229004,
|
| 721 |
+
"step": 23552
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.24480038250059766,
|
| 725 |
+
"eval_cos_loss": 0.44172694999724627,
|
| 726 |
+
"eval_loss": 1.5377833917737007,
|
| 727 |
+
"eval_mse_loss": 1.316919919103384,
|
| 728 |
+
"flow/cos_sim": 0.5582730043679476,
|
| 729 |
+
"flow/improvement_ratio": 0.9642701335251331,
|
| 730 |
+
"flow/mag_ratio_mean": 0.5346939843147993,
|
| 731 |
+
"flow/mag_ratio_std": 0.22327208751812577,
|
| 732 |
+
"step": 23552
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.24480038250059766,
|
| 736 |
+
"eval_cos_loss": 0.44172694999724627,
|
| 737 |
+
"eval_loss": 1.5377833917737007,
|
| 738 |
+
"eval_mse_loss": 1.316919919103384,
|
| 739 |
+
"eval_runtime": 3.1395,
|
| 740 |
+
"eval_samples_per_second": 637.04,
|
| 741 |
+
"eval_steps_per_second": 10.193,
|
| 742 |
+
"flow/cos_sim": 0.5582730043679476,
|
| 743 |
+
"flow/improvement_ratio": 0.9642701335251331,
|
| 744 |
+
"flow/mag_ratio_mean": 0.5346939843147993,
|
| 745 |
+
"flow/mag_ratio_std": 0.22327208751812577,
|
| 746 |
+
"step": 23552
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.255443877391928,
|
| 750 |
+
"grad_norm": 0.2300369143486023,
|
| 751 |
+
"learning_rate": 0.0008742123561119935,
|
| 752 |
+
"loss": 1.569944143295288,
|
| 753 |
+
"step": 24576
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.255443877391928,
|
| 757 |
+
"eval_cos_loss": 0.447942478582263,
|
| 758 |
+
"eval_loss": 1.553868442773819,
|
| 759 |
+
"eval_mse_loss": 1.3298972100019455,
|
| 760 |
+
"flow/cos_sim": 0.5520575055852532,
|
| 761 |
+
"flow/improvement_ratio": 0.9638102632015944,
|
| 762 |
+
"flow/mag_ratio_mean": 0.5306164929643273,
|
| 763 |
+
"flow/mag_ratio_std": 0.22182104969397187,
|
| 764 |
+
"step": 24576
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.255443877391928,
|
| 768 |
+
"eval_cos_loss": 0.447942478582263,
|
| 769 |
+
"eval_loss": 1.553868442773819,
|
| 770 |
+
"eval_mse_loss": 1.3298972100019455,
|
| 771 |
+
"eval_runtime": 2.5886,
|
| 772 |
+
"eval_samples_per_second": 772.612,
|
| 773 |
+
"eval_steps_per_second": 12.362,
|
| 774 |
+
"flow/cos_sim": 0.5520575055852532,
|
| 775 |
+
"flow/improvement_ratio": 0.9638102632015944,
|
| 776 |
+
"flow/mag_ratio_mean": 0.5306164929643273,
|
| 777 |
+
"flow/mag_ratio_std": 0.22182104969397187,
|
| 778 |
+
"step": 24576
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.26608737228325835,
|
| 782 |
+
"grad_norm": 0.2177908569574356,
|
| 783 |
+
"learning_rate": 0.0008625491011983832,
|
| 784 |
+
"loss": 1.5683772563934326,
|
| 785 |
+
"step": 25600
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 0.26608737228325835,
|
| 789 |
+
"eval_cos_loss": 0.45118876080960035,
|
| 790 |
+
"eval_loss": 1.5609249621629715,
|
| 791 |
+
"eval_mse_loss": 1.3353305757045746,
|
| 792 |
+
"flow/cos_sim": 0.5488111022859812,
|
| 793 |
+
"flow/improvement_ratio": 0.9652206618338823,
|
| 794 |
+
"flow/mag_ratio_mean": 0.5250881398096681,
|
| 795 |
+
"flow/mag_ratio_std": 0.22340481635183096,
|
| 796 |
+
"step": 25600
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.26608737228325835,
|
| 800 |
+
"eval_cos_loss": 0.45118876080960035,
|
| 801 |
+
"eval_loss": 1.5609249621629715,
|
| 802 |
+
"eval_mse_loss": 1.3353305757045746,
|
| 803 |
+
"eval_runtime": 2.5832,
|
| 804 |
+
"eval_samples_per_second": 774.241,
|
| 805 |
+
"eval_steps_per_second": 12.388,
|
| 806 |
+
"flow/cos_sim": 0.5488111022859812,
|
| 807 |
+
"flow/improvement_ratio": 0.9652206618338823,
|
| 808 |
+
"flow/mag_ratio_mean": 0.5250881398096681,
|
| 809 |
+
"flow/mag_ratio_std": 0.22340481635183096,
|
| 810 |
+
"step": 25600
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.27673086717458867,
|
| 814 |
+
"grad_norm": 0.13252000510692596,
|
| 815 |
+
"learning_rate": 0.0008504421682637403,
|
| 816 |
+
"loss": 1.5673582553863525,
|
| 817 |
+
"step": 26624
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.27673086717458867,
|
| 821 |
+
"eval_cos_loss": 0.44563145097345114,
|
| 822 |
+
"eval_loss": 1.5506689585745335,
|
| 823 |
+
"eval_mse_loss": 1.3278532326221466,
|
| 824 |
+
"flow/cos_sim": 0.5543686226010323,
|
| 825 |
+
"flow/improvement_ratio": 0.9666622839868069,
|
| 826 |
+
"flow/mag_ratio_mean": 0.5269411941990256,
|
| 827 |
+
"flow/mag_ratio_std": 0.21877468656748533,
|
| 828 |
+
"step": 26624
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.27673086717458867,
|
| 832 |
+
"eval_cos_loss": 0.44563145097345114,
|
| 833 |
+
"eval_loss": 1.5506689585745335,
|
| 834 |
+
"eval_mse_loss": 1.3278532326221466,
|
| 835 |
+
"eval_runtime": 2.617,
|
| 836 |
+
"eval_samples_per_second": 764.244,
|
| 837 |
+
"eval_steps_per_second": 12.228,
|
| 838 |
+
"flow/cos_sim": 0.5543686226010323,
|
| 839 |
+
"flow/improvement_ratio": 0.9666622839868069,
|
| 840 |
+
"flow/mag_ratio_mean": 0.5269411941990256,
|
| 841 |
+
"flow/mag_ratio_std": 0.21877468656748533,
|
| 842 |
+
"step": 26624
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.287374362065919,
|
| 846 |
+
"grad_norm": 0.2598721981048584,
|
| 847 |
+
"learning_rate": 0.0008379296157504366,
|
| 848 |
+
"loss": 1.564971923828125,
|
| 849 |
+
"step": 27648
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.287374362065919,
|
| 853 |
+
"eval_cos_loss": 0.4419550793245435,
|
| 854 |
+
"eval_loss": 1.5396056547760963,
|
| 855 |
+
"eval_mse_loss": 1.3186281062662601,
|
| 856 |
+
"flow/cos_sim": 0.5580449867993593,
|
| 857 |
+
"flow/improvement_ratio": 0.9673310127109289,
|
| 858 |
+
"flow/mag_ratio_mean": 0.5319117670878768,
|
| 859 |
+
"flow/mag_ratio_std": 0.22065124148502946,
|
| 860 |
+
"step": 27648
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.287374362065919,
|
| 864 |
+
"eval_cos_loss": 0.4419550793245435,
|
| 865 |
+
"eval_loss": 1.5396056547760963,
|
| 866 |
+
"eval_mse_loss": 1.3186281062662601,
|
| 867 |
+
"eval_runtime": 3.0156,
|
| 868 |
+
"eval_samples_per_second": 663.219,
|
| 869 |
+
"eval_steps_per_second": 10.612,
|
| 870 |
+
"flow/cos_sim": 0.5580449867993593,
|
| 871 |
+
"flow/improvement_ratio": 0.9673310127109289,
|
| 872 |
+
"flow/mag_ratio_mean": 0.5319117670878768,
|
| 873 |
+
"flow/mag_ratio_std": 0.22065124148502946,
|
| 874 |
+
"step": 27648
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.2980178569572493,
|
| 878 |
+
"grad_norm": 0.274239718914032,
|
| 879 |
+
"learning_rate": 0.0008250147265053921,
|
| 880 |
+
"loss": 1.564111590385437,
|
| 881 |
+
"step": 28672
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.2980178569572493,
|
| 885 |
+
"eval_cos_loss": 0.44205061066895723,
|
| 886 |
+
"eval_loss": 1.5401594452559948,
|
| 887 |
+
"eval_mse_loss": 1.319134145975113,
|
| 888 |
+
"flow/cos_sim": 0.5579493436962366,
|
| 889 |
+
"flow/improvement_ratio": 0.9671094436198473,
|
| 890 |
+
"flow/mag_ratio_mean": 0.5367719177156687,
|
| 891 |
+
"flow/mag_ratio_std": 0.2205441533587873,
|
| 892 |
+
"step": 28672
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.2980178569572493,
|
| 896 |
+
"eval_cos_loss": 0.44205061066895723,
|
| 897 |
+
"eval_loss": 1.5401594452559948,
|
| 898 |
+
"eval_mse_loss": 1.319134145975113,
|
| 899 |
+
"eval_runtime": 2.6525,
|
| 900 |
+
"eval_samples_per_second": 754.016,
|
| 901 |
+
"eval_steps_per_second": 12.064,
|
| 902 |
+
"flow/cos_sim": 0.5579493436962366,
|
| 903 |
+
"flow/improvement_ratio": 0.9671094436198473,
|
| 904 |
+
"flow/mag_ratio_mean": 0.5367719177156687,
|
| 905 |
+
"flow/mag_ratio_std": 0.2205441533587873,
|
| 906 |
+
"step": 28672
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.3086613518485797,
|
| 910 |
+
"grad_norm": 0.2095516473054886,
|
| 911 |
+
"learning_rate": 0.000811699689939724,
|
| 912 |
+
"loss": 1.5625946521759033,
|
| 913 |
+
"step": 29696
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.3086613518485797,
|
| 917 |
+
"eval_cos_loss": 0.44621053244918585,
|
| 918 |
+
"eval_loss": 1.5496392995119095,
|
| 919 |
+
"eval_mse_loss": 1.3265340402722359,
|
| 920 |
+
"flow/cos_sim": 0.5537894666194916,
|
| 921 |
+
"flow/improvement_ratio": 0.9614951889961958,
|
| 922 |
+
"flow/mag_ratio_mean": 0.5277672996744514,
|
| 923 |
+
"flow/mag_ratio_std": 0.22011788561940193,
|
| 924 |
+
"step": 29696
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.3086613518485797,
|
| 928 |
+
"eval_cos_loss": 0.44621053244918585,
|
| 929 |
+
"eval_loss": 1.5496392995119095,
|
| 930 |
+
"eval_mse_loss": 1.3265340402722359,
|
| 931 |
+
"eval_runtime": 2.5404,
|
| 932 |
+
"eval_samples_per_second": 787.269,
|
| 933 |
+
"eval_steps_per_second": 12.596,
|
| 934 |
+
"flow/cos_sim": 0.5537894666194916,
|
| 935 |
+
"flow/improvement_ratio": 0.9614951889961958,
|
| 936 |
+
"flow/mag_ratio_mean": 0.5277672996744514,
|
| 937 |
+
"flow/mag_ratio_std": 0.22011788561940193,
|
| 938 |
+
"step": 29696
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.31930484673991,
|
| 942 |
+
"grad_norm": 0.15808935463428497,
|
| 943 |
+
"learning_rate": 0.0007980128200054721,
|
| 944 |
+
"loss": 1.5646651983261108,
|
| 945 |
+
"step": 30720
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.31930484673991,
|
| 949 |
+
"eval_cos_loss": 0.44148214533925056,
|
| 950 |
+
"eval_loss": 1.5364415682852268,
|
| 951 |
+
"eval_mse_loss": 1.3157004974782467,
|
| 952 |
+
"flow/cos_sim": 0.5585179291665554,
|
| 953 |
+
"flow/improvement_ratio": 0.9648495689034462,
|
| 954 |
+
"flow/mag_ratio_mean": 0.537518884986639,
|
| 955 |
+
"flow/mag_ratio_std": 0.22350562876090407,
|
| 956 |
+
"step": 30720
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.31930484673991,
|
| 960 |
+
"eval_cos_loss": 0.44148214533925056,
|
| 961 |
+
"eval_loss": 1.5364415682852268,
|
| 962 |
+
"eval_mse_loss": 1.3157004974782467,
|
| 963 |
+
"eval_runtime": 2.5821,
|
| 964 |
+
"eval_samples_per_second": 774.564,
|
| 965 |
+
"eval_steps_per_second": 12.393,
|
| 966 |
+
"flow/cos_sim": 0.5585179291665554,
|
| 967 |
+
"flow/improvement_ratio": 0.9648495689034462,
|
| 968 |
+
"flow/mag_ratio_mean": 0.537518884986639,
|
| 969 |
+
"flow/mag_ratio_std": 0.22350562876090407,
|
| 970 |
+
"step": 30720
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.3299483416312403,
|
| 974 |
+
"grad_norm": 0.1921176314353943,
|
| 975 |
+
"learning_rate": 0.0007839843253324567,
|
| 976 |
+
"loss": 1.563474416732788,
|
| 977 |
+
"step": 31744
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.3299483416312403,
|
| 981 |
+
"eval_cos_loss": 0.4404078619554639,
|
| 982 |
+
"eval_loss": 1.534349039196968,
|
| 983 |
+
"eval_mse_loss": 1.3141451105475426,
|
| 984 |
+
"flow/cos_sim": 0.5595921669155359,
|
| 985 |
+
"flow/improvement_ratio": 0.9649890139698982,
|
| 986 |
+
"flow/mag_ratio_mean": 0.5285989735275507,
|
| 987 |
+
"flow/mag_ratio_std": 0.22230371600016952,
|
| 988 |
+
"step": 31744
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 0.3299483416312403,
|
| 992 |
+
"eval_cos_loss": 0.4404078619554639,
|
| 993 |
+
"eval_loss": 1.534349039196968,
|
| 994 |
+
"eval_mse_loss": 1.3141451105475426,
|
| 995 |
+
"eval_runtime": 2.5419,
|
| 996 |
+
"eval_samples_per_second": 786.808,
|
| 997 |
+
"eval_steps_per_second": 12.589,
|
| 998 |
+
"flow/cos_sim": 0.5595921669155359,
|
| 999 |
+
"flow/improvement_ratio": 0.9649890139698982,
|
| 1000 |
+
"flow/mag_ratio_mean": 0.5285989735275507,
|
| 1001 |
+
"flow/mag_ratio_std": 0.22230371600016952,
|
| 1002 |
+
"step": 31744
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.34059183652257063,
|
| 1006 |
+
"grad_norm": 0.19485324621200562,
|
| 1007 |
+
"learning_rate": 0.0007696035173607825,
|
| 1008 |
+
"loss": 1.5625982284545898,
|
| 1009 |
+
"step": 32768
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 0.34059183652257063,
|
| 1013 |
+
"eval_cos_loss": 0.4499282343313098,
|
| 1014 |
+
"eval_loss": 1.5567349456250668,
|
| 1015 |
+
"eval_mse_loss": 1.3317708261311054,
|
| 1016 |
+
"flow/cos_sim": 0.5500718057155609,
|
| 1017 |
+
"flow/improvement_ratio": 0.9645203202962875,
|
| 1018 |
+
"flow/mag_ratio_mean": 0.5316947773098946,
|
| 1019 |
+
"flow/mag_ratio_std": 0.22561145247891545,
|
| 1020 |
+
"step": 32768
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.34059183652257063,
|
| 1024 |
+
"eval_cos_loss": 0.4499282343313098,
|
| 1025 |
+
"eval_loss": 1.5567349456250668,
|
| 1026 |
+
"eval_mse_loss": 1.3317708261311054,
|
| 1027 |
+
"eval_runtime": 2.5397,
|
| 1028 |
+
"eval_samples_per_second": 787.504,
|
| 1029 |
+
"eval_steps_per_second": 12.6,
|
| 1030 |
+
"flow/cos_sim": 0.5500718057155609,
|
| 1031 |
+
"flow/improvement_ratio": 0.9645203202962875,
|
| 1032 |
+
"flow/mag_ratio_mean": 0.5316947773098946,
|
| 1033 |
+
"flow/mag_ratio_std": 0.22561145247891545,
|
| 1034 |
+
"step": 32768
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.351235331413901,
|
| 1038 |
+
"grad_norm": 0.18854600191116333,
|
| 1039 |
+
"learning_rate": 0.0007549156025151,
|
| 1040 |
+
"loss": 1.5604270696640015,
|
| 1041 |
+
"step": 33792
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.351235331413901,
|
| 1045 |
+
"eval_cos_loss": 0.44613189715892076,
|
| 1046 |
+
"eval_loss": 1.5455855540931225,
|
| 1047 |
+
"eval_mse_loss": 1.322519600391388,
|
| 1048 |
+
"flow/cos_sim": 0.5538681279867887,
|
| 1049 |
+
"flow/improvement_ratio": 0.9590773209929466,
|
| 1050 |
+
"flow/mag_ratio_mean": 0.5333189619705081,
|
| 1051 |
+
"flow/mag_ratio_std": 0.22768286149948835,
|
| 1052 |
+
"step": 33792
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.351235331413901,
|
| 1056 |
+
"eval_cos_loss": 0.44613189715892076,
|
| 1057 |
+
"eval_loss": 1.5455855540931225,
|
| 1058 |
+
"eval_mse_loss": 1.322519600391388,
|
| 1059 |
+
"eval_runtime": 2.5511,
|
| 1060 |
+
"eval_samples_per_second": 783.987,
|
| 1061 |
+
"eval_steps_per_second": 12.544,
|
| 1062 |
+
"flow/cos_sim": 0.5538681279867887,
|
| 1063 |
+
"flow/improvement_ratio": 0.9590773209929466,
|
| 1064 |
+
"flow/mag_ratio_mean": 0.5333189619705081,
|
| 1065 |
+
"flow/mag_ratio_std": 0.22768286149948835,
|
| 1066 |
+
"step": 33792
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.3618788263052313,
|
| 1070 |
+
"grad_norm": 0.1909618228673935,
|
| 1071 |
+
"learning_rate": 0.0007399093898115421,
|
| 1072 |
+
"loss": 1.559531807899475,
|
| 1073 |
+
"step": 34816
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.3618788263052313,
|
| 1077 |
+
"eval_cos_loss": 0.4420803328976035,
|
| 1078 |
+
"eval_loss": 1.5389960557222366,
|
| 1079 |
+
"eval_mse_loss": 1.3179558925330639,
|
| 1080 |
+
"flow/cos_sim": 0.5579197406768799,
|
| 1081 |
+
"flow/improvement_ratio": 0.959646550938487,
|
| 1082 |
+
"flow/mag_ratio_mean": 0.5326429791748524,
|
| 1083 |
+
"flow/mag_ratio_std": 0.22111017350107431,
|
| 1084 |
+
"step": 34816
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 0.3618788263052313,
|
| 1088 |
+
"eval_cos_loss": 0.4420803328976035,
|
| 1089 |
+
"eval_loss": 1.5389960557222366,
|
| 1090 |
+
"eval_mse_loss": 1.3179558925330639,
|
| 1091 |
+
"eval_runtime": 2.5395,
|
| 1092 |
+
"eval_samples_per_second": 787.558,
|
| 1093 |
+
"eval_steps_per_second": 12.601,
|
| 1094 |
+
"flow/cos_sim": 0.5579197406768799,
|
| 1095 |
+
"flow/improvement_ratio": 0.959646550938487,
|
| 1096 |
+
"flow/mag_ratio_mean": 0.5326429791748524,
|
| 1097 |
+
"flow/mag_ratio_std": 0.22111017350107431,
|
| 1098 |
+
"step": 34816
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.37252232119656165,
|
| 1102 |
+
"grad_norm": 0.1629696786403656,
|
| 1103 |
+
"learning_rate": 0.0007246320516499633,
|
| 1104 |
+
"loss": 1.5580956935882568,
|
| 1105 |
+
"step": 35840
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 0.37252232119656165,
|
| 1109 |
+
"eval_cos_loss": 0.44205798115581274,
|
| 1110 |
+
"eval_loss": 1.5348509810864925,
|
| 1111 |
+
"eval_mse_loss": 1.3138219900429249,
|
| 1112 |
+
"flow/cos_sim": 0.5579419694840908,
|
| 1113 |
+
"flow/improvement_ratio": 0.9621348176151514,
|
| 1114 |
+
"flow/mag_ratio_mean": 0.5318824276328087,
|
| 1115 |
+
"flow/mag_ratio_std": 0.226469362154603,
|
| 1116 |
+
"step": 35840
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.37252232119656165,
|
| 1120 |
+
"eval_cos_loss": 0.44205798115581274,
|
| 1121 |
+
"eval_loss": 1.5348509810864925,
|
| 1122 |
+
"eval_mse_loss": 1.3138219900429249,
|
| 1123 |
+
"eval_runtime": 2.557,
|
| 1124 |
+
"eval_samples_per_second": 782.177,
|
| 1125 |
+
"eval_steps_per_second": 12.515,
|
| 1126 |
+
"flow/cos_sim": 0.5579419694840908,
|
| 1127 |
+
"flow/improvement_ratio": 0.9621348176151514,
|
| 1128 |
+
"flow/mag_ratio_mean": 0.5318824276328087,
|
| 1129 |
+
"flow/mag_ratio_std": 0.226469362154603,
|
| 1130 |
+
"step": 35840
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.38316581608789196,
|
| 1134 |
+
"grad_norm": 0.2269536405801773,
|
| 1135 |
+
"learning_rate": 0.0007090719479543767,
|
| 1136 |
+
"loss": 1.557045578956604,
|
| 1137 |
+
"step": 36864
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.38316581608789196,
|
| 1141 |
+
"eval_cos_loss": 0.4411419341340661,
|
| 1142 |
+
"eval_loss": 1.5364714972674847,
|
| 1143 |
+
"eval_mse_loss": 1.315900530666113,
|
| 1144 |
+
"flow/cos_sim": 0.5588581711053848,
|
| 1145 |
+
"flow/improvement_ratio": 0.965403363108635,
|
| 1146 |
+
"flow/mag_ratio_mean": 0.5407936815172434,
|
| 1147 |
+
"flow/mag_ratio_std": 0.22326642088592052,
|
| 1148 |
+
"step": 36864
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.38316581608789196,
|
| 1152 |
+
"eval_cos_loss": 0.4411419341340661,
|
| 1153 |
+
"eval_loss": 1.5364714972674847,
|
| 1154 |
+
"eval_mse_loss": 1.315900530666113,
|
| 1155 |
+
"eval_runtime": 2.5319,
|
| 1156 |
+
"eval_samples_per_second": 789.932,
|
| 1157 |
+
"eval_steps_per_second": 12.639,
|
| 1158 |
+
"flow/cos_sim": 0.5588581711053848,
|
| 1159 |
+
"flow/improvement_ratio": 0.965403363108635,
|
| 1160 |
+
"flow/mag_ratio_mean": 0.5407936815172434,
|
| 1161 |
+
"flow/mag_ratio_std": 0.22326642088592052,
|
| 1162 |
+
"step": 36864
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.39380931097922234,
|
| 1166 |
+
"grad_norm": 0.18450024724006653,
|
| 1167 |
+
"learning_rate": 0.0006932779922946351,
|
| 1168 |
+
"loss": 1.5578693151474,
|
| 1169 |
+
"step": 37888
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 0.39380931097922234,
|
| 1173 |
+
"eval_cos_loss": 0.44131703954190016,
|
| 1174 |
+
"eval_loss": 1.5340029932558537,
|
| 1175 |
+
"eval_mse_loss": 1.3133444860577583,
|
| 1176 |
+
"flow/cos_sim": 0.5586829409003258,
|
| 1177 |
+
"flow/improvement_ratio": 0.9628860391676426,
|
| 1178 |
+
"flow/mag_ratio_mean": 0.537332147359848,
|
| 1179 |
+
"flow/mag_ratio_std": 0.22678001504391432,
|
| 1180 |
+
"step": 37888
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.39380931097922234,
|
| 1184 |
+
"eval_cos_loss": 0.44131703954190016,
|
| 1185 |
+
"eval_loss": 1.5340029932558537,
|
| 1186 |
+
"eval_mse_loss": 1.3133444860577583,
|
| 1187 |
+
"eval_runtime": 2.555,
|
| 1188 |
+
"eval_samples_per_second": 782.783,
|
| 1189 |
+
"eval_steps_per_second": 12.525,
|
| 1190 |
+
"flow/cos_sim": 0.5586829409003258,
|
| 1191 |
+
"flow/improvement_ratio": 0.9628860391676426,
|
| 1192 |
+
"flow/mag_ratio_mean": 0.537332147359848,
|
| 1193 |
+
"flow/mag_ratio_std": 0.22678001504391432,
|
| 1194 |
+
"step": 37888
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.40445280587055266,
|
| 1198 |
+
"grad_norm": 0.20654521882534027,
|
| 1199 |
+
"learning_rate": 0.0006772381509746807,
|
| 1200 |
+
"loss": 1.5568833351135254,
|
| 1201 |
+
"step": 38912
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.40445280587055266,
|
| 1205 |
+
"eval_cos_loss": 0.4440508605912328,
|
| 1206 |
+
"eval_loss": 1.540926594287157,
|
| 1207 |
+
"eval_mse_loss": 1.3189011700451374,
|
| 1208 |
+
"flow/cos_sim": 0.5559491030871868,
|
| 1209 |
+
"flow/improvement_ratio": 0.964597575366497,
|
| 1210 |
+
"flow/mag_ratio_mean": 0.5335862170904875,
|
| 1211 |
+
"flow/mag_ratio_std": 0.22821834543719888,
|
| 1212 |
+
"step": 38912
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 0.40445280587055266,
|
| 1216 |
+
"eval_cos_loss": 0.4440508605912328,
|
| 1217 |
+
"eval_loss": 1.540926594287157,
|
| 1218 |
+
"eval_mse_loss": 1.3189011700451374,
|
| 1219 |
+
"eval_runtime": 2.9265,
|
| 1220 |
+
"eval_samples_per_second": 683.411,
|
| 1221 |
+
"eval_steps_per_second": 10.935,
|
| 1222 |
+
"flow/cos_sim": 0.5559491030871868,
|
| 1223 |
+
"flow/improvement_ratio": 0.964597575366497,
|
| 1224 |
+
"flow/mag_ratio_mean": 0.5335862170904875,
|
| 1225 |
+
"flow/mag_ratio_std": 0.22821834543719888,
|
| 1226 |
+
"step": 38912
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.415096300761883,
|
| 1230 |
+
"grad_norm": 0.20580987632274628,
|
| 1231 |
+
"learning_rate": 0.0006609868783930164,
|
| 1232 |
+
"loss": 1.5565650463104248,
|
| 1233 |
+
"step": 39936
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.415096300761883,
|
| 1237 |
+
"eval_cos_loss": 0.4446534486487508,
|
| 1238 |
+
"eval_loss": 1.5423276983201504,
|
| 1239 |
+
"eval_mse_loss": 1.3200009688735008,
|
| 1240 |
+
"flow/cos_sim": 0.5553465932607651,
|
| 1241 |
+
"flow/improvement_ratio": 0.9601697400212288,
|
| 1242 |
+
"flow/mag_ratio_mean": 0.5308061949908733,
|
| 1243 |
+
"flow/mag_ratio_std": 0.22724535362794995,
|
| 1244 |
+
"step": 39936
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.415096300761883,
|
| 1248 |
+
"eval_cos_loss": 0.4446534486487508,
|
| 1249 |
+
"eval_loss": 1.5423276983201504,
|
| 1250 |
+
"eval_mse_loss": 1.3200009688735008,
|
| 1251 |
+
"eval_runtime": 2.5315,
|
| 1252 |
+
"eval_samples_per_second": 790.058,
|
| 1253 |
+
"eval_steps_per_second": 12.641,
|
| 1254 |
+
"flow/cos_sim": 0.5553465932607651,
|
| 1255 |
+
"flow/improvement_ratio": 0.9601697400212288,
|
| 1256 |
+
"flow/mag_ratio_mean": 0.5308061949908733,
|
| 1257 |
+
"flow/mag_ratio_std": 0.22724535362794995,
|
| 1258 |
+
"step": 39936
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.4257397956532133,
|
| 1262 |
+
"grad_norm": 0.17006264626979828,
|
| 1263 |
+
"learning_rate": 0.0006445597062966236,
|
| 1264 |
+
"loss": 1.5565887689590454,
|
| 1265 |
+
"step": 40960
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 0.4257397956532133,
|
| 1269 |
+
"eval_cos_loss": 0.4380533881485462,
|
| 1270 |
+
"eval_loss": 1.5283529199659824,
|
| 1271 |
+
"eval_mse_loss": 1.3093262203037739,
|
| 1272 |
+
"flow/cos_sim": 0.5619466044008732,
|
| 1273 |
+
"flow/improvement_ratio": 0.9661596808582544,
|
| 1274 |
+
"flow/mag_ratio_mean": 0.5353248585015535,
|
| 1275 |
+
"flow/mag_ratio_std": 0.22140436619520187,
|
| 1276 |
+
"step": 40960
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.4257397956532133,
|
| 1280 |
+
"eval_cos_loss": 0.4380533881485462,
|
| 1281 |
+
"eval_loss": 1.5283529199659824,
|
| 1282 |
+
"eval_mse_loss": 1.3093262203037739,
|
| 1283 |
+
"eval_runtime": 2.6667,
|
| 1284 |
+
"eval_samples_per_second": 749.988,
|
| 1285 |
+
"eval_steps_per_second": 12.0,
|
| 1286 |
+
"flow/cos_sim": 0.5619466044008732,
|
| 1287 |
+
"flow/improvement_ratio": 0.9661596808582544,
|
| 1288 |
+
"flow/mag_ratio_mean": 0.5353248585015535,
|
| 1289 |
+
"flow/mag_ratio_std": 0.22140436619520187,
|
| 1290 |
+
"step": 40960
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.43638329054454367,
|
| 1294 |
+
"grad_norm": 0.2264794260263443,
|
| 1295 |
+
"learning_rate": 0.0006279604223844502,
|
| 1296 |
+
"loss": 1.5556617975234985,
|
| 1297 |
+
"step": 41984
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.43638329054454367,
|
| 1301 |
+
"eval_cos_loss": 0.4411089513450861,
|
| 1302 |
+
"eval_loss": 1.5351563543081284,
|
| 1303 |
+
"eval_mse_loss": 1.3146018758416176,
|
| 1304 |
+
"flow/cos_sim": 0.5588910467922688,
|
| 1305 |
+
"flow/improvement_ratio": 0.9700996112078428,
|
| 1306 |
+
"flow/mag_ratio_mean": 0.535472328774631,
|
| 1307 |
+
"flow/mag_ratio_std": 0.2249652906320989,
|
| 1308 |
+
"step": 41984
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 0.43638329054454367,
|
| 1312 |
+
"eval_cos_loss": 0.4411089513450861,
|
| 1313 |
+
"eval_loss": 1.5351563543081284,
|
| 1314 |
+
"eval_mse_loss": 1.3146018758416176,
|
| 1315 |
+
"eval_runtime": 2.6645,
|
| 1316 |
+
"eval_samples_per_second": 750.607,
|
| 1317 |
+
"eval_steps_per_second": 12.01,
|
| 1318 |
+
"flow/cos_sim": 0.5588910467922688,
|
| 1319 |
+
"flow/improvement_ratio": 0.9700996112078428,
|
| 1320 |
+
"flow/mag_ratio_mean": 0.535472328774631,
|
| 1321 |
+
"flow/mag_ratio_std": 0.2249652906320989,
|
| 1322 |
+
"step": 41984
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.447026785435874,
|
| 1326 |
+
"grad_norm": 0.27407148480415344,
|
| 1327 |
+
"learning_rate": 0.0006111923466049098,
|
| 1328 |
+
"loss": 1.5525274276733398,
|
| 1329 |
+
"step": 43008
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.447026785435874,
|
| 1333 |
+
"eval_cos_loss": 0.43889701180160046,
|
| 1334 |
+
"eval_loss": 1.5268253944814205,
|
| 1335 |
+
"eval_mse_loss": 1.3073768950998783,
|
| 1336 |
+
"flow/cos_sim": 0.561102925799787,
|
| 1337 |
+
"flow/improvement_ratio": 0.9647987205535173,
|
| 1338 |
+
"flow/mag_ratio_mean": 0.5332341426983476,
|
| 1339 |
+
"flow/mag_ratio_std": 0.22829985432326794,
|
| 1340 |
+
"step": 43008
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.447026785435874,
|
| 1344 |
+
"eval_cos_loss": 0.43889701180160046,
|
| 1345 |
+
"eval_loss": 1.5268253944814205,
|
| 1346 |
+
"eval_mse_loss": 1.3073768950998783,
|
| 1347 |
+
"eval_runtime": 2.7939,
|
| 1348 |
+
"eval_samples_per_second": 715.846,
|
| 1349 |
+
"eval_steps_per_second": 11.454,
|
| 1350 |
+
"flow/cos_sim": 0.561102925799787,
|
| 1351 |
+
"flow/improvement_ratio": 0.9647987205535173,
|
| 1352 |
+
"flow/mag_ratio_mean": 0.5332341426983476,
|
| 1353 |
+
"flow/mag_ratio_std": 0.22829985432326794,
|
| 1354 |
+
"step": 43008
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.4576702803272043,
|
| 1358 |
+
"grad_norm": 0.1399686485528946,
|
| 1359 |
+
"learning_rate": 0.0005942916270463306,
|
| 1360 |
+
"loss": 1.5545454025268555,
|
| 1361 |
+
"step": 44032
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.4576702803272043,
|
| 1365 |
+
"eval_cos_loss": 0.44924431946128607,
|
| 1366 |
+
"eval_loss": 1.553151711821556,
|
| 1367 |
+
"eval_mse_loss": 1.3285295516252518,
|
| 1368 |
+
"flow/cos_sim": 0.550755743868649,
|
| 1369 |
+
"flow/improvement_ratio": 0.9646315854042768,
|
| 1370 |
+
"flow/mag_ratio_mean": 0.5285015730187297,
|
| 1371 |
+
"flow/mag_ratio_std": 0.22717531491070986,
|
| 1372 |
+
"step": 44032
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 0.4576702803272043,
|
| 1376 |
+
"eval_cos_loss": 0.44924431946128607,
|
| 1377 |
+
"eval_loss": 1.553151711821556,
|
| 1378 |
+
"eval_mse_loss": 1.3285295516252518,
|
| 1379 |
+
"eval_runtime": 2.6015,
|
| 1380 |
+
"eval_samples_per_second": 768.789,
|
| 1381 |
+
"eval_steps_per_second": 12.301,
|
| 1382 |
+
"flow/cos_sim": 0.550755743868649,
|
| 1383 |
+
"flow/improvement_ratio": 0.9646315854042768,
|
| 1384 |
+
"flow/mag_ratio_mean": 0.5285015730187297,
|
| 1385 |
+
"flow/mag_ratio_std": 0.22717531491070986,
|
| 1386 |
+
"step": 44032
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 0.4683137752185346,
|
| 1390 |
+
"grad_norm": 0.17331954836845398,
|
| 1391 |
+
"learning_rate": 0.0005772784249462365,
|
| 1392 |
+
"loss": 1.5531715154647827,
|
| 1393 |
+
"step": 45056
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 0.4683137752185346,
|
| 1397 |
+
"eval_cos_loss": 0.44831305276602507,
|
| 1398 |
+
"eval_loss": 1.55314514413476,
|
| 1399 |
+
"eval_mse_loss": 1.328988615423441,
|
| 1400 |
+
"flow/cos_sim": 0.5516869705170393,
|
| 1401 |
+
"flow/improvement_ratio": 0.9640028644353151,
|
| 1402 |
+
"flow/mag_ratio_mean": 0.53109060972929,
|
| 1403 |
+
"flow/mag_ratio_std": 0.22524388320744038,
|
| 1404 |
+
"step": 45056
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.4683137752185346,
|
| 1408 |
+
"eval_cos_loss": 0.44831305276602507,
|
| 1409 |
+
"eval_loss": 1.55314514413476,
|
| 1410 |
+
"eval_mse_loss": 1.328988615423441,
|
| 1411 |
+
"eval_runtime": 2.6383,
|
| 1412 |
+
"eval_samples_per_second": 758.058,
|
| 1413 |
+
"eval_steps_per_second": 12.129,
|
| 1414 |
+
"flow/cos_sim": 0.5516869705170393,
|
| 1415 |
+
"flow/improvement_ratio": 0.9640028644353151,
|
| 1416 |
+
"flow/mag_ratio_mean": 0.53109060972929,
|
| 1417 |
+
"flow/mag_ratio_std": 0.22524388320744038,
|
| 1418 |
+
"step": 45056
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.478957270109865,
|
| 1422 |
+
"grad_norm": 0.20901691913604736,
|
| 1423 |
+
"learning_rate": 0.0005601730357250316,
|
| 1424 |
+
"loss": 1.5534908771514893,
|
| 1425 |
+
"step": 46080
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.478957270109865,
|
| 1429 |
+
"eval_cos_loss": 0.4380967328324914,
|
| 1430 |
+
"eval_loss": 1.5263510905206203,
|
| 1431 |
+
"eval_mse_loss": 1.3073027282953262,
|
| 1432 |
+
"flow/cos_sim": 0.5619033649563789,
|
| 1433 |
+
"flow/improvement_ratio": 0.9646317362785339,
|
| 1434 |
+
"flow/mag_ratio_mean": 0.5367344031110406,
|
| 1435 |
+
"flow/mag_ratio_std": 0.22356789046898484,
|
| 1436 |
+
"step": 46080
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.478957270109865,
|
| 1440 |
+
"eval_cos_loss": 0.4380967328324914,
|
| 1441 |
+
"eval_loss": 1.5263510905206203,
|
| 1442 |
+
"eval_mse_loss": 1.3073027282953262,
|
| 1443 |
+
"eval_runtime": 3.1077,
|
| 1444 |
+
"eval_samples_per_second": 643.555,
|
| 1445 |
+
"eval_steps_per_second": 10.297,
|
| 1446 |
+
"flow/cos_sim": 0.5619033649563789,
|
| 1447 |
+
"flow/improvement_ratio": 0.9646317362785339,
|
| 1448 |
+
"flow/mag_ratio_mean": 0.5367344031110406,
|
| 1449 |
+
"flow/mag_ratio_std": 0.22356789046898484,
|
| 1450 |
+
"step": 46080
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.4896007650011953,
|
| 1454 |
+
"grad_norm": 0.18614411354064941,
|
| 1455 |
+
"learning_rate": 0.0005430126677168879,
|
| 1456 |
+
"loss": 1.5517550706863403,
|
| 1457 |
+
"step": 47104
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.4896007650011953,
|
| 1461 |
+
"eval_cos_loss": 0.4381159236654639,
|
| 1462 |
+
"eval_loss": 1.5279735252261162,
|
| 1463 |
+
"eval_mse_loss": 1.308915562927723,
|
| 1464 |
+
"flow/cos_sim": 0.5618840865790844,
|
| 1465 |
+
"flow/improvement_ratio": 0.9676383044570684,
|
| 1466 |
+
"flow/mag_ratio_mean": 0.5347359916195273,
|
| 1467 |
+
"flow/mag_ratio_std": 0.22025129199028015,
|
| 1468 |
+
"step": 47104
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 0.4896007650011953,
|
| 1472 |
+
"eval_cos_loss": 0.4381159236654639,
|
| 1473 |
+
"eval_loss": 1.5279735252261162,
|
| 1474 |
+
"eval_mse_loss": 1.308915562927723,
|
| 1475 |
+
"eval_runtime": 2.5963,
|
| 1476 |
+
"eval_samples_per_second": 770.338,
|
| 1477 |
+
"eval_steps_per_second": 12.325,
|
| 1478 |
+
"flow/cos_sim": 0.5618840865790844,
|
| 1479 |
+
"flow/improvement_ratio": 0.9676383044570684,
|
| 1480 |
+
"flow/mag_ratio_mean": 0.5347359916195273,
|
| 1481 |
+
"flow/mag_ratio_std": 0.22025129199028015,
|
| 1482 |
+
"step": 47104
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 0.5002442598925256,
|
| 1486 |
+
"grad_norm": 0.20864352583885193,
|
| 1487 |
+
"learning_rate": 0.0005257842461318475,
|
| 1488 |
+
"loss": 1.5500738620758057,
|
| 1489 |
+
"step": 48128
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 0.5002442598925256,
|
| 1493 |
+
"eval_cos_loss": 0.4427802488207817,
|
| 1494 |
+
"eval_loss": 1.5378683991730213,
|
| 1495 |
+
"eval_mse_loss": 1.3164782784879208,
|
| 1496 |
+
"flow/cos_sim": 0.5572197437286377,
|
| 1497 |
+
"flow/improvement_ratio": 0.9660468604415655,
|
| 1498 |
+
"flow/mag_ratio_mean": 0.5358876623213291,
|
| 1499 |
+
"flow/mag_ratio_std": 0.22517190361395478,
|
| 1500 |
+
"step": 48128
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.5002442598925256,
|
| 1504 |
+
"eval_cos_loss": 0.4427802488207817,
|
| 1505 |
+
"eval_loss": 1.5378683991730213,
|
| 1506 |
+
"eval_mse_loss": 1.3164782784879208,
|
| 1507 |
+
"eval_runtime": 2.7741,
|
| 1508 |
+
"eval_samples_per_second": 720.965,
|
| 1509 |
+
"eval_steps_per_second": 11.535,
|
| 1510 |
+
"flow/cos_sim": 0.5572197437286377,
|
| 1511 |
+
"flow/improvement_ratio": 0.9660468604415655,
|
| 1512 |
+
"flow/mag_ratio_mean": 0.5358876623213291,
|
| 1513 |
+
"flow/mag_ratio_std": 0.22517190361395478,
|
| 1514 |
+
"step": 48128
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.510887754783856,
|
| 1518 |
+
"grad_norm": 0.18199937045574188,
|
| 1519 |
+
"learning_rate": 0.0005085250659563913,
|
| 1520 |
+
"loss": 1.5504491329193115,
|
| 1521 |
+
"step": 49152
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.510887754783856,
|
| 1525 |
+
"eval_cos_loss": 0.4387433025985956,
|
| 1526 |
+
"eval_loss": 1.5289665646851063,
|
| 1527 |
+
"eval_mse_loss": 1.309594914317131,
|
| 1528 |
+
"flow/cos_sim": 0.561256805434823,
|
| 1529 |
+
"flow/improvement_ratio": 0.9640381913632154,
|
| 1530 |
+
"flow/mag_ratio_mean": 0.5361321400851011,
|
| 1531 |
+
"flow/mag_ratio_std": 0.22544911736622453,
|
| 1532 |
+
"step": 49152
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 0.510887754783856,
|
| 1536 |
+
"eval_cos_loss": 0.4387433025985956,
|
| 1537 |
+
"eval_loss": 1.5289665646851063,
|
| 1538 |
+
"eval_mse_loss": 1.309594914317131,
|
| 1539 |
+
"eval_runtime": 2.5843,
|
| 1540 |
+
"eval_samples_per_second": 773.896,
|
| 1541 |
+
"eval_steps_per_second": 12.382,
|
| 1542 |
+
"flow/cos_sim": 0.561256805434823,
|
| 1543 |
+
"flow/improvement_ratio": 0.9640381913632154,
|
| 1544 |
+
"flow/mag_ratio_mean": 0.5361321400851011,
|
| 1545 |
+
"flow/mag_ratio_std": 0.22544911736622453,
|
| 1546 |
+
"step": 49152
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 0.5215312496751863,
|
| 1550 |
+
"grad_norm": 0.21871572732925415,
|
| 1551 |
+
"learning_rate": 0.0004912557160435426,
|
| 1552 |
+
"loss": 1.551537036895752,
|
| 1553 |
+
"step": 50176
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 0.5215312496751863,
|
| 1557 |
+
"eval_cos_loss": 0.4424938661977649,
|
| 1558 |
+
"eval_loss": 1.5380571633577347,
|
| 1559 |
+
"eval_mse_loss": 1.316810242831707,
|
| 1560 |
+
"flow/cos_sim": 0.5575060974806547,
|
| 1561 |
+
"flow/improvement_ratio": 0.9636888317763805,
|
| 1562 |
+
"flow/mag_ratio_mean": 0.5369405504316092,
|
| 1563 |
+
"flow/mag_ratio_std": 0.22526462702080607,
|
| 1564 |
+
"step": 50176
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.5215312496751863,
|
| 1568 |
+
"eval_cos_loss": 0.4424938661977649,
|
| 1569 |
+
"eval_loss": 1.5380571633577347,
|
| 1570 |
+
"eval_mse_loss": 1.316810242831707,
|
| 1571 |
+
"eval_runtime": 2.6421,
|
| 1572 |
+
"eval_samples_per_second": 756.983,
|
| 1573 |
+
"eval_steps_per_second": 12.112,
|
| 1574 |
+
"flow/cos_sim": 0.5575060974806547,
|
| 1575 |
+
"flow/improvement_ratio": 0.9636888317763805,
|
| 1576 |
+
"flow/mag_ratio_mean": 0.5369405504316092,
|
| 1577 |
+
"flow/mag_ratio_std": 0.22526462702080607,
|
| 1578 |
+
"step": 50176
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.5321747445665167,
|
| 1582 |
+
"grad_norm": 0.19360916316509247,
|
| 1583 |
+
"learning_rate": 0.000474013640007982,
|
| 1584 |
+
"loss": 1.550221562385559,
|
| 1585 |
+
"step": 51200
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.5321747445665167,
|
| 1589 |
+
"eval_cos_loss": 0.44469246733933687,
|
| 1590 |
+
"eval_loss": 1.5433855392038822,
|
| 1591 |
+
"eval_mse_loss": 1.3210392966866493,
|
| 1592 |
+
"flow/cos_sim": 0.5553075838834047,
|
| 1593 |
+
"flow/improvement_ratio": 0.9643128626048565,
|
| 1594 |
+
"flow/mag_ratio_mean": 0.5328638143837452,
|
| 1595 |
+
"flow/mag_ratio_std": 0.2267028819769621,
|
| 1596 |
+
"step": 51200
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 0.5321747445665167,
|
| 1600 |
+
"eval_cos_loss": 0.44469246733933687,
|
| 1601 |
+
"eval_loss": 1.5433855392038822,
|
| 1602 |
+
"eval_mse_loss": 1.3210392966866493,
|
| 1603 |
+
"eval_runtime": 2.5901,
|
| 1604 |
+
"eval_samples_per_second": 772.182,
|
| 1605 |
+
"eval_steps_per_second": 12.355,
|
| 1606 |
+
"flow/cos_sim": 0.5553075838834047,
|
| 1607 |
+
"flow/improvement_ratio": 0.9643128626048565,
|
| 1608 |
+
"flow/mag_ratio_mean": 0.5328638143837452,
|
| 1609 |
+
"flow/mag_ratio_std": 0.2267028819769621,
|
| 1610 |
+
"step": 51200
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.542818239457847,
|
| 1614 |
+
"grad_norm": 0.25140267610549927,
|
| 1615 |
+
"learning_rate": 0.0004567857008049507,
|
| 1616 |
+
"loss": 1.5501980781555176,
|
| 1617 |
+
"step": 52224
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.542818239457847,
|
| 1621 |
+
"eval_cos_loss": 0.44003486074507236,
|
| 1622 |
+
"eval_loss": 1.5340061485767365,
|
| 1623 |
+
"eval_mse_loss": 1.3139887191355228,
|
| 1624 |
+
"flow/cos_sim": 0.5599651224911213,
|
| 1625 |
+
"flow/improvement_ratio": 0.9662698730826378,
|
| 1626 |
+
"flow/mag_ratio_mean": 0.5311327101662755,
|
| 1627 |
+
"flow/mag_ratio_std": 0.21956392657011747,
|
| 1628 |
+
"step": 52224
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.542818239457847,
|
| 1632 |
+
"eval_cos_loss": 0.44003486074507236,
|
| 1633 |
+
"eval_loss": 1.5340061485767365,
|
| 1634 |
+
"eval_mse_loss": 1.3139887191355228,
|
| 1635 |
+
"eval_runtime": 2.5514,
|
| 1636 |
+
"eval_samples_per_second": 783.875,
|
| 1637 |
+
"eval_steps_per_second": 12.542,
|
| 1638 |
+
"flow/cos_sim": 0.5599651224911213,
|
| 1639 |
+
"flow/improvement_ratio": 0.9662698730826378,
|
| 1640 |
+
"flow/mag_ratio_mean": 0.5311327101662755,
|
| 1641 |
+
"flow/mag_ratio_std": 0.21956392657011747,
|
| 1642 |
+
"step": 52224
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.5534617343491773,
|
| 1646 |
+
"grad_norm": 0.1874593198299408,
|
| 1647 |
+
"learning_rate": 0.0004396260548863663,
|
| 1648 |
+
"loss": 1.5494704246520996,
|
| 1649 |
+
"step": 53248
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 0.5534617343491773,
|
| 1653 |
+
"eval_cos_loss": 0.4375583464279771,
|
| 1654 |
+
"eval_loss": 1.5273119732737541,
|
| 1655 |
+
"eval_mse_loss": 1.3085327930748463,
|
| 1656 |
+
"flow/cos_sim": 0.5624416321516037,
|
| 1657 |
+
"flow/improvement_ratio": 0.9654307011514902,
|
| 1658 |
+
"flow/mag_ratio_mean": 0.5390398278832436,
|
| 1659 |
+
"flow/mag_ratio_std": 0.2204155451618135,
|
| 1660 |
+
"step": 53248
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 0.5534617343491773,
|
| 1664 |
+
"eval_cos_loss": 0.4375583464279771,
|
| 1665 |
+
"eval_loss": 1.5273119732737541,
|
| 1666 |
+
"eval_mse_loss": 1.3085327930748463,
|
| 1667 |
+
"eval_runtime": 2.9751,
|
| 1668 |
+
"eval_samples_per_second": 672.236,
|
| 1669 |
+
"eval_steps_per_second": 10.756,
|
| 1670 |
+
"flow/cos_sim": 0.5624416321516037,
|
| 1671 |
+
"flow/improvement_ratio": 0.9654307011514902,
|
| 1672 |
+
"flow/mag_ratio_mean": 0.5390398278832436,
|
| 1673 |
+
"flow/mag_ratio_std": 0.2204155451618135,
|
| 1674 |
+
"step": 53248
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.5641052292405077,
|
| 1678 |
+
"grad_norm": 0.260960191488266,
|
| 1679 |
+
"learning_rate": 0.000422521628012444,
|
| 1680 |
+
"loss": 1.5488193035125732,
|
| 1681 |
+
"step": 54272
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 0.5641052292405077,
|
| 1685 |
+
"eval_cos_loss": 0.4423528155311942,
|
| 1686 |
+
"eval_loss": 1.5401365533471107,
|
| 1687 |
+
"eval_mse_loss": 1.318960152566433,
|
| 1688 |
+
"flow/cos_sim": 0.5576471537351608,
|
| 1689 |
+
"flow/improvement_ratio": 0.9645018931478262,
|
| 1690 |
+
"flow/mag_ratio_mean": 0.5321623589843512,
|
| 1691 |
+
"flow/mag_ratio_std": 0.22070467984303832,
|
| 1692 |
+
"step": 54272
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 0.5641052292405077,
|
| 1696 |
+
"eval_cos_loss": 0.4423528155311942,
|
| 1697 |
+
"eval_loss": 1.5401365533471107,
|
| 1698 |
+
"eval_mse_loss": 1.318960152566433,
|
| 1699 |
+
"eval_runtime": 2.7006,
|
| 1700 |
+
"eval_samples_per_second": 740.582,
|
| 1701 |
+
"eval_steps_per_second": 11.849,
|
| 1702 |
+
"flow/cos_sim": 0.5576471537351608,
|
| 1703 |
+
"flow/improvement_ratio": 0.9645018931478262,
|
| 1704 |
+
"flow/mag_ratio_mean": 0.5321623589843512,
|
| 1705 |
+
"flow/mag_ratio_std": 0.22070467984303832,
|
| 1706 |
+
"step": 54272
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.574748724131838,
|
| 1710 |
+
"grad_norm": 0.18275974690914154,
|
| 1711 |
+
"learning_rate": 0.00040552618837104806,
|
| 1712 |
+
"loss": 1.5481247901916504,
|
| 1713 |
+
"step": 55296
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 0.574748724131838,
|
| 1717 |
+
"eval_cos_loss": 0.4375903755426407,
|
| 1718 |
+
"eval_loss": 1.52578229829669,
|
| 1719 |
+
"eval_mse_loss": 1.3069871068000793,
|
| 1720 |
+
"flow/cos_sim": 0.5624096170067787,
|
| 1721 |
+
"flow/improvement_ratio": 0.9644604399800301,
|
| 1722 |
+
"flow/mag_ratio_mean": 0.5396961104124784,
|
| 1723 |
+
"flow/mag_ratio_std": 0.22439152654260397,
|
| 1724 |
+
"step": 55296
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.574748724131838,
|
| 1728 |
+
"eval_cos_loss": 0.4375903755426407,
|
| 1729 |
+
"eval_loss": 1.52578229829669,
|
| 1730 |
+
"eval_mse_loss": 1.3069871068000793,
|
| 1731 |
+
"eval_runtime": 2.5932,
|
| 1732 |
+
"eval_samples_per_second": 771.259,
|
| 1733 |
+
"eval_steps_per_second": 12.34,
|
| 1734 |
+
"flow/cos_sim": 0.5624096170067787,
|
| 1735 |
+
"flow/improvement_ratio": 0.9644604399800301,
|
| 1736 |
+
"flow/mag_ratio_mean": 0.5396961104124784,
|
| 1737 |
+
"flow/mag_ratio_std": 0.22439152654260397,
|
| 1738 |
+
"step": 55296
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.5853922190231683,
|
| 1742 |
+
"grad_norm": 0.19556212425231934,
|
| 1743 |
+
"learning_rate": 0.00038862678683408014,
|
| 1744 |
+
"loss": 1.548845648765564,
|
| 1745 |
+
"step": 56320
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.5853922190231683,
|
| 1749 |
+
"eval_cos_loss": 0.44385355431586504,
|
| 1750 |
+
"eval_loss": 1.5408159419894218,
|
| 1751 |
+
"eval_mse_loss": 1.3188891597092152,
|
| 1752 |
+
"flow/cos_sim": 0.5561465304344893,
|
| 1753 |
+
"flow/improvement_ratio": 0.9655895195901394,
|
| 1754 |
+
"flow/mag_ratio_mean": 0.5379520216956735,
|
| 1755 |
+
"flow/mag_ratio_std": 0.22702415706589818,
|
| 1756 |
+
"step": 56320
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 0.5853922190231683,
|
| 1760 |
+
"eval_cos_loss": 0.44385355431586504,
|
| 1761 |
+
"eval_loss": 1.5408159419894218,
|
| 1762 |
+
"eval_mse_loss": 1.3188891597092152,
|
| 1763 |
+
"eval_runtime": 2.811,
|
| 1764 |
+
"eval_samples_per_second": 711.502,
|
| 1765 |
+
"eval_steps_per_second": 11.384,
|
| 1766 |
+
"flow/cos_sim": 0.5561465304344893,
|
| 1767 |
+
"flow/improvement_ratio": 0.9655895195901394,
|
| 1768 |
+
"flow/mag_ratio_mean": 0.5379520216956735,
|
| 1769 |
+
"flow/mag_ratio_std": 0.22702415706589818,
|
| 1770 |
+
"step": 56320
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 0.5960357139144986,
|
| 1774 |
+
"grad_norm": 0.154473677277565,
|
| 1775 |
+
"learning_rate": 0.00037187654708719937,
|
| 1776 |
+
"loss": 1.5476142168045044,
|
| 1777 |
+
"step": 57344
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 0.5960357139144986,
|
| 1781 |
+
"eval_cos_loss": 0.44051590468734503,
|
| 1782 |
+
"eval_loss": 1.5316696986556053,
|
| 1783 |
+
"eval_mse_loss": 1.3114117458462715,
|
| 1784 |
+
"flow/cos_sim": 0.5594841903075576,
|
| 1785 |
+
"flow/improvement_ratio": 0.9659801628440619,
|
| 1786 |
+
"flow/mag_ratio_mean": 0.5374981416389346,
|
| 1787 |
+
"flow/mag_ratio_std": 0.2266282932832837,
|
| 1788 |
+
"step": 57344
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.5960357139144986,
|
| 1792 |
+
"eval_cos_loss": 0.44051590468734503,
|
| 1793 |
+
"eval_loss": 1.5316696986556053,
|
| 1794 |
+
"eval_mse_loss": 1.3114117458462715,
|
| 1795 |
+
"eval_runtime": 2.9501,
|
| 1796 |
+
"eval_samples_per_second": 677.94,
|
| 1797 |
+
"eval_steps_per_second": 10.847,
|
| 1798 |
+
"flow/cos_sim": 0.5594841903075576,
|
| 1799 |
+
"flow/improvement_ratio": 0.9659801628440619,
|
| 1800 |
+
"flow/mag_ratio_mean": 0.5374981416389346,
|
| 1801 |
+
"flow/mag_ratio_std": 0.2266282932832837,
|
| 1802 |
+
"step": 57344
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.6066792088058289,
|
| 1806 |
+
"grad_norm": 0.19920258224010468,
|
| 1807 |
+
"learning_rate": 0.00035526270682447326,
|
| 1808 |
+
"loss": 1.546189785003662,
|
| 1809 |
+
"step": 58368
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.6066792088058289,
|
| 1813 |
+
"eval_cos_loss": 0.45052160415798426,
|
| 1814 |
+
"eval_loss": 1.5585628859698772,
|
| 1815 |
+
"eval_mse_loss": 1.333302080631256,
|
| 1816 |
+
"flow/cos_sim": 0.5494783949106932,
|
| 1817 |
+
"flow/improvement_ratio": 0.9626638870686293,
|
| 1818 |
+
"flow/mag_ratio_mean": 0.5275079058483243,
|
| 1819 |
+
"flow/mag_ratio_std": 0.22685475973412395,
|
| 1820 |
+
"step": 58368
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.6066792088058289,
|
| 1824 |
+
"eval_cos_loss": 0.45052160415798426,
|
| 1825 |
+
"eval_loss": 1.5585628859698772,
|
| 1826 |
+
"eval_mse_loss": 1.333302080631256,
|
| 1827 |
+
"eval_runtime": 2.5748,
|
| 1828 |
+
"eval_samples_per_second": 776.758,
|
| 1829 |
+
"eval_steps_per_second": 12.428,
|
| 1830 |
+
"flow/cos_sim": 0.5494783949106932,
|
| 1831 |
+
"flow/improvement_ratio": 0.9626638870686293,
|
| 1832 |
+
"flow/mag_ratio_mean": 0.5275079058483243,
|
| 1833 |
+
"flow/mag_ratio_std": 0.22685475973412395,
|
| 1834 |
+
"step": 58368
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.6173227036971594,
|
| 1838 |
+
"grad_norm": 0.15236690640449524,
|
| 1839 |
+
"learning_rate": 0.0003388374920626505,
|
| 1840 |
+
"loss": 1.546614408493042,
|
| 1841 |
+
"step": 59392
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 0.6173227036971594,
|
| 1845 |
+
"eval_cos_loss": 0.44333774503320456,
|
| 1846 |
+
"eval_loss": 1.5386833399534225,
|
| 1847 |
+
"eval_mse_loss": 1.3170144706964493,
|
| 1848 |
+
"flow/cos_sim": 0.5566621888428926,
|
| 1849 |
+
"flow/improvement_ratio": 0.9644087161868811,
|
| 1850 |
+
"flow/mag_ratio_mean": 0.536553805693984,
|
| 1851 |
+
"flow/mag_ratio_std": 0.22319983318448067,
|
| 1852 |
+
"step": 59392
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 0.6173227036971594,
|
| 1856 |
+
"eval_cos_loss": 0.44333774503320456,
|
| 1857 |
+
"eval_loss": 1.5386833399534225,
|
| 1858 |
+
"eval_mse_loss": 1.3170144706964493,
|
| 1859 |
+
"eval_runtime": 2.5638,
|
| 1860 |
+
"eval_samples_per_second": 780.079,
|
| 1861 |
+
"eval_steps_per_second": 12.481,
|
| 1862 |
+
"flow/cos_sim": 0.5566621888428926,
|
| 1863 |
+
"flow/improvement_ratio": 0.9644087161868811,
|
| 1864 |
+
"flow/mag_ratio_mean": 0.536553805693984,
|
| 1865 |
+
"flow/mag_ratio_std": 0.22319983318448067,
|
| 1866 |
+
"step": 59392
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.6279661985884897,
|
| 1870 |
+
"grad_norm": 0.20959317684173584,
|
| 1871 |
+
"learning_rate": 0.0003225883881381223,
|
| 1872 |
+
"loss": 1.5455403327941895,
|
| 1873 |
+
"step": 60416
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.6279661985884897,
|
| 1877 |
+
"eval_cos_loss": 0.4418268231675029,
|
| 1878 |
+
"eval_loss": 1.5369167998433113,
|
| 1879 |
+
"eval_mse_loss": 1.3160033896565437,
|
| 1880 |
+
"flow/cos_sim": 0.5581732373684645,
|
| 1881 |
+
"flow/improvement_ratio": 0.9641634412109852,
|
| 1882 |
+
"flow/mag_ratio_mean": 0.5371082350611687,
|
| 1883 |
+
"flow/mag_ratio_std": 0.22345659835264087,
|
| 1884 |
+
"step": 60416
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 0.6279661985884897,
|
| 1888 |
+
"eval_cos_loss": 0.4418268231675029,
|
| 1889 |
+
"eval_loss": 1.5369167998433113,
|
| 1890 |
+
"eval_mse_loss": 1.3160033896565437,
|
| 1891 |
+
"eval_runtime": 3.0215,
|
| 1892 |
+
"eval_samples_per_second": 661.926,
|
| 1893 |
+
"eval_steps_per_second": 10.591,
|
| 1894 |
+
"flow/cos_sim": 0.5581732373684645,
|
| 1895 |
+
"flow/improvement_ratio": 0.9641634412109852,
|
| 1896 |
+
"flow/mag_ratio_mean": 0.5371082350611687,
|
| 1897 |
+
"flow/mag_ratio_std": 0.22345659835264087,
|
| 1898 |
+
"step": 60416
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.63860969347982,
|
| 1902 |
+
"grad_norm": 0.14656518399715424,
|
| 1903 |
+
"learning_rate": 0.00030656647450995957,
|
| 1904 |
+
"loss": 1.5472609996795654,
|
| 1905 |
+
"step": 61440
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 0.63860969347982,
|
| 1909 |
+
"eval_cos_loss": 0.4388773338869214,
|
| 1910 |
+
"eval_loss": 1.5277933366596699,
|
| 1911 |
+
"eval_mse_loss": 1.308354664593935,
|
| 1912 |
+
"flow/cos_sim": 0.5611227098852396,
|
| 1913 |
+
"flow/improvement_ratio": 0.9615725018084049,
|
| 1914 |
+
"flow/mag_ratio_mean": 0.5404465068131685,
|
| 1915 |
+
"flow/mag_ratio_std": 0.22536235908046365,
|
| 1916 |
+
"step": 61440
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 0.63860969347982,
|
| 1920 |
+
"eval_cos_loss": 0.4388773338869214,
|
| 1921 |
+
"eval_loss": 1.5277933366596699,
|
| 1922 |
+
"eval_mse_loss": 1.308354664593935,
|
| 1923 |
+
"eval_runtime": 2.725,
|
| 1924 |
+
"eval_samples_per_second": 733.933,
|
| 1925 |
+
"eval_steps_per_second": 11.743,
|
| 1926 |
+
"flow/cos_sim": 0.5611227098852396,
|
| 1927 |
+
"flow/improvement_ratio": 0.9615725018084049,
|
| 1928 |
+
"flow/mag_ratio_mean": 0.5404465068131685,
|
| 1929 |
+
"flow/mag_ratio_std": 0.22536235908046365,
|
| 1930 |
+
"step": 61440
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.6492531883711503,
|
| 1934 |
+
"grad_norm": 0.16984502971172333,
|
| 1935 |
+
"learning_rate": 0.0002907595437968689,
|
| 1936 |
+
"loss": 1.546900987625122,
|
| 1937 |
+
"step": 62464
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 0.6492531883711503,
|
| 1941 |
+
"eval_cos_loss": 0.44708416890352964,
|
| 1942 |
+
"eval_loss": 1.550130132585764,
|
| 1943 |
+
"eval_mse_loss": 1.3265880458056927,
|
| 1944 |
+
"flow/cos_sim": 0.5529158180579543,
|
| 1945 |
+
"flow/improvement_ratio": 0.964706290513277,
|
| 1946 |
+
"flow/mag_ratio_mean": 0.5327735636383295,
|
| 1947 |
+
"flow/mag_ratio_std": 0.22531110560521483,
|
| 1948 |
+
"step": 62464
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 0.6492531883711503,
|
| 1952 |
+
"eval_cos_loss": 0.44708416890352964,
|
| 1953 |
+
"eval_loss": 1.550130132585764,
|
| 1954 |
+
"eval_mse_loss": 1.3265880458056927,
|
| 1955 |
+
"eval_runtime": 2.9226,
|
| 1956 |
+
"eval_samples_per_second": 684.317,
|
| 1957 |
+
"eval_steps_per_second": 10.949,
|
| 1958 |
+
"flow/cos_sim": 0.5529158180579543,
|
| 1959 |
+
"flow/improvement_ratio": 0.964706290513277,
|
| 1960 |
+
"flow/mag_ratio_mean": 0.5327735636383295,
|
| 1961 |
+
"flow/mag_ratio_std": 0.22531110560521483,
|
| 1962 |
+
"step": 62464
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.6598966832624806,
|
| 1966 |
+
"grad_norm": 0.16193066537380219,
|
| 1967 |
+
"learning_rate": 0.00027521728547552225,
|
| 1968 |
+
"loss": 1.5464342832565308,
|
| 1969 |
+
"step": 63488
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.6598966832624806,
|
| 1973 |
+
"eval_cos_loss": 0.44685271102935076,
|
| 1974 |
+
"eval_loss": 1.5440139174461365,
|
| 1975 |
+
"eval_mse_loss": 1.3205875717103481,
|
| 1976 |
+
"flow/cos_sim": 0.5531473197042942,
|
| 1977 |
+
"flow/improvement_ratio": 0.9620692692697048,
|
| 1978 |
+
"flow/mag_ratio_mean": 0.5350175518542528,
|
| 1979 |
+
"flow/mag_ratio_std": 0.23371326178312302,
|
| 1980 |
+
"step": 63488
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 0.6598966832624806,
|
| 1984 |
+
"eval_cos_loss": 0.44685271102935076,
|
| 1985 |
+
"eval_loss": 1.5440139174461365,
|
| 1986 |
+
"eval_mse_loss": 1.3205875717103481,
|
| 1987 |
+
"eval_runtime": 2.588,
|
| 1988 |
+
"eval_samples_per_second": 772.809,
|
| 1989 |
+
"eval_steps_per_second": 12.365,
|
| 1990 |
+
"flow/cos_sim": 0.5531473197042942,
|
| 1991 |
+
"flow/improvement_ratio": 0.9620692692697048,
|
| 1992 |
+
"flow/mag_ratio_mean": 0.5350175518542528,
|
| 1993 |
+
"flow/mag_ratio_std": 0.23371326178312302,
|
| 1994 |
+
"step": 63488
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.670540178153811,
|
| 1998 |
+
"grad_norm": 0.15779773890972137,
|
| 1999 |
+
"learning_rate": 0.00025992785762254565,
|
| 2000 |
+
"loss": 1.545114278793335,
|
| 2001 |
+
"step": 64512
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 0.670540178153811,
|
| 2005 |
+
"eval_cos_loss": 0.4412845829501748,
|
| 2006 |
+
"eval_loss": 1.5324720852077007,
|
| 2007 |
+
"eval_mse_loss": 1.311829797923565,
|
| 2008 |
+
"flow/cos_sim": 0.5587154757231474,
|
| 2009 |
+
"flow/improvement_ratio": 0.9654579609632492,
|
| 2010 |
+
"flow/mag_ratio_mean": 0.5386016722768545,
|
| 2011 |
+
"flow/mag_ratio_std": 0.22591237584128976,
|
| 2012 |
+
"step": 64512
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.670540178153811,
|
| 2016 |
+
"eval_cos_loss": 0.4412845829501748,
|
| 2017 |
+
"eval_loss": 1.5324720852077007,
|
| 2018 |
+
"eval_mse_loss": 1.311829797923565,
|
| 2019 |
+
"eval_runtime": 2.5359,
|
| 2020 |
+
"eval_samples_per_second": 788.675,
|
| 2021 |
+
"eval_steps_per_second": 12.619,
|
| 2022 |
+
"flow/cos_sim": 0.5587154757231474,
|
| 2023 |
+
"flow/improvement_ratio": 0.9654579609632492,
|
| 2024 |
+
"flow/mag_ratio_mean": 0.5386016722768545,
|
| 2025 |
+
"flow/mag_ratio_std": 0.22591237584128976,
|
| 2026 |
+
"step": 64512
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.6811836730451413,
|
| 2030 |
+
"grad_norm": 0.13823935389518738,
|
| 2031 |
+
"learning_rate": 0.00024493932293160253,
|
| 2032 |
+
"loss": 1.5480190515518188,
|
| 2033 |
+
"step": 65536
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.6811836730451413,
|
| 2037 |
+
"eval_cos_loss": 0.441369004547596,
|
| 2038 |
+
"eval_loss": 1.5341729335486889,
|
| 2039 |
+
"eval_mse_loss": 1.3134884350001812,
|
| 2040 |
+
"flow/cos_sim": 0.5586309880018234,
|
| 2041 |
+
"flow/improvement_ratio": 0.9630204811692238,
|
| 2042 |
+
"flow/mag_ratio_mean": 0.5338549390435219,
|
| 2043 |
+
"flow/mag_ratio_std": 0.2280183294788003,
|
| 2044 |
+
"step": 65536
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 0.6811836730451413,
|
| 2048 |
+
"eval_cos_loss": 0.441369004547596,
|
| 2049 |
+
"eval_loss": 1.5341729335486889,
|
| 2050 |
+
"eval_mse_loss": 1.3134884350001812,
|
| 2051 |
+
"eval_runtime": 2.5158,
|
| 2052 |
+
"eval_samples_per_second": 794.978,
|
| 2053 |
+
"eval_steps_per_second": 12.72,
|
| 2054 |
+
"flow/cos_sim": 0.5586309880018234,
|
| 2055 |
+
"flow/improvement_ratio": 0.9630204811692238,
|
| 2056 |
+
"flow/mag_ratio_mean": 0.5338549390435219,
|
| 2057 |
+
"flow/mag_ratio_std": 0.2280183294788003,
|
| 2058 |
+
"step": 65536
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 0.6918271679364716,
|
| 2062 |
+
"grad_norm": 0.15880072116851807,
|
| 2063 |
+
"learning_rate": 0.00023024026137121335,
|
| 2064 |
+
"loss": 1.5464882850646973,
|
| 2065 |
+
"step": 66560
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 0.6918271679364716,
|
| 2069 |
+
"eval_cos_loss": 0.4457838609814644,
|
| 2070 |
+
"eval_loss": 1.543992355465889,
|
| 2071 |
+
"eval_mse_loss": 1.3211004212498665,
|
| 2072 |
+
"flow/cos_sim": 0.554216225631535,
|
| 2073 |
+
"flow/improvement_ratio": 0.9604913741350174,
|
| 2074 |
+
"flow/mag_ratio_mean": 0.5307466676458716,
|
| 2075 |
+
"flow/mag_ratio_std": 0.22845542011782527,
|
| 2076 |
+
"step": 66560
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 0.6918271679364716,
|
| 2080 |
+
"eval_cos_loss": 0.4457838609814644,
|
| 2081 |
+
"eval_loss": 1.543992355465889,
|
| 2082 |
+
"eval_mse_loss": 1.3211004212498665,
|
| 2083 |
+
"eval_runtime": 2.821,
|
| 2084 |
+
"eval_samples_per_second": 708.957,
|
| 2085 |
+
"eval_steps_per_second": 11.343,
|
| 2086 |
+
"flow/cos_sim": 0.554216225631535,
|
| 2087 |
+
"flow/improvement_ratio": 0.9604913741350174,
|
| 2088 |
+
"flow/mag_ratio_mean": 0.5307466676458716,
|
| 2089 |
+
"flow/mag_ratio_std": 0.22845542011782527,
|
| 2090 |
+
"step": 66560
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.702470662827802,
|
| 2094 |
+
"grad_norm": 0.16780522465705872,
|
| 2095 |
+
"learning_rate": 0.0002158768798040357,
|
| 2096 |
+
"loss": 1.5465317964553833,
|
| 2097 |
+
"step": 67584
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 0.702470662827802,
|
| 2101 |
+
"eval_cos_loss": 0.4425074281170964,
|
| 2102 |
+
"eval_loss": 1.5348311252892017,
|
| 2103 |
+
"eval_mse_loss": 1.3135774172842503,
|
| 2104 |
+
"flow/cos_sim": 0.557492695748806,
|
| 2105 |
+
"flow/improvement_ratio": 0.9596080742776394,
|
| 2106 |
+
"flow/mag_ratio_mean": 0.5359784308820963,
|
| 2107 |
+
"flow/mag_ratio_std": 0.22894051391631365,
|
| 2108 |
+
"step": 67584
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 0.702470662827802,
|
| 2112 |
+
"eval_cos_loss": 0.4425074281170964,
|
| 2113 |
+
"eval_loss": 1.5348311252892017,
|
| 2114 |
+
"eval_mse_loss": 1.3135774172842503,
|
| 2115 |
+
"eval_runtime": 3.1093,
|
| 2116 |
+
"eval_samples_per_second": 643.239,
|
| 2117 |
+
"eval_steps_per_second": 10.292,
|
| 2118 |
+
"flow/cos_sim": 0.557492695748806,
|
| 2119 |
+
"flow/improvement_ratio": 0.9596080742776394,
|
| 2120 |
+
"flow/mag_ratio_mean": 0.5359784308820963,
|
| 2121 |
+
"flow/mag_ratio_std": 0.22894051391631365,
|
| 2122 |
+
"step": 67584
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 0.7131141577191323,
|
| 2126 |
+
"grad_norm": 0.14971283078193665,
|
| 2127 |
+
"learning_rate": 0.0002018382345138871,
|
| 2128 |
+
"loss": 1.5448497533798218,
|
| 2129 |
+
"step": 68608
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 0.7131141577191323,
|
| 2133 |
+
"eval_cos_loss": 0.4460556413978338,
|
| 2134 |
+
"eval_loss": 1.544582311064005,
|
| 2135 |
+
"eval_mse_loss": 1.3215544857084751,
|
| 2136 |
+
"flow/cos_sim": 0.5539443735033274,
|
| 2137 |
+
"flow/improvement_ratio": 0.9611221943050623,
|
| 2138 |
+
"flow/mag_ratio_mean": 0.5354085844010115,
|
| 2139 |
+
"flow/mag_ratio_std": 0.23012477485463023,
|
| 2140 |
+
"step": 68608
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 0.7131141577191323,
|
| 2144 |
+
"eval_cos_loss": 0.4460556413978338,
|
| 2145 |
+
"eval_loss": 1.544582311064005,
|
| 2146 |
+
"eval_mse_loss": 1.3215544857084751,
|
| 2147 |
+
"eval_runtime": 2.5817,
|
| 2148 |
+
"eval_samples_per_second": 774.689,
|
| 2149 |
+
"eval_steps_per_second": 12.395,
|
| 2150 |
+
"flow/cos_sim": 0.5539443735033274,
|
| 2151 |
+
"flow/improvement_ratio": 0.9611221943050623,
|
| 2152 |
+
"flow/mag_ratio_mean": 0.5354085844010115,
|
| 2153 |
+
"flow/mag_ratio_std": 0.23012477485463023,
|
| 2154 |
+
"step": 68608
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 0.7237576526104627,
|
| 2158 |
+
"grad_norm": 0.14716410636901855,
|
| 2159 |
+
"learning_rate": 0.00018816845632858647,
|
| 2160 |
+
"loss": 1.5454319715499878,
|
| 2161 |
+
"step": 69632
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 0.7237576526104627,
|
| 2165 |
+
"eval_cos_loss": 0.4387012077495456,
|
| 2166 |
+
"eval_loss": 1.527622751891613,
|
| 2167 |
+
"eval_mse_loss": 1.3082721531391144,
|
| 2168 |
+
"flow/cos_sim": 0.5612988974899054,
|
| 2169 |
+
"flow/improvement_ratio": 0.9638887159526348,
|
| 2170 |
+
"flow/mag_ratio_mean": 0.5365499863401055,
|
| 2171 |
+
"flow/mag_ratio_std": 0.2257627071812749,
|
| 2172 |
+
"step": 69632
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 0.7237576526104627,
|
| 2176 |
+
"eval_cos_loss": 0.4387012077495456,
|
| 2177 |
+
"eval_loss": 1.527622751891613,
|
| 2178 |
+
"eval_mse_loss": 1.3082721531391144,
|
| 2179 |
+
"eval_runtime": 2.5421,
|
| 2180 |
+
"eval_samples_per_second": 786.736,
|
| 2181 |
+
"eval_steps_per_second": 12.588,
|
| 2182 |
+
"flow/cos_sim": 0.5612988974899054,
|
| 2183 |
+
"flow/improvement_ratio": 0.9638887159526348,
|
| 2184 |
+
"flow/mag_ratio_mean": 0.5365499863401055,
|
| 2185 |
+
"flow/mag_ratio_std": 0.2257627071812749,
|
| 2186 |
+
"step": 69632
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.734401147501793,
|
| 2190 |
+
"grad_norm": 0.1667480319738388,
|
| 2191 |
+
"learning_rate": 0.00017485713000071836,
|
| 2192 |
+
"loss": 1.5442575216293335,
|
| 2193 |
+
"step": 70656
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.734401147501793,
|
| 2197 |
+
"eval_cos_loss": 0.44433039985597134,
|
| 2198 |
+
"eval_loss": 1.540200136601925,
|
| 2199 |
+
"eval_mse_loss": 1.3180349357426167,
|
| 2200 |
+
"flow/cos_sim": 0.5556696448475122,
|
| 2201 |
+
"flow/improvement_ratio": 0.9625885803252459,
|
| 2202 |
+
"flow/mag_ratio_mean": 0.5337581913918257,
|
| 2203 |
+
"flow/mag_ratio_std": 0.22786249266937375,
|
| 2204 |
+
"step": 70656
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.734401147501793,
|
| 2208 |
+
"eval_cos_loss": 0.44433039985597134,
|
| 2209 |
+
"eval_loss": 1.540200136601925,
|
| 2210 |
+
"eval_mse_loss": 1.3180349357426167,
|
| 2211 |
+
"eval_runtime": 2.5514,
|
| 2212 |
+
"eval_samples_per_second": 783.877,
|
| 2213 |
+
"eval_steps_per_second": 12.542,
|
| 2214 |
+
"flow/cos_sim": 0.5556696448475122,
|
| 2215 |
+
"flow/improvement_ratio": 0.9625885803252459,
|
| 2216 |
+
"flow/mag_ratio_mean": 0.5337581913918257,
|
| 2217 |
+
"flow/mag_ratio_std": 0.22786249266937375,
|
| 2218 |
+
"step": 70656
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 0.7450446423931233,
|
| 2222 |
+
"grad_norm": 0.21780510246753693,
|
| 2223 |
+
"learning_rate": 0.00016194610001300043,
|
| 2224 |
+
"loss": 1.5446763038635254,
|
| 2225 |
+
"step": 71680
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 0.7450446423931233,
|
| 2229 |
+
"eval_cos_loss": 0.44072897639125586,
|
| 2230 |
+
"eval_loss": 1.53380636125803,
|
| 2231 |
+
"eval_mse_loss": 1.3134418688714504,
|
| 2232 |
+
"flow/cos_sim": 0.559271028265357,
|
| 2233 |
+
"flow/improvement_ratio": 0.9642956424504519,
|
| 2234 |
+
"flow/mag_ratio_mean": 0.5349069200456142,
|
| 2235 |
+
"flow/mag_ratio_std": 0.22263477742671967,
|
| 2236 |
+
"step": 71680
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 0.7450446423931233,
|
| 2240 |
+
"eval_cos_loss": 0.44072897639125586,
|
| 2241 |
+
"eval_loss": 1.53380636125803,
|
| 2242 |
+
"eval_mse_loss": 1.3134418688714504,
|
| 2243 |
+
"eval_runtime": 2.544,
|
| 2244 |
+
"eval_samples_per_second": 786.162,
|
| 2245 |
+
"eval_steps_per_second": 12.579,
|
| 2246 |
+
"flow/cos_sim": 0.559271028265357,
|
| 2247 |
+
"flow/improvement_ratio": 0.9642956424504519,
|
| 2248 |
+
"flow/mag_ratio_mean": 0.5349069200456142,
|
| 2249 |
+
"flow/mag_ratio_std": 0.22263477742671967,
|
| 2250 |
+
"step": 71680
|
| 2251 |
+
}
|
| 2252 |
+
],
|
| 2253 |
+
"logging_steps": 1024,
|
| 2254 |
+
"max_steps": 96209,
|
| 2255 |
+
"num_input_tokens_seen": 0,
|
| 2256 |
+
"num_train_epochs": 1,
|
| 2257 |
+
"save_steps": 1024,
|
| 2258 |
+
"stateful_callbacks": {
|
| 2259 |
+
"TrainerControl": {
|
| 2260 |
+
"args": {
|
| 2261 |
+
"should_epoch_stop": false,
|
| 2262 |
+
"should_evaluate": false,
|
| 2263 |
+
"should_log": false,
|
| 2264 |
+
"should_save": true,
|
| 2265 |
+
"should_training_stop": false
|
| 2266 |
+
},
|
| 2267 |
+
"attributes": {}
|
| 2268 |
+
}
|
| 2269 |
+
},
|
| 2270 |
+
"total_flos": 0.0,
|
| 2271 |
+
"train_batch_size": 64,
|
| 2272 |
+
"trial_name": null,
|
| 2273 |
+
"trial_params": null
|
| 2274 |
+
}
|
checkpoints-v2.6-c/checkpoint-71680/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd8764e705e3d3a01283602f7bbb7db8de6272269197e8d0fb5f615cd86459fe
|
| 3 |
+
size 5137
|