End of training
Browse files- README.md +2 -1
- all_results.json +16 -0
- train_results.json +16 -0
- trainer_state.json +3490 -0
- training_loss.png +0 -0
README.md
CHANGED
|
@@ -4,6 +4,7 @@ license: apache-2.0
|
|
| 4 |
base_model: Qwen/Qwen3-8B
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
|
|
|
| 7 |
- generated_from_trainer
|
| 8 |
model-index:
|
| 9 |
- name: code-contests-sandboxes-traces-terminus-2_global-batch-size_32
|
|
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# code-contests-sandboxes-traces-terminus-2_global-batch-size_32
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B) on
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 4 |
base_model: Qwen/Qwen3-8B
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
| 7 |
+
- full
|
| 8 |
- generated_from_trainer
|
| 9 |
model-index:
|
| 10 |
- name: code-contests-sandboxes-traces-terminus-2_global-batch-size_32
|
|
|
|
| 16 |
|
| 17 |
# code-contests-sandboxes-traces-terminus-2_global-batch-size_32
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B) on the mlfoundations-dev/code-contests-sandboxes-traces-terminus-2 dataset.
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"achieved_tflops_per_gpu": 2.6369613385336814,
|
| 3 |
+
"achieved_tflops_per_gpu_theoretical": 304.9114106540851,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"loss_nan_ranks": 0,
|
| 6 |
+
"loss_rank_avg": 0.3492066562175751,
|
| 7 |
+
"mfu_percent": 0.18635769176916478,
|
| 8 |
+
"mfu_percent_theoretical": 21.54850958686114,
|
| 9 |
+
"total_flos": 7.336290139934556e+17,
|
| 10 |
+
"train_loss": 0.37341142744301986,
|
| 11 |
+
"train_runtime": 17388.125,
|
| 12 |
+
"train_samples_per_second": 2.876,
|
| 13 |
+
"train_steps_per_second": 0.09,
|
| 14 |
+
"valid_targets_mean": 3627.8,
|
| 15 |
+
"valid_targets_min": 954
|
| 16 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"achieved_tflops_per_gpu": 2.6369613385336814,
|
| 3 |
+
"achieved_tflops_per_gpu_theoretical": 304.9114106540851,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"loss_nan_ranks": 0,
|
| 6 |
+
"loss_rank_avg": 0.3492066562175751,
|
| 7 |
+
"mfu_percent": 0.18635769176916478,
|
| 8 |
+
"mfu_percent_theoretical": 21.54850958686114,
|
| 9 |
+
"total_flos": 7.336290139934556e+17,
|
| 10 |
+
"train_loss": 0.37341142744301986,
|
| 11 |
+
"train_runtime": 17388.125,
|
| 12 |
+
"train_samples_per_second": 2.876,
|
| 13 |
+
"train_steps_per_second": 0.09,
|
| 14 |
+
"valid_targets_mean": 3627.8,
|
| 15 |
+
"valid_targets_min": 954
|
| 16 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3490 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1565,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.016,
|
| 14 |
+
"grad_norm": 5.69872959186611,
|
| 15 |
+
"learning_rate": 1.0191082802547772e-06,
|
| 16 |
+
"loss": 0.6974,
|
| 17 |
+
"loss_nan_ranks": 0,
|
| 18 |
+
"loss_rank_avg": 0.3614201545715332,
|
| 19 |
+
"step": 5,
|
| 20 |
+
"valid_targets_mean": 3866.1,
|
| 21 |
+
"valid_targets_min": 867
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.032,
|
| 25 |
+
"grad_norm": 5.164438631475304,
|
| 26 |
+
"learning_rate": 2.2929936305732485e-06,
|
| 27 |
+
"loss": 0.6778,
|
| 28 |
+
"loss_nan_ranks": 0,
|
| 29 |
+
"loss_rank_avg": 0.3798177242279053,
|
| 30 |
+
"step": 10,
|
| 31 |
+
"valid_targets_mean": 3772.9,
|
| 32 |
+
"valid_targets_min": 1172
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.048,
|
| 36 |
+
"grad_norm": 3.1544536534246252,
|
| 37 |
+
"learning_rate": 3.56687898089172e-06,
|
| 38 |
+
"loss": 0.6826,
|
| 39 |
+
"loss_nan_ranks": 0,
|
| 40 |
+
"loss_rank_avg": 0.33693352341651917,
|
| 41 |
+
"step": 15,
|
| 42 |
+
"valid_targets_mean": 5926.4,
|
| 43 |
+
"valid_targets_min": 955
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.064,
|
| 47 |
+
"grad_norm": 2.0390987903237137,
|
| 48 |
+
"learning_rate": 4.840764331210192e-06,
|
| 49 |
+
"loss": 0.6153,
|
| 50 |
+
"loss_nan_ranks": 0,
|
| 51 |
+
"loss_rank_avg": 0.3763325810432434,
|
| 52 |
+
"step": 20,
|
| 53 |
+
"valid_targets_mean": 5537.4,
|
| 54 |
+
"valid_targets_min": 684
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.08,
|
| 58 |
+
"grad_norm": 1.2724968440026243,
|
| 59 |
+
"learning_rate": 6.114649681528663e-06,
|
| 60 |
+
"loss": 0.5933,
|
| 61 |
+
"loss_nan_ranks": 0,
|
| 62 |
+
"loss_rank_avg": 0.2624075710773468,
|
| 63 |
+
"step": 25,
|
| 64 |
+
"valid_targets_mean": 3464.9,
|
| 65 |
+
"valid_targets_min": 557
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.096,
|
| 69 |
+
"grad_norm": 0.8015927473817406,
|
| 70 |
+
"learning_rate": 7.388535031847134e-06,
|
| 71 |
+
"loss": 0.5901,
|
| 72 |
+
"loss_nan_ranks": 0,
|
| 73 |
+
"loss_rank_avg": 0.2664138078689575,
|
| 74 |
+
"step": 30,
|
| 75 |
+
"valid_targets_mean": 4067.4,
|
| 76 |
+
"valid_targets_min": 422
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.112,
|
| 80 |
+
"grad_norm": 0.7590646795996507,
|
| 81 |
+
"learning_rate": 8.662420382165606e-06,
|
| 82 |
+
"loss": 0.5455,
|
| 83 |
+
"loss_nan_ranks": 0,
|
| 84 |
+
"loss_rank_avg": 0.27429863810539246,
|
| 85 |
+
"step": 35,
|
| 86 |
+
"valid_targets_mean": 3136.5,
|
| 87 |
+
"valid_targets_min": 596
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.128,
|
| 91 |
+
"grad_norm": 0.5768938436627202,
|
| 92 |
+
"learning_rate": 9.936305732484078e-06,
|
| 93 |
+
"loss": 0.5396,
|
| 94 |
+
"loss_nan_ranks": 0,
|
| 95 |
+
"loss_rank_avg": 0.25026804208755493,
|
| 96 |
+
"step": 40,
|
| 97 |
+
"valid_targets_mean": 3707.1,
|
| 98 |
+
"valid_targets_min": 905
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.144,
|
| 102 |
+
"grad_norm": 0.545771347093659,
|
| 103 |
+
"learning_rate": 1.1210191082802548e-05,
|
| 104 |
+
"loss": 0.5283,
|
| 105 |
+
"loss_nan_ranks": 0,
|
| 106 |
+
"loss_rank_avg": 0.2582656741142273,
|
| 107 |
+
"step": 45,
|
| 108 |
+
"valid_targets_mean": 3181.2,
|
| 109 |
+
"valid_targets_min": 1120
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.16,
|
| 113 |
+
"grad_norm": 0.5029841857856351,
|
| 114 |
+
"learning_rate": 1.248407643312102e-05,
|
| 115 |
+
"loss": 0.5368,
|
| 116 |
+
"loss_nan_ranks": 0,
|
| 117 |
+
"loss_rank_avg": 0.35082149505615234,
|
| 118 |
+
"step": 50,
|
| 119 |
+
"valid_targets_mean": 5725.4,
|
| 120 |
+
"valid_targets_min": 665
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.176,
|
| 124 |
+
"grad_norm": 0.4722335098176793,
|
| 125 |
+
"learning_rate": 1.375796178343949e-05,
|
| 126 |
+
"loss": 0.5508,
|
| 127 |
+
"loss_nan_ranks": 0,
|
| 128 |
+
"loss_rank_avg": 0.3100292384624481,
|
| 129 |
+
"step": 55,
|
| 130 |
+
"valid_targets_mean": 4060.9,
|
| 131 |
+
"valid_targets_min": 671
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.192,
|
| 135 |
+
"grad_norm": 0.4393102585880352,
|
| 136 |
+
"learning_rate": 1.5031847133757964e-05,
|
| 137 |
+
"loss": 0.4724,
|
| 138 |
+
"loss_nan_ranks": 0,
|
| 139 |
+
"loss_rank_avg": 0.2322624772787094,
|
| 140 |
+
"step": 60,
|
| 141 |
+
"valid_targets_mean": 3568.0,
|
| 142 |
+
"valid_targets_min": 727
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.208,
|
| 146 |
+
"grad_norm": 0.46984637985358174,
|
| 147 |
+
"learning_rate": 1.6305732484076436e-05,
|
| 148 |
+
"loss": 0.4841,
|
| 149 |
+
"loss_nan_ranks": 0,
|
| 150 |
+
"loss_rank_avg": 0.22150176763534546,
|
| 151 |
+
"step": 65,
|
| 152 |
+
"valid_targets_mean": 2860.4,
|
| 153 |
+
"valid_targets_min": 1026
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.224,
|
| 157 |
+
"grad_norm": 0.38610287129563536,
|
| 158 |
+
"learning_rate": 1.7579617834394907e-05,
|
| 159 |
+
"loss": 0.4838,
|
| 160 |
+
"loss_nan_ranks": 0,
|
| 161 |
+
"loss_rank_avg": 0.18804149329662323,
|
| 162 |
+
"step": 70,
|
| 163 |
+
"valid_targets_mean": 3148.8,
|
| 164 |
+
"valid_targets_min": 1224
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.24,
|
| 168 |
+
"grad_norm": 0.5071493123995814,
|
| 169 |
+
"learning_rate": 1.8853503184713376e-05,
|
| 170 |
+
"loss": 0.4685,
|
| 171 |
+
"loss_nan_ranks": 0,
|
| 172 |
+
"loss_rank_avg": 0.22837616503238678,
|
| 173 |
+
"step": 75,
|
| 174 |
+
"valid_targets_mean": 2380.9,
|
| 175 |
+
"valid_targets_min": 760
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.256,
|
| 179 |
+
"grad_norm": 0.39991978336257084,
|
| 180 |
+
"learning_rate": 2.0127388535031848e-05,
|
| 181 |
+
"loss": 0.473,
|
| 182 |
+
"loss_nan_ranks": 0,
|
| 183 |
+
"loss_rank_avg": 0.23189273476600647,
|
| 184 |
+
"step": 80,
|
| 185 |
+
"valid_targets_mean": 3358.9,
|
| 186 |
+
"valid_targets_min": 811
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.272,
|
| 190 |
+
"grad_norm": 0.3750148456903095,
|
| 191 |
+
"learning_rate": 2.140127388535032e-05,
|
| 192 |
+
"loss": 0.4527,
|
| 193 |
+
"loss_nan_ranks": 0,
|
| 194 |
+
"loss_rank_avg": 0.22813287377357483,
|
| 195 |
+
"step": 85,
|
| 196 |
+
"valid_targets_mean": 5757.9,
|
| 197 |
+
"valid_targets_min": 890
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.288,
|
| 201 |
+
"grad_norm": 0.403265544100627,
|
| 202 |
+
"learning_rate": 2.267515923566879e-05,
|
| 203 |
+
"loss": 0.4428,
|
| 204 |
+
"loss_nan_ranks": 0,
|
| 205 |
+
"loss_rank_avg": 0.18977582454681396,
|
| 206 |
+
"step": 90,
|
| 207 |
+
"valid_targets_mean": 3059.4,
|
| 208 |
+
"valid_targets_min": 790
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.304,
|
| 212 |
+
"grad_norm": 0.4536180824055258,
|
| 213 |
+
"learning_rate": 2.3949044585987263e-05,
|
| 214 |
+
"loss": 0.438,
|
| 215 |
+
"loss_nan_ranks": 0,
|
| 216 |
+
"loss_rank_avg": 0.23432332277297974,
|
| 217 |
+
"step": 95,
|
| 218 |
+
"valid_targets_mean": 3082.3,
|
| 219 |
+
"valid_targets_min": 803
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.32,
|
| 223 |
+
"grad_norm": 0.3512124065455802,
|
| 224 |
+
"learning_rate": 2.5222929936305732e-05,
|
| 225 |
+
"loss": 0.4555,
|
| 226 |
+
"loss_nan_ranks": 0,
|
| 227 |
+
"loss_rank_avg": 0.21950744092464447,
|
| 228 |
+
"step": 100,
|
| 229 |
+
"valid_targets_mean": 4653.7,
|
| 230 |
+
"valid_targets_min": 1002
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.336,
|
| 234 |
+
"grad_norm": 0.3089899886308942,
|
| 235 |
+
"learning_rate": 2.6496815286624204e-05,
|
| 236 |
+
"loss": 0.4491,
|
| 237 |
+
"loss_nan_ranks": 0,
|
| 238 |
+
"loss_rank_avg": 0.19772186875343323,
|
| 239 |
+
"step": 105,
|
| 240 |
+
"valid_targets_mean": 4689.6,
|
| 241 |
+
"valid_targets_min": 701
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.352,
|
| 245 |
+
"grad_norm": 0.34589835913470146,
|
| 246 |
+
"learning_rate": 2.7770700636942676e-05,
|
| 247 |
+
"loss": 0.4341,
|
| 248 |
+
"loss_nan_ranks": 0,
|
| 249 |
+
"loss_rank_avg": 0.22734415531158447,
|
| 250 |
+
"step": 110,
|
| 251 |
+
"valid_targets_mean": 5929.0,
|
| 252 |
+
"valid_targets_min": 737
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.368,
|
| 256 |
+
"grad_norm": 0.333900665380794,
|
| 257 |
+
"learning_rate": 2.9044585987261148e-05,
|
| 258 |
+
"loss": 0.4147,
|
| 259 |
+
"loss_nan_ranks": 0,
|
| 260 |
+
"loss_rank_avg": 0.15204599499702454,
|
| 261 |
+
"step": 115,
|
| 262 |
+
"valid_targets_mean": 4069.6,
|
| 263 |
+
"valid_targets_min": 807
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.384,
|
| 267 |
+
"grad_norm": 0.4286792724989212,
|
| 268 |
+
"learning_rate": 3.0318471337579623e-05,
|
| 269 |
+
"loss": 0.4335,
|
| 270 |
+
"loss_nan_ranks": 0,
|
| 271 |
+
"loss_rank_avg": 0.2184617519378662,
|
| 272 |
+
"step": 120,
|
| 273 |
+
"valid_targets_mean": 2641.9,
|
| 274 |
+
"valid_targets_min": 651
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.4,
|
| 278 |
+
"grad_norm": 0.4994416444087381,
|
| 279 |
+
"learning_rate": 3.1592356687898095e-05,
|
| 280 |
+
"loss": 0.4316,
|
| 281 |
+
"loss_nan_ranks": 0,
|
| 282 |
+
"loss_rank_avg": 0.13226771354675293,
|
| 283 |
+
"step": 125,
|
| 284 |
+
"valid_targets_mean": 2113.8,
|
| 285 |
+
"valid_targets_min": 619
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.416,
|
| 289 |
+
"grad_norm": 0.4533713886076989,
|
| 290 |
+
"learning_rate": 3.286624203821656e-05,
|
| 291 |
+
"loss": 0.4146,
|
| 292 |
+
"loss_nan_ranks": 0,
|
| 293 |
+
"loss_rank_avg": 0.18592208623886108,
|
| 294 |
+
"step": 130,
|
| 295 |
+
"valid_targets_mean": 2183.4,
|
| 296 |
+
"valid_targets_min": 624
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.432,
|
| 300 |
+
"grad_norm": 0.4063059240583294,
|
| 301 |
+
"learning_rate": 3.414012738853504e-05,
|
| 302 |
+
"loss": 0.4358,
|
| 303 |
+
"loss_nan_ranks": 0,
|
| 304 |
+
"loss_rank_avg": 0.25601351261138916,
|
| 305 |
+
"step": 135,
|
| 306 |
+
"valid_targets_mean": 4597.6,
|
| 307 |
+
"valid_targets_min": 1140
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.448,
|
| 311 |
+
"grad_norm": 0.36028533700369003,
|
| 312 |
+
"learning_rate": 3.541401273885351e-05,
|
| 313 |
+
"loss": 0.4056,
|
| 314 |
+
"loss_nan_ranks": 0,
|
| 315 |
+
"loss_rank_avg": 0.21660026907920837,
|
| 316 |
+
"step": 140,
|
| 317 |
+
"valid_targets_mean": 4851.8,
|
| 318 |
+
"valid_targets_min": 920
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.464,
|
| 322 |
+
"grad_norm": 0.37494642923573474,
|
| 323 |
+
"learning_rate": 3.6687898089171976e-05,
|
| 324 |
+
"loss": 0.4364,
|
| 325 |
+
"loss_nan_ranks": 0,
|
| 326 |
+
"loss_rank_avg": 0.2517198324203491,
|
| 327 |
+
"step": 145,
|
| 328 |
+
"valid_targets_mean": 4430.9,
|
| 329 |
+
"valid_targets_min": 1251
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.48,
|
| 333 |
+
"grad_norm": 0.416006623617401,
|
| 334 |
+
"learning_rate": 3.796178343949045e-05,
|
| 335 |
+
"loss": 0.4304,
|
| 336 |
+
"loss_nan_ranks": 0,
|
| 337 |
+
"loss_rank_avg": 0.257464200258255,
|
| 338 |
+
"step": 150,
|
| 339 |
+
"valid_targets_mean": 5269.2,
|
| 340 |
+
"valid_targets_min": 699
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.496,
|
| 344 |
+
"grad_norm": 0.4201384306903681,
|
| 345 |
+
"learning_rate": 3.923566878980892e-05,
|
| 346 |
+
"loss": 0.419,
|
| 347 |
+
"loss_nan_ranks": 0,
|
| 348 |
+
"loss_rank_avg": 0.26475703716278076,
|
| 349 |
+
"step": 155,
|
| 350 |
+
"valid_targets_mean": 4737.1,
|
| 351 |
+
"valid_targets_min": 702
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.512,
|
| 355 |
+
"grad_norm": 0.37008181636461424,
|
| 356 |
+
"learning_rate": 3.999980086219931e-05,
|
| 357 |
+
"loss": 0.4124,
|
| 358 |
+
"loss_nan_ranks": 0,
|
| 359 |
+
"loss_rank_avg": 0.16652914881706238,
|
| 360 |
+
"step": 160,
|
| 361 |
+
"valid_targets_mean": 2959.7,
|
| 362 |
+
"valid_targets_min": 586
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.528,
|
| 366 |
+
"grad_norm": 0.4076042174730073,
|
| 367 |
+
"learning_rate": 3.9997560607483595e-05,
|
| 368 |
+
"loss": 0.4087,
|
| 369 |
+
"loss_nan_ranks": 0,
|
| 370 |
+
"loss_rank_avg": 0.16840457916259766,
|
| 371 |
+
"step": 165,
|
| 372 |
+
"valid_targets_mean": 3294.4,
|
| 373 |
+
"valid_targets_min": 732
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.544,
|
| 377 |
+
"grad_norm": 0.32841247716849165,
|
| 378 |
+
"learning_rate": 3.999283145555291e-05,
|
| 379 |
+
"loss": 0.4026,
|
| 380 |
+
"loss_nan_ranks": 0,
|
| 381 |
+
"loss_rank_avg": 0.19586911797523499,
|
| 382 |
+
"step": 170,
|
| 383 |
+
"valid_targets_mean": 5450.7,
|
| 384 |
+
"valid_targets_min": 693
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.56,
|
| 388 |
+
"grad_norm": 0.41779856510643243,
|
| 389 |
+
"learning_rate": 3.998561399499772e-05,
|
| 390 |
+
"loss": 0.3959,
|
| 391 |
+
"loss_nan_ranks": 0,
|
| 392 |
+
"loss_rank_avg": 0.174172043800354,
|
| 393 |
+
"step": 175,
|
| 394 |
+
"valid_targets_mean": 4048.8,
|
| 395 |
+
"valid_targets_min": 916
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.576,
|
| 399 |
+
"grad_norm": 0.4741559609057003,
|
| 400 |
+
"learning_rate": 3.997590912410345e-05,
|
| 401 |
+
"loss": 0.406,
|
| 402 |
+
"loss_nan_ranks": 0,
|
| 403 |
+
"loss_rank_avg": 0.17835122346878052,
|
| 404 |
+
"step": 180,
|
| 405 |
+
"valid_targets_mean": 2691.6,
|
| 406 |
+
"valid_targets_min": 878
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.592,
|
| 410 |
+
"grad_norm": 0.4233303846560544,
|
| 411 |
+
"learning_rate": 3.996371805073874e-05,
|
| 412 |
+
"loss": 0.4124,
|
| 413 |
+
"loss_nan_ranks": 0,
|
| 414 |
+
"loss_rank_avg": 0.19907709956169128,
|
| 415 |
+
"step": 185,
|
| 416 |
+
"valid_targets_mean": 3547.6,
|
| 417 |
+
"valid_targets_min": 582
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.608,
|
| 421 |
+
"grad_norm": 0.5519863740267512,
|
| 422 |
+
"learning_rate": 3.994904229220507e-05,
|
| 423 |
+
"loss": 0.4131,
|
| 424 |
+
"loss_nan_ranks": 0,
|
| 425 |
+
"loss_rank_avg": 0.2815743386745453,
|
| 426 |
+
"step": 190,
|
| 427 |
+
"valid_targets_mean": 5913.5,
|
| 428 |
+
"valid_targets_min": 789
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.624,
|
| 432 |
+
"grad_norm": 0.3853735917618442,
|
| 433 |
+
"learning_rate": 3.9931883675047966e-05,
|
| 434 |
+
"loss": 0.4107,
|
| 435 |
+
"loss_nan_ranks": 0,
|
| 436 |
+
"loss_rank_avg": 0.2598858177661896,
|
| 437 |
+
"step": 195,
|
| 438 |
+
"valid_targets_mean": 5340.4,
|
| 439 |
+
"valid_targets_min": 1088
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.64,
|
| 443 |
+
"grad_norm": 0.38155567667116275,
|
| 444 |
+
"learning_rate": 3.991224433482961e-05,
|
| 445 |
+
"loss": 0.3849,
|
| 446 |
+
"loss_nan_ranks": 0,
|
| 447 |
+
"loss_rank_avg": 0.16318747401237488,
|
| 448 |
+
"step": 200,
|
| 449 |
+
"valid_targets_mean": 3579.7,
|
| 450 |
+
"valid_targets_min": 598
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.656,
|
| 454 |
+
"grad_norm": 0.3511821690187007,
|
| 455 |
+
"learning_rate": 3.98901267158631e-05,
|
| 456 |
+
"loss": 0.3904,
|
| 457 |
+
"loss_nan_ranks": 0,
|
| 458 |
+
"loss_rank_avg": 0.13865548372268677,
|
| 459 |
+
"step": 205,
|
| 460 |
+
"valid_targets_mean": 3389.2,
|
| 461 |
+
"valid_targets_min": 740
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.672,
|
| 465 |
+
"grad_norm": 0.35229572545058246,
|
| 466 |
+
"learning_rate": 3.98655335709082e-05,
|
| 467 |
+
"loss": 0.4191,
|
| 468 |
+
"loss_nan_ranks": 0,
|
| 469 |
+
"loss_rank_avg": 0.13957738876342773,
|
| 470 |
+
"step": 210,
|
| 471 |
+
"valid_targets_mean": 2704.1,
|
| 472 |
+
"valid_targets_min": 784
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.688,
|
| 476 |
+
"grad_norm": 0.35386211696296843,
|
| 477 |
+
"learning_rate": 3.9838467960828745e-05,
|
| 478 |
+
"loss": 0.4035,
|
| 479 |
+
"loss_nan_ranks": 0,
|
| 480 |
+
"loss_rank_avg": 0.20394352078437805,
|
| 481 |
+
"step": 215,
|
| 482 |
+
"valid_targets_mean": 4380.9,
|
| 483 |
+
"valid_targets_min": 546
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.704,
|
| 487 |
+
"grad_norm": 0.3704950358052972,
|
| 488 |
+
"learning_rate": 3.9808933254211665e-05,
|
| 489 |
+
"loss": 0.4189,
|
| 490 |
+
"loss_nan_ranks": 0,
|
| 491 |
+
"loss_rank_avg": 0.21493899822235107,
|
| 492 |
+
"step": 220,
|
| 493 |
+
"valid_targets_mean": 4052.7,
|
| 494 |
+
"valid_targets_min": 692
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.72,
|
| 498 |
+
"grad_norm": 0.4196268184599092,
|
| 499 |
+
"learning_rate": 3.977693312694778e-05,
|
| 500 |
+
"loss": 0.3944,
|
| 501 |
+
"loss_nan_ranks": 0,
|
| 502 |
+
"loss_rank_avg": 0.188992440700531,
|
| 503 |
+
"step": 225,
|
| 504 |
+
"valid_targets_mean": 3296.3,
|
| 505 |
+
"valid_targets_min": 752
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.736,
|
| 509 |
+
"grad_norm": 0.3908305933846935,
|
| 510 |
+
"learning_rate": 3.974247156177423e-05,
|
| 511 |
+
"loss": 0.4177,
|
| 512 |
+
"loss_nan_ranks": 0,
|
| 513 |
+
"loss_rank_avg": 0.16833311319351196,
|
| 514 |
+
"step": 230,
|
| 515 |
+
"valid_targets_mean": 3082.8,
|
| 516 |
+
"valid_targets_min": 623
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.752,
|
| 520 |
+
"grad_norm": 0.4976376129445208,
|
| 521 |
+
"learning_rate": 3.970555284777883e-05,
|
| 522 |
+
"loss": 0.4163,
|
| 523 |
+
"loss_nan_ranks": 0,
|
| 524 |
+
"loss_rank_avg": 0.21885736286640167,
|
| 525 |
+
"step": 235,
|
| 526 |
+
"valid_targets_mean": 3845.2,
|
| 527 |
+
"valid_targets_min": 787
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.768,
|
| 531 |
+
"grad_norm": 0.4175782511172779,
|
| 532 |
+
"learning_rate": 3.9666181579866244e-05,
|
| 533 |
+
"loss": 0.4175,
|
| 534 |
+
"loss_nan_ranks": 0,
|
| 535 |
+
"loss_rank_avg": 0.25375479459762573,
|
| 536 |
+
"step": 240,
|
| 537 |
+
"valid_targets_mean": 4226.0,
|
| 538 |
+
"valid_targets_min": 701
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.784,
|
| 542 |
+
"grad_norm": 0.38370492586706956,
|
| 543 |
+
"learning_rate": 3.962436265818611e-05,
|
| 544 |
+
"loss": 0.4195,
|
| 545 |
+
"loss_nan_ranks": 0,
|
| 546 |
+
"loss_rank_avg": 0.22604212164878845,
|
| 547 |
+
"step": 245,
|
| 548 |
+
"valid_targets_mean": 4141.8,
|
| 549 |
+
"valid_targets_min": 832
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.8,
|
| 553 |
+
"grad_norm": 0.3352818127708385,
|
| 554 |
+
"learning_rate": 3.9580101287523105e-05,
|
| 555 |
+
"loss": 0.4106,
|
| 556 |
+
"loss_nan_ranks": 0,
|
| 557 |
+
"loss_rank_avg": 0.13106878101825714,
|
| 558 |
+
"step": 250,
|
| 559 |
+
"valid_targets_mean": 2769.6,
|
| 560 |
+
"valid_targets_min": 632
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.816,
|
| 564 |
+
"grad_norm": 0.35918802953549456,
|
| 565 |
+
"learning_rate": 3.953340297664928e-05,
|
| 566 |
+
"loss": 0.4023,
|
| 567 |
+
"loss_nan_ranks": 0,
|
| 568 |
+
"loss_rank_avg": 0.20013932883739471,
|
| 569 |
+
"step": 255,
|
| 570 |
+
"valid_targets_mean": 5036.6,
|
| 571 |
+
"valid_targets_min": 845
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.832,
|
| 575 |
+
"grad_norm": 0.46404548019076985,
|
| 576 |
+
"learning_rate": 3.948427353763829e-05,
|
| 577 |
+
"loss": 0.4259,
|
| 578 |
+
"loss_nan_ranks": 0,
|
| 579 |
+
"loss_rank_avg": 0.2261233776807785,
|
| 580 |
+
"step": 260,
|
| 581 |
+
"valid_targets_mean": 3119.8,
|
| 582 |
+
"valid_targets_min": 1146
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.848,
|
| 586 |
+
"grad_norm": 0.6034877544521583,
|
| 587 |
+
"learning_rate": 3.943271908514216e-05,
|
| 588 |
+
"loss": 0.3996,
|
| 589 |
+
"loss_nan_ranks": 0,
|
| 590 |
+
"loss_rank_avg": 0.20489788055419922,
|
| 591 |
+
"step": 265,
|
| 592 |
+
"valid_targets_mean": 3835.4,
|
| 593 |
+
"valid_targets_min": 585
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.864,
|
| 597 |
+
"grad_norm": 0.3776305065926428,
|
| 598 |
+
"learning_rate": 3.937874603563015e-05,
|
| 599 |
+
"loss": 0.3949,
|
| 600 |
+
"loss_nan_ranks": 0,
|
| 601 |
+
"loss_rank_avg": 0.17166143655776978,
|
| 602 |
+
"step": 270,
|
| 603 |
+
"valid_targets_mean": 4574.4,
|
| 604 |
+
"valid_targets_min": 574
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.88,
|
| 608 |
+
"grad_norm": 0.43899093850067783,
|
| 609 |
+
"learning_rate": 3.932236110659023e-05,
|
| 610 |
+
"loss": 0.4057,
|
| 611 |
+
"loss_nan_ranks": 0,
|
| 612 |
+
"loss_rank_avg": 0.1660895049571991,
|
| 613 |
+
"step": 275,
|
| 614 |
+
"valid_targets_mean": 2570.5,
|
| 615 |
+
"valid_targets_min": 852
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.896,
|
| 619 |
+
"grad_norm": 0.4836232981284088,
|
| 620 |
+
"learning_rate": 3.9263571315692976e-05,
|
| 621 |
+
"loss": 0.42,
|
| 622 |
+
"loss_nan_ranks": 0,
|
| 623 |
+
"loss_rank_avg": 0.1986091136932373,
|
| 624 |
+
"step": 280,
|
| 625 |
+
"valid_targets_mean": 4396.8,
|
| 626 |
+
"valid_targets_min": 779
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.912,
|
| 630 |
+
"grad_norm": 0.4429870800850433,
|
| 631 |
+
"learning_rate": 3.920238397991818e-05,
|
| 632 |
+
"loss": 0.4315,
|
| 633 |
+
"loss_nan_ranks": 0,
|
| 634 |
+
"loss_rank_avg": 0.188503235578537,
|
| 635 |
+
"step": 285,
|
| 636 |
+
"valid_targets_mean": 2363.9,
|
| 637 |
+
"valid_targets_min": 489
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 0.928,
|
| 641 |
+
"grad_norm": 0.4051710874433319,
|
| 642 |
+
"learning_rate": 3.913880671464418e-05,
|
| 643 |
+
"loss": 0.4256,
|
| 644 |
+
"loss_nan_ranks": 0,
|
| 645 |
+
"loss_rank_avg": 0.1489955335855484,
|
| 646 |
+
"step": 290,
|
| 647 |
+
"valid_targets_mean": 2600.9,
|
| 648 |
+
"valid_targets_min": 884
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.944,
|
| 652 |
+
"grad_norm": 0.3825741990919869,
|
| 653 |
+
"learning_rate": 3.907284743270001e-05,
|
| 654 |
+
"loss": 0.3981,
|
| 655 |
+
"loss_nan_ranks": 0,
|
| 656 |
+
"loss_rank_avg": 0.16852974891662598,
|
| 657 |
+
"step": 295,
|
| 658 |
+
"valid_targets_mean": 2941.3,
|
| 659 |
+
"valid_targets_min": 610
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 0.96,
|
| 663 |
+
"grad_norm": 0.35337156229499256,
|
| 664 |
+
"learning_rate": 3.900451434338062e-05,
|
| 665 |
+
"loss": 0.4173,
|
| 666 |
+
"loss_nan_ranks": 0,
|
| 667 |
+
"loss_rank_avg": 0.19359055161476135,
|
| 668 |
+
"step": 300,
|
| 669 |
+
"valid_targets_mean": 5110.8,
|
| 670 |
+
"valid_targets_min": 725
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.976,
|
| 674 |
+
"grad_norm": 0.3630355820027396,
|
| 675 |
+
"learning_rate": 3.893381595142511e-05,
|
| 676 |
+
"loss": 0.4033,
|
| 677 |
+
"loss_nan_ranks": 0,
|
| 678 |
+
"loss_rank_avg": 0.2465812861919403,
|
| 679 |
+
"step": 305,
|
| 680 |
+
"valid_targets_mean": 4703.4,
|
| 681 |
+
"valid_targets_min": 679
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.992,
|
| 685 |
+
"grad_norm": 0.332578670862645,
|
| 686 |
+
"learning_rate": 3.886076105595825e-05,
|
| 687 |
+
"loss": 0.3935,
|
| 688 |
+
"loss_nan_ranks": 0,
|
| 689 |
+
"loss_rank_avg": 0.16100963950157166,
|
| 690 |
+
"step": 310,
|
| 691 |
+
"valid_targets_mean": 4606.3,
|
| 692 |
+
"valid_targets_min": 771
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.0064,
|
| 696 |
+
"grad_norm": 0.3699048218897067,
|
| 697 |
+
"learning_rate": 3.878535874939532e-05,
|
| 698 |
+
"loss": 0.3777,
|
| 699 |
+
"loss_nan_ranks": 0,
|
| 700 |
+
"loss_rank_avg": 0.19179093837738037,
|
| 701 |
+
"step": 315,
|
| 702 |
+
"valid_targets_mean": 4326.3,
|
| 703 |
+
"valid_targets_min": 1151
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.0224,
|
| 707 |
+
"grad_norm": 0.35993800947449023,
|
| 708 |
+
"learning_rate": 3.870761841631051e-05,
|
| 709 |
+
"loss": 0.3994,
|
| 710 |
+
"loss_nan_ranks": 0,
|
| 711 |
+
"loss_rank_avg": 0.17964524030685425,
|
| 712 |
+
"step": 320,
|
| 713 |
+
"valid_targets_mean": 3957.6,
|
| 714 |
+
"valid_targets_min": 703
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.0384,
|
| 718 |
+
"grad_norm": 0.4176440488699317,
|
| 719 |
+
"learning_rate": 3.862754973226887e-05,
|
| 720 |
+
"loss": 0.3892,
|
| 721 |
+
"loss_nan_ranks": 0,
|
| 722 |
+
"loss_rank_avg": 0.15542760491371155,
|
| 723 |
+
"step": 325,
|
| 724 |
+
"valid_targets_mean": 2751.8,
|
| 725 |
+
"valid_targets_min": 379
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.0544,
|
| 729 |
+
"grad_norm": 0.40074497319972846,
|
| 730 |
+
"learning_rate": 3.85451626626221e-05,
|
| 731 |
+
"loss": 0.3831,
|
| 732 |
+
"loss_nan_ranks": 0,
|
| 733 |
+
"loss_rank_avg": 0.17138831317424774,
|
| 734 |
+
"step": 330,
|
| 735 |
+
"valid_targets_mean": 2887.1,
|
| 736 |
+
"valid_targets_min": 871
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.0704,
|
| 740 |
+
"grad_norm": 0.34919217958278553,
|
| 741 |
+
"learning_rate": 3.846046746126827e-05,
|
| 742 |
+
"loss": 0.3805,
|
| 743 |
+
"loss_nan_ranks": 0,
|
| 744 |
+
"loss_rank_avg": 0.1968536376953125,
|
| 745 |
+
"step": 335,
|
| 746 |
+
"valid_targets_mean": 5225.8,
|
| 747 |
+
"valid_targets_min": 766
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 1.0864,
|
| 751 |
+
"grad_norm": 0.3824861551768386,
|
| 752 |
+
"learning_rate": 3.837347466937562e-05,
|
| 753 |
+
"loss": 0.394,
|
| 754 |
+
"loss_nan_ranks": 0,
|
| 755 |
+
"loss_rank_avg": 0.1652269959449768,
|
| 756 |
+
"step": 340,
|
| 757 |
+
"valid_targets_mean": 3592.2,
|
| 758 |
+
"valid_targets_min": 371
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.1024,
|
| 762 |
+
"grad_norm": 0.48727554921232685,
|
| 763 |
+
"learning_rate": 3.828419511407062e-05,
|
| 764 |
+
"loss": 0.3926,
|
| 765 |
+
"loss_nan_ranks": 0,
|
| 766 |
+
"loss_rank_avg": 0.2775905728340149,
|
| 767 |
+
"step": 345,
|
| 768 |
+
"valid_targets_mean": 3552.9,
|
| 769 |
+
"valid_targets_min": 790
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.1184,
|
| 773 |
+
"grad_norm": 0.331694685748023,
|
| 774 |
+
"learning_rate": 3.819263990709037e-05,
|
| 775 |
+
"loss": 0.3926,
|
| 776 |
+
"loss_nan_ranks": 0,
|
| 777 |
+
"loss_rank_avg": 0.17502962052822113,
|
| 778 |
+
"step": 350,
|
| 779 |
+
"valid_targets_mean": 4362.4,
|
| 780 |
+
"valid_targets_min": 934
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.1344,
|
| 784 |
+
"grad_norm": 0.4491397219899234,
|
| 785 |
+
"learning_rate": 3.809882044339971e-05,
|
| 786 |
+
"loss": 0.4077,
|
| 787 |
+
"loss_nan_ranks": 0,
|
| 788 |
+
"loss_rank_avg": 0.19338899850845337,
|
| 789 |
+
"step": 355,
|
| 790 |
+
"valid_targets_mean": 3287.8,
|
| 791 |
+
"valid_targets_min": 630
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 1.1504,
|
| 795 |
+
"grad_norm": 0.34406293508467406,
|
| 796 |
+
"learning_rate": 3.800274839977293e-05,
|
| 797 |
+
"loss": 0.3706,
|
| 798 |
+
"loss_nan_ranks": 0,
|
| 799 |
+
"loss_rank_avg": 0.18019019067287445,
|
| 800 |
+
"step": 360,
|
| 801 |
+
"valid_targets_mean": 5227.4,
|
| 802 |
+
"valid_targets_min": 764
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.1663999999999999,
|
| 806 |
+
"grad_norm": 0.38705503675482716,
|
| 807 |
+
"learning_rate": 3.790443573334055e-05,
|
| 808 |
+
"loss": 0.3704,
|
| 809 |
+
"loss_nan_ranks": 0,
|
| 810 |
+
"loss_rank_avg": 0.18869172036647797,
|
| 811 |
+
"step": 365,
|
| 812 |
+
"valid_targets_mean": 4310.2,
|
| 813 |
+
"valid_targets_min": 760
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.1824,
|
| 817 |
+
"grad_norm": 0.42059597448080804,
|
| 818 |
+
"learning_rate": 3.780389468010106e-05,
|
| 819 |
+
"loss": 0.3843,
|
| 820 |
+
"loss_nan_ranks": 0,
|
| 821 |
+
"loss_rank_avg": 0.2906254827976227,
|
| 822 |
+
"step": 370,
|
| 823 |
+
"valid_targets_mean": 5065.8,
|
| 824 |
+
"valid_targets_min": 939
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.1984,
|
| 828 |
+
"grad_norm": 0.3896344271054351,
|
| 829 |
+
"learning_rate": 3.7701137753398075e-05,
|
| 830 |
+
"loss": 0.3806,
|
| 831 |
+
"loss_nan_ranks": 0,
|
| 832 |
+
"loss_rank_avg": 0.21188490092754364,
|
| 833 |
+
"step": 375,
|
| 834 |
+
"valid_targets_mean": 4391.9,
|
| 835 |
+
"valid_targets_min": 816
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 1.2144,
|
| 839 |
+
"grad_norm": 0.35064060810634223,
|
| 840 |
+
"learning_rate": 3.759617774236292e-05,
|
| 841 |
+
"loss": 0.3849,
|
| 842 |
+
"loss_nan_ranks": 0,
|
| 843 |
+
"loss_rank_avg": 0.21180033683776855,
|
| 844 |
+
"step": 380,
|
| 845 |
+
"valid_targets_mean": 5338.6,
|
| 846 |
+
"valid_targets_min": 805
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.2304,
|
| 850 |
+
"grad_norm": 0.3999618540862411,
|
| 851 |
+
"learning_rate": 3.748902771032288e-05,
|
| 852 |
+
"loss": 0.3863,
|
| 853 |
+
"loss_nan_ranks": 0,
|
| 854 |
+
"loss_rank_avg": 0.14589013159275055,
|
| 855 |
+
"step": 385,
|
| 856 |
+
"valid_targets_mean": 2809.4,
|
| 857 |
+
"valid_targets_min": 718
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 1.2464,
|
| 861 |
+
"grad_norm": 0.3425508807463841,
|
| 862 |
+
"learning_rate": 3.737970099317535e-05,
|
| 863 |
+
"loss": 0.3805,
|
| 864 |
+
"loss_nan_ranks": 0,
|
| 865 |
+
"loss_rank_avg": 0.19089946150779724,
|
| 866 |
+
"step": 390,
|
| 867 |
+
"valid_targets_mean": 4076.4,
|
| 868 |
+
"valid_targets_min": 404
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.2624,
|
| 872 |
+
"grad_norm": 0.3232819729746728,
|
| 873 |
+
"learning_rate": 3.726821119772803e-05,
|
| 874 |
+
"loss": 0.3981,
|
| 875 |
+
"loss_nan_ranks": 0,
|
| 876 |
+
"loss_rank_avg": 0.18869639933109283,
|
| 877 |
+
"step": 395,
|
| 878 |
+
"valid_targets_mean": 4955.0,
|
| 879 |
+
"valid_targets_min": 779
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.2784,
|
| 883 |
+
"grad_norm": 0.32513134903898305,
|
| 884 |
+
"learning_rate": 3.7154572200005446e-05,
|
| 885 |
+
"loss": 0.3655,
|
| 886 |
+
"loss_nan_ranks": 0,
|
| 887 |
+
"loss_rank_avg": 0.158743754029274,
|
| 888 |
+
"step": 400,
|
| 889 |
+
"valid_targets_mean": 4409.9,
|
| 890 |
+
"valid_targets_min": 673
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.2944,
|
| 894 |
+
"grad_norm": 0.41954006335784794,
|
| 895 |
+
"learning_rate": 3.703879814352193e-05,
|
| 896 |
+
"loss": 0.3948,
|
| 897 |
+
"loss_nan_ranks": 0,
|
| 898 |
+
"loss_rank_avg": 0.2265087366104126,
|
| 899 |
+
"step": 405,
|
| 900 |
+
"valid_targets_mean": 4266.3,
|
| 901 |
+
"valid_targets_min": 649
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.3104,
|
| 905 |
+
"grad_norm": 0.31152266082551305,
|
| 906 |
+
"learning_rate": 3.6920903437521305e-05,
|
| 907 |
+
"loss": 0.3539,
|
| 908 |
+
"loss_nan_ranks": 0,
|
| 909 |
+
"loss_rank_avg": 0.14884796738624573,
|
| 910 |
+
"step": 410,
|
| 911 |
+
"valid_targets_mean": 4143.8,
|
| 912 |
+
"valid_targets_min": 645
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.3264,
|
| 916 |
+
"grad_norm": 0.3874174052877045,
|
| 917 |
+
"learning_rate": 3.680090275518352e-05,
|
| 918 |
+
"loss": 0.39,
|
| 919 |
+
"loss_nan_ranks": 0,
|
| 920 |
+
"loss_rank_avg": 0.14476105570793152,
|
| 921 |
+
"step": 415,
|
| 922 |
+
"valid_targets_mean": 2301.5,
|
| 923 |
+
"valid_targets_min": 979
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.3424,
|
| 927 |
+
"grad_norm": 0.4076202345092236,
|
| 928 |
+
"learning_rate": 3.667881103179844e-05,
|
| 929 |
+
"loss": 0.406,
|
| 930 |
+
"loss_nan_ranks": 0,
|
| 931 |
+
"loss_rank_avg": 0.1450413465499878,
|
| 932 |
+
"step": 420,
|
| 933 |
+
"valid_targets_mean": 2344.8,
|
| 934 |
+
"valid_targets_min": 764
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.3584,
|
| 938 |
+
"grad_norm": 0.36629328269123096,
|
| 939 |
+
"learning_rate": 3.655464346290697e-05,
|
| 940 |
+
"loss": 0.3553,
|
| 941 |
+
"loss_nan_ranks": 0,
|
| 942 |
+
"loss_rank_avg": 0.1950920820236206,
|
| 943 |
+
"step": 425,
|
| 944 |
+
"valid_targets_mean": 3503.9,
|
| 945 |
+
"valid_targets_min": 786
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1.3744,
|
| 949 |
+
"grad_norm": 0.41610685727170865,
|
| 950 |
+
"learning_rate": 3.642841550240983e-05,
|
| 951 |
+
"loss": 0.3875,
|
| 952 |
+
"loss_nan_ranks": 0,
|
| 953 |
+
"loss_rank_avg": 0.1455891728401184,
|
| 954 |
+
"step": 430,
|
| 955 |
+
"valid_targets_mean": 2554.6,
|
| 956 |
+
"valid_targets_min": 799
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.3904,
|
| 960 |
+
"grad_norm": 0.366729970723816,
|
| 961 |
+
"learning_rate": 3.630014286064419e-05,
|
| 962 |
+
"loss": 0.3591,
|
| 963 |
+
"loss_nan_ranks": 0,
|
| 964 |
+
"loss_rank_avg": 0.1807524859905243,
|
| 965 |
+
"step": 435,
|
| 966 |
+
"valid_targets_mean": 3847.1,
|
| 967 |
+
"valid_targets_min": 689
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.4064,
|
| 971 |
+
"grad_norm": 0.32794983321715615,
|
| 972 |
+
"learning_rate": 3.6169841502428285e-05,
|
| 973 |
+
"loss": 0.3735,
|
| 974 |
+
"loss_nan_ranks": 0,
|
| 975 |
+
"loss_rank_avg": 0.14754189550876617,
|
| 976 |
+
"step": 440,
|
| 977 |
+
"valid_targets_mean": 3818.2,
|
| 978 |
+
"valid_targets_min": 541
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.4224,
|
| 982 |
+
"grad_norm": 0.44011771339583505,
|
| 983 |
+
"learning_rate": 3.603752764507454e-05,
|
| 984 |
+
"loss": 0.4034,
|
| 985 |
+
"loss_nan_ranks": 0,
|
| 986 |
+
"loss_rank_avg": 0.20290511846542358,
|
| 987 |
+
"step": 445,
|
| 988 |
+
"valid_targets_mean": 2867.4,
|
| 989 |
+
"valid_targets_min": 924
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.4384000000000001,
|
| 993 |
+
"grad_norm": 0.3875724203422835,
|
| 994 |
+
"learning_rate": 3.5903217756371066e-05,
|
| 995 |
+
"loss": 0.39,
|
| 996 |
+
"loss_nan_ranks": 0,
|
| 997 |
+
"loss_rank_avg": 0.24192467331886292,
|
| 998 |
+
"step": 450,
|
| 999 |
+
"valid_targets_mean": 4539.0,
|
| 1000 |
+
"valid_targets_min": 686
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.4544000000000001,
|
| 1004 |
+
"grad_norm": 0.35423009985165277,
|
| 1005 |
+
"learning_rate": 3.576692855253213e-05,
|
| 1006 |
+
"loss": 0.3669,
|
| 1007 |
+
"loss_nan_ranks": 0,
|
| 1008 |
+
"loss_rank_avg": 0.1834782361984253,
|
| 1009 |
+
"step": 455,
|
| 1010 |
+
"valid_targets_mean": 4385.8,
|
| 1011 |
+
"valid_targets_min": 740
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 1.4704,
|
| 1015 |
+
"grad_norm": 0.34329690274477853,
|
| 1016 |
+
"learning_rate": 3.562867699611764e-05,
|
| 1017 |
+
"loss": 0.3522,
|
| 1018 |
+
"loss_nan_ranks": 0,
|
| 1019 |
+
"loss_rank_avg": 0.16957998275756836,
|
| 1020 |
+
"step": 460,
|
| 1021 |
+
"valid_targets_mean": 4358.4,
|
| 1022 |
+
"valid_targets_min": 705
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.4864,
|
| 1026 |
+
"grad_norm": 0.4969730194262121,
|
| 1027 |
+
"learning_rate": 3.5488480293922e-05,
|
| 1028 |
+
"loss": 0.3526,
|
| 1029 |
+
"loss_nan_ranks": 0,
|
| 1030 |
+
"loss_rank_avg": 0.14323380589485168,
|
| 1031 |
+
"step": 465,
|
| 1032 |
+
"valid_targets_mean": 3332.0,
|
| 1033 |
+
"valid_targets_min": 979
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 1.5024,
|
| 1037 |
+
"grad_norm": 0.3878219097905555,
|
| 1038 |
+
"learning_rate": 3.5346355894832515e-05,
|
| 1039 |
+
"loss": 0.3929,
|
| 1040 |
+
"loss_nan_ranks": 0,
|
| 1041 |
+
"loss_rank_avg": 0.25531578063964844,
|
| 1042 |
+
"step": 470,
|
| 1043 |
+
"valid_targets_mean": 4687.0,
|
| 1044 |
+
"valid_targets_min": 699
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.5184,
|
| 1048 |
+
"grad_norm": 0.41901557134182976,
|
| 1049 |
+
"learning_rate": 3.520232148765774e-05,
|
| 1050 |
+
"loss": 0.3579,
|
| 1051 |
+
"loss_nan_ranks": 0,
|
| 1052 |
+
"loss_rank_avg": 0.15701517462730408,
|
| 1053 |
+
"step": 475,
|
| 1054 |
+
"valid_targets_mean": 3616.7,
|
| 1055 |
+
"valid_targets_min": 660
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 1.5344,
|
| 1059 |
+
"grad_norm": 0.3337566112172416,
|
| 1060 |
+
"learning_rate": 3.505639499892591e-05,
|
| 1061 |
+
"loss": 0.3606,
|
| 1062 |
+
"loss_nan_ranks": 0,
|
| 1063 |
+
"loss_rank_avg": 0.14276567101478577,
|
| 1064 |
+
"step": 480,
|
| 1065 |
+
"valid_targets_mean": 4460.1,
|
| 1066 |
+
"valid_targets_min": 705
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.5504,
|
| 1070 |
+
"grad_norm": 0.376009534019795,
|
| 1071 |
+
"learning_rate": 3.490859459065382e-05,
|
| 1072 |
+
"loss": 0.3762,
|
| 1073 |
+
"loss_nan_ranks": 0,
|
| 1074 |
+
"loss_rank_avg": 0.18227124214172363,
|
| 1075 |
+
"step": 485,
|
| 1076 |
+
"valid_targets_mean": 4140.9,
|
| 1077 |
+
"valid_targets_min": 927
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 1.5664,
|
| 1081 |
+
"grad_norm": 0.3657848949268241,
|
| 1082 |
+
"learning_rate": 3.475893865808633e-05,
|
| 1083 |
+
"loss": 0.3859,
|
| 1084 |
+
"loss_nan_ranks": 0,
|
| 1085 |
+
"loss_rank_avg": 0.1826404631137848,
|
| 1086 |
+
"step": 490,
|
| 1087 |
+
"valid_targets_mean": 3706.3,
|
| 1088 |
+
"valid_targets_min": 692
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.5824,
|
| 1092 |
+
"grad_norm": 0.4325898106752329,
|
| 1093 |
+
"learning_rate": 3.4607445827406984e-05,
|
| 1094 |
+
"loss": 0.4181,
|
| 1095 |
+
"loss_nan_ranks": 0,
|
| 1096 |
+
"loss_rank_avg": 0.23097142577171326,
|
| 1097 |
+
"step": 495,
|
| 1098 |
+
"valid_targets_mean": 3480.9,
|
| 1099 |
+
"valid_targets_min": 832
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 1.5984,
|
| 1103 |
+
"grad_norm": 0.2802872024830924,
|
| 1104 |
+
"learning_rate": 3.445413495341971e-05,
|
| 1105 |
+
"loss": 0.363,
|
| 1106 |
+
"loss_nan_ranks": 0,
|
| 1107 |
+
"loss_rank_avg": 0.13881602883338928,
|
| 1108 |
+
"step": 500,
|
| 1109 |
+
"valid_targets_mean": 5370.1,
|
| 1110 |
+
"valid_targets_min": 430
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 1.6143999999999998,
|
| 1114 |
+
"grad_norm": 0.43602069232557744,
|
| 1115 |
+
"learning_rate": 3.429902511720216e-05,
|
| 1116 |
+
"loss": 0.3927,
|
| 1117 |
+
"loss_nan_ranks": 0,
|
| 1118 |
+
"loss_rank_avg": 0.23958256840705872,
|
| 1119 |
+
"step": 505,
|
| 1120 |
+
"valid_targets_mean": 3277.2,
|
| 1121 |
+
"valid_targets_min": 1050
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 1.6303999999999998,
|
| 1125 |
+
"grad_norm": 0.41936014097791685,
|
| 1126 |
+
"learning_rate": 3.4142135623730954e-05,
|
| 1127 |
+
"loss": 0.3791,
|
| 1128 |
+
"loss_nan_ranks": 0,
|
| 1129 |
+
"loss_rank_avg": 0.15627846121788025,
|
| 1130 |
+
"step": 510,
|
| 1131 |
+
"valid_targets_mean": 4432.1,
|
| 1132 |
+
"valid_targets_min": 558
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 1.6463999999999999,
|
| 1136 |
+
"grad_norm": 0.3358679962436761,
|
| 1137 |
+
"learning_rate": 3.398348599947888e-05,
|
| 1138 |
+
"loss": 0.3754,
|
| 1139 |
+
"loss_nan_ranks": 0,
|
| 1140 |
+
"loss_rank_avg": 0.132685586810112,
|
| 1141 |
+
"step": 515,
|
| 1142 |
+
"valid_targets_mean": 4297.8,
|
| 1143 |
+
"valid_targets_min": 707
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.6623999999999999,
|
| 1147 |
+
"grad_norm": 0.338992772693464,
|
| 1148 |
+
"learning_rate": 3.3823095989984697e-05,
|
| 1149 |
+
"loss": 0.41,
|
| 1150 |
+
"loss_nan_ranks": 0,
|
| 1151 |
+
"loss_rank_avg": 0.1532086730003357,
|
| 1152 |
+
"step": 520,
|
| 1153 |
+
"valid_targets_mean": 3837.9,
|
| 1154 |
+
"valid_targets_min": 779
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.6784,
|
| 1158 |
+
"grad_norm": 0.39162023399835316,
|
| 1159 |
+
"learning_rate": 3.366098555739557e-05,
|
| 1160 |
+
"loss": 0.3623,
|
| 1161 |
+
"loss_nan_ranks": 0,
|
| 1162 |
+
"loss_rank_avg": 0.16554811596870422,
|
| 1163 |
+
"step": 525,
|
| 1164 |
+
"valid_targets_mean": 4507.1,
|
| 1165 |
+
"valid_targets_min": 867
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 1.6944,
|
| 1169 |
+
"grad_norm": 0.41290820076532175,
|
| 1170 |
+
"learning_rate": 3.349717487798261e-05,
|
| 1171 |
+
"loss": 0.3722,
|
| 1172 |
+
"loss_nan_ranks": 0,
|
| 1173 |
+
"loss_rank_avg": 0.20561575889587402,
|
| 1174 |
+
"step": 530,
|
| 1175 |
+
"valid_targets_mean": 3697.0,
|
| 1176 |
+
"valid_targets_min": 764
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 1.7104,
|
| 1180 |
+
"grad_norm": 0.3494036698637042,
|
| 1181 |
+
"learning_rate": 3.3331684339629706e-05,
|
| 1182 |
+
"loss": 0.3734,
|
| 1183 |
+
"loss_nan_ranks": 0,
|
| 1184 |
+
"loss_rank_avg": 0.15987402200698853,
|
| 1185 |
+
"step": 535,
|
| 1186 |
+
"valid_targets_mean": 4119.2,
|
| 1187 |
+
"valid_targets_min": 668
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 1.7264,
|
| 1191 |
+
"grad_norm": 0.47807369844807185,
|
| 1192 |
+
"learning_rate": 3.3164534539296056e-05,
|
| 1193 |
+
"loss": 0.3747,
|
| 1194 |
+
"loss_nan_ranks": 0,
|
| 1195 |
+
"loss_rank_avg": 0.16826443374156952,
|
| 1196 |
+
"step": 540,
|
| 1197 |
+
"valid_targets_mean": 3707.9,
|
| 1198 |
+
"valid_targets_min": 763
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 1.7424,
|
| 1202 |
+
"grad_norm": 0.3361689358143466,
|
| 1203 |
+
"learning_rate": 3.299574628045269e-05,
|
| 1204 |
+
"loss": 0.3615,
|
| 1205 |
+
"loss_nan_ranks": 0,
|
| 1206 |
+
"loss_rank_avg": 0.1361405998468399,
|
| 1207 |
+
"step": 545,
|
| 1208 |
+
"valid_targets_mean": 3589.8,
|
| 1209 |
+
"valid_targets_min": 599
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.7584,
|
| 1213 |
+
"grad_norm": 0.37484427799298453,
|
| 1214 |
+
"learning_rate": 3.282534057049322e-05,
|
| 1215 |
+
"loss": 0.3349,
|
| 1216 |
+
"loss_nan_ranks": 0,
|
| 1217 |
+
"loss_rank_avg": 0.16627568006515503,
|
| 1218 |
+
"step": 550,
|
| 1219 |
+
"valid_targets_mean": 3740.7,
|
| 1220 |
+
"valid_targets_min": 957
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.7744,
|
| 1224 |
+
"grad_norm": 0.37363103188547103,
|
| 1225 |
+
"learning_rate": 3.265333861811933e-05,
|
| 1226 |
+
"loss": 0.3792,
|
| 1227 |
+
"loss_nan_ranks": 0,
|
| 1228 |
+
"loss_rank_avg": 0.18092834949493408,
|
| 1229 |
+
"step": 555,
|
| 1230 |
+
"valid_targets_mean": 3324.7,
|
| 1231 |
+
"valid_targets_min": 695
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 1.7904,
|
| 1235 |
+
"grad_norm": 0.3839625364420311,
|
| 1236 |
+
"learning_rate": 3.2479761830701075e-05,
|
| 1237 |
+
"loss": 0.3918,
|
| 1238 |
+
"loss_nan_ranks": 0,
|
| 1239 |
+
"loss_rank_avg": 0.21519896388053894,
|
| 1240 |
+
"step": 560,
|
| 1241 |
+
"valid_targets_mean": 3921.2,
|
| 1242 |
+
"valid_targets_min": 606
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.8064,
|
| 1246 |
+
"grad_norm": 0.37568764545151495,
|
| 1247 |
+
"learning_rate": 3.230463181161254e-05,
|
| 1248 |
+
"loss": 0.3933,
|
| 1249 |
+
"loss_nan_ranks": 0,
|
| 1250 |
+
"loss_rank_avg": 0.1830562949180603,
|
| 1251 |
+
"step": 565,
|
| 1252 |
+
"valid_targets_mean": 3310.1,
|
| 1253 |
+
"valid_targets_min": 711
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 1.8224,
|
| 1257 |
+
"grad_norm": 0.3288521673537427,
|
| 1258 |
+
"learning_rate": 3.212797035754311e-05,
|
| 1259 |
+
"loss": 0.3873,
|
| 1260 |
+
"loss_nan_ranks": 0,
|
| 1261 |
+
"loss_rank_avg": 0.1765051782131195,
|
| 1262 |
+
"step": 570,
|
| 1263 |
+
"valid_targets_mean": 4186.1,
|
| 1264 |
+
"valid_targets_min": 656
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 1.8384,
|
| 1268 |
+
"grad_norm": 0.327471713865092,
|
| 1269 |
+
"learning_rate": 3.194979945578461e-05,
|
| 1270 |
+
"loss": 0.3926,
|
| 1271 |
+
"loss_nan_ranks": 0,
|
| 1272 |
+
"loss_rank_avg": 0.18825772404670715,
|
| 1273 |
+
"step": 575,
|
| 1274 |
+
"valid_targets_mean": 5375.1,
|
| 1275 |
+
"valid_targets_min": 724
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.8544,
|
| 1279 |
+
"grad_norm": 0.3373202744506304,
|
| 1280 |
+
"learning_rate": 3.177014128149479e-05,
|
| 1281 |
+
"loss": 0.3546,
|
| 1282 |
+
"loss_nan_ranks": 0,
|
| 1283 |
+
"loss_rank_avg": 0.19015300273895264,
|
| 1284 |
+
"step": 580,
|
| 1285 |
+
"valid_targets_mean": 4340.6,
|
| 1286 |
+
"valid_targets_min": 679
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 1.8704,
|
| 1290 |
+
"grad_norm": 0.3469982372901262,
|
| 1291 |
+
"learning_rate": 3.158901819493742e-05,
|
| 1292 |
+
"loss": 0.3814,
|
| 1293 |
+
"loss_nan_ranks": 0,
|
| 1294 |
+
"loss_rank_avg": 0.22717589139938354,
|
| 1295 |
+
"step": 585,
|
| 1296 |
+
"valid_targets_mean": 6144.8,
|
| 1297 |
+
"valid_targets_min": 626
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.8864,
|
| 1301 |
+
"grad_norm": 0.37258492423632356,
|
| 1302 |
+
"learning_rate": 3.1406452738699284e-05,
|
| 1303 |
+
"loss": 0.3693,
|
| 1304 |
+
"loss_nan_ranks": 0,
|
| 1305 |
+
"loss_rank_avg": 0.1856401562690735,
|
| 1306 |
+
"step": 590,
|
| 1307 |
+
"valid_targets_mean": 3395.8,
|
| 1308 |
+
"valid_targets_min": 659
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 1.9024,
|
| 1312 |
+
"grad_norm": 0.3379882445810335,
|
| 1313 |
+
"learning_rate": 3.122246763488457e-05,
|
| 1314 |
+
"loss": 0.3629,
|
| 1315 |
+
"loss_nan_ranks": 0,
|
| 1316 |
+
"loss_rank_avg": 0.1680774986743927,
|
| 1317 |
+
"step": 595,
|
| 1318 |
+
"valid_targets_mean": 4609.5,
|
| 1319 |
+
"valid_targets_min": 585
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.9184,
|
| 1323 |
+
"grad_norm": 0.3436516413868104,
|
| 1324 |
+
"learning_rate": 3.103708578228686e-05,
|
| 1325 |
+
"loss": 0.3835,
|
| 1326 |
+
"loss_nan_ranks": 0,
|
| 1327 |
+
"loss_rank_avg": 0.1666383296251297,
|
| 1328 |
+
"step": 600,
|
| 1329 |
+
"valid_targets_mean": 3579.3,
|
| 1330 |
+
"valid_targets_min": 943
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 1.9344000000000001,
|
| 1334 |
+
"grad_norm": 0.4313763145949012,
|
| 1335 |
+
"learning_rate": 3.085033025353915e-05,
|
| 1336 |
+
"loss": 0.3808,
|
| 1337 |
+
"loss_nan_ranks": 0,
|
| 1338 |
+
"loss_rank_avg": 0.19704148173332214,
|
| 1339 |
+
"step": 605,
|
| 1340 |
+
"valid_targets_mean": 2658.1,
|
| 1341 |
+
"valid_targets_min": 639
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.9504000000000001,
|
| 1345 |
+
"grad_norm": 0.37657788301827405,
|
| 1346 |
+
"learning_rate": 3.066222429224221e-05,
|
| 1347 |
+
"loss": 0.3776,
|
| 1348 |
+
"loss_nan_ranks": 0,
|
| 1349 |
+
"loss_rank_avg": 0.16659879684448242,
|
| 1350 |
+
"step": 610,
|
| 1351 |
+
"valid_targets_mean": 3304.2,
|
| 1352 |
+
"valid_targets_min": 661
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.9664000000000001,
|
| 1356 |
+
"grad_norm": 0.4058483842532848,
|
| 1357 |
+
"learning_rate": 3.047279131007173e-05,
|
| 1358 |
+
"loss": 0.3716,
|
| 1359 |
+
"loss_nan_ranks": 0,
|
| 1360 |
+
"loss_rank_avg": 0.20012781023979187,
|
| 1361 |
+
"step": 615,
|
| 1362 |
+
"valid_targets_mean": 2792.8,
|
| 1363 |
+
"valid_targets_min": 827
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 1.9824000000000002,
|
| 1367 |
+
"grad_norm": 0.412660563012823,
|
| 1368 |
+
"learning_rate": 3.0282054883864434e-05,
|
| 1369 |
+
"loss": 0.3624,
|
| 1370 |
+
"loss_nan_ranks": 0,
|
| 1371 |
+
"loss_rank_avg": 0.1812940537929535,
|
| 1372 |
+
"step": 620,
|
| 1373 |
+
"valid_targets_mean": 2277.3,
|
| 1374 |
+
"valid_targets_min": 666
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.9984,
|
| 1378 |
+
"grad_norm": 0.32332051288280694,
|
| 1379 |
+
"learning_rate": 3.009003875268379e-05,
|
| 1380 |
+
"loss": 0.3744,
|
| 1381 |
+
"loss_nan_ranks": 0,
|
| 1382 |
+
"loss_rank_avg": 0.12370559573173523,
|
| 1383 |
+
"step": 625,
|
| 1384 |
+
"valid_targets_mean": 2962.4,
|
| 1385 |
+
"valid_targets_min": 866
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 2.0128,
|
| 1389 |
+
"grad_norm": 0.3552382763657628,
|
| 1390 |
+
"learning_rate": 2.9896766814865355e-05,
|
| 1391 |
+
"loss": 0.3532,
|
| 1392 |
+
"loss_nan_ranks": 0,
|
| 1393 |
+
"loss_rank_avg": 0.14251808822155,
|
| 1394 |
+
"step": 630,
|
| 1395 |
+
"valid_targets_mean": 2889.8,
|
| 1396 |
+
"valid_targets_min": 829
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.0288,
|
| 1400 |
+
"grad_norm": 0.3663974186319015,
|
| 1401 |
+
"learning_rate": 2.970226312504246e-05,
|
| 1402 |
+
"loss": 0.3675,
|
| 1403 |
+
"loss_nan_ranks": 0,
|
| 1404 |
+
"loss_rank_avg": 0.15324479341506958,
|
| 1405 |
+
"step": 635,
|
| 1406 |
+
"valid_targets_mean": 3010.3,
|
| 1407 |
+
"valid_targets_min": 768
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 2.0448,
|
| 1411 |
+
"grad_norm": 0.39187020362125585,
|
| 1412 |
+
"learning_rate": 2.9506551891152334e-05,
|
| 1413 |
+
"loss": 0.378,
|
| 1414 |
+
"loss_nan_ranks": 0,
|
| 1415 |
+
"loss_rank_avg": 0.26267358660697937,
|
| 1416 |
+
"step": 640,
|
| 1417 |
+
"valid_targets_mean": 4856.9,
|
| 1418 |
+
"valid_targets_min": 570
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 2.0608,
|
| 1422 |
+
"grad_norm": 0.31957017059314285,
|
| 1423 |
+
"learning_rate": 2.930965747142319e-05,
|
| 1424 |
+
"loss": 0.3305,
|
| 1425 |
+
"loss_nan_ranks": 0,
|
| 1426 |
+
"loss_rank_avg": 0.16857537627220154,
|
| 1427 |
+
"step": 645,
|
| 1428 |
+
"valid_targets_mean": 4703.2,
|
| 1429 |
+
"valid_targets_min": 829
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 2.0768,
|
| 1433 |
+
"grad_norm": 0.40369627273889996,
|
| 1434 |
+
"learning_rate": 2.9111604371342593e-05,
|
| 1435 |
+
"loss": 0.337,
|
| 1436 |
+
"loss_nan_ranks": 0,
|
| 1437 |
+
"loss_rank_avg": 0.15384268760681152,
|
| 1438 |
+
"step": 650,
|
| 1439 |
+
"valid_targets_mean": 2815.1,
|
| 1440 |
+
"valid_targets_min": 699
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 2.0928,
|
| 1444 |
+
"grad_norm": 0.46412451317350417,
|
| 1445 |
+
"learning_rate": 2.891241724060752e-05,
|
| 1446 |
+
"loss": 0.3493,
|
| 1447 |
+
"loss_nan_ranks": 0,
|
| 1448 |
+
"loss_rank_avg": 0.18567049503326416,
|
| 1449 |
+
"step": 655,
|
| 1450 |
+
"valid_targets_mean": 4328.1,
|
| 1451 |
+
"valid_targets_min": 814
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 2.1088,
|
| 1455 |
+
"grad_norm": 0.354455219931927,
|
| 1456 |
+
"learning_rate": 2.8712120870056455e-05,
|
| 1457 |
+
"loss": 0.3565,
|
| 1458 |
+
"loss_nan_ranks": 0,
|
| 1459 |
+
"loss_rank_avg": 0.17629089951515198,
|
| 1460 |
+
"step": 660,
|
| 1461 |
+
"valid_targets_mean": 4897.3,
|
| 1462 |
+
"valid_targets_min": 740
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 2.1248,
|
| 1466 |
+
"grad_norm": 0.32558707469954473,
|
| 1467 |
+
"learning_rate": 2.851074018858389e-05,
|
| 1468 |
+
"loss": 0.3671,
|
| 1469 |
+
"loss_nan_ranks": 0,
|
| 1470 |
+
"loss_rank_avg": 0.17213571071624756,
|
| 1471 |
+
"step": 665,
|
| 1472 |
+
"valid_targets_mean": 5233.5,
|
| 1473 |
+
"valid_targets_min": 792
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 2.1408,
|
| 1477 |
+
"grad_norm": 0.42136092165410943,
|
| 1478 |
+
"learning_rate": 2.8308300260037734e-05,
|
| 1479 |
+
"loss": 0.3606,
|
| 1480 |
+
"loss_nan_ranks": 0,
|
| 1481 |
+
"loss_rank_avg": 0.17917993664741516,
|
| 1482 |
+
"step": 670,
|
| 1483 |
+
"valid_targets_mean": 3339.1,
|
| 1484 |
+
"valid_targets_min": 1000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 2.1568,
|
| 1488 |
+
"grad_norm": 0.3456749974395744,
|
| 1489 |
+
"learning_rate": 2.8104826280099796e-05,
|
| 1490 |
+
"loss": 0.3769,
|
| 1491 |
+
"loss_nan_ranks": 0,
|
| 1492 |
+
"loss_rank_avg": 0.16497279703617096,
|
| 1493 |
+
"step": 675,
|
| 1494 |
+
"valid_targets_mean": 4999.3,
|
| 1495 |
+
"valid_targets_min": 1058
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 2.1728,
|
| 1499 |
+
"grad_norm": 0.47636949292183406,
|
| 1500 |
+
"learning_rate": 2.7900343573150003e-05,
|
| 1501 |
+
"loss": 0.3549,
|
| 1502 |
+
"loss_nan_ranks": 0,
|
| 1503 |
+
"loss_rank_avg": 0.16123411059379578,
|
| 1504 |
+
"step": 680,
|
| 1505 |
+
"valid_targets_mean": 3640.6,
|
| 1506 |
+
"valid_targets_min": 521
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 2.1888,
|
| 1510 |
+
"grad_norm": 0.3802344366234923,
|
| 1511 |
+
"learning_rate": 2.7694877589114442e-05,
|
| 1512 |
+
"loss": 0.3675,
|
| 1513 |
+
"loss_nan_ranks": 0,
|
| 1514 |
+
"loss_rank_avg": 0.15201343595981598,
|
| 1515 |
+
"step": 685,
|
| 1516 |
+
"valid_targets_mean": 3488.5,
|
| 1517 |
+
"valid_targets_min": 661
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.2048,
|
| 1521 |
+
"grad_norm": 0.3889010529697492,
|
| 1522 |
+
"learning_rate": 2.748845390029794e-05,
|
| 1523 |
+
"loss": 0.3462,
|
| 1524 |
+
"loss_nan_ranks": 0,
|
| 1525 |
+
"loss_rank_avg": 0.136835515499115,
|
| 1526 |
+
"step": 690,
|
| 1527 |
+
"valid_targets_mean": 3073.3,
|
| 1528 |
+
"valid_targets_min": 1053
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.2208,
|
| 1532 |
+
"grad_norm": 0.31268069502172785,
|
| 1533 |
+
"learning_rate": 2.728109819820129e-05,
|
| 1534 |
+
"loss": 0.353,
|
| 1535 |
+
"loss_nan_ranks": 0,
|
| 1536 |
+
"loss_rank_avg": 0.1584779918193817,
|
| 1537 |
+
"step": 695,
|
| 1538 |
+
"valid_targets_mean": 5115.0,
|
| 1539 |
+
"valid_targets_min": 643
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 2.2368,
|
| 1543 |
+
"grad_norm": 0.3517489537279622,
|
| 1544 |
+
"learning_rate": 2.7072836290323698e-05,
|
| 1545 |
+
"loss": 0.3614,
|
| 1546 |
+
"loss_nan_ranks": 0,
|
| 1547 |
+
"loss_rank_avg": 0.13909167051315308,
|
| 1548 |
+
"step": 700,
|
| 1549 |
+
"valid_targets_mean": 4441.3,
|
| 1550 |
+
"valid_targets_min": 846
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 2.2528,
|
| 1554 |
+
"grad_norm": 0.42108941278672635,
|
| 1555 |
+
"learning_rate": 2.6863694096950763e-05,
|
| 1556 |
+
"loss": 0.3447,
|
| 1557 |
+
"loss_nan_ranks": 0,
|
| 1558 |
+
"loss_rank_avg": 0.17374692857265472,
|
| 1559 |
+
"step": 705,
|
| 1560 |
+
"valid_targets_mean": 4347.6,
|
| 1561 |
+
"valid_targets_min": 738
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 2.2688,
|
| 1565 |
+
"grad_norm": 0.3854671122527939,
|
| 1566 |
+
"learning_rate": 2.6653697647928485e-05,
|
| 1567 |
+
"loss": 0.3676,
|
| 1568 |
+
"loss_nan_ranks": 0,
|
| 1569 |
+
"loss_rank_avg": 0.18790662288665771,
|
| 1570 |
+
"step": 710,
|
| 1571 |
+
"valid_targets_mean": 3574.8,
|
| 1572 |
+
"valid_targets_min": 971
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 2.2848,
|
| 1576 |
+
"grad_norm": 0.3325381979319747,
|
| 1577 |
+
"learning_rate": 2.644287307942352e-05,
|
| 1578 |
+
"loss": 0.357,
|
| 1579 |
+
"loss_nan_ranks": 0,
|
| 1580 |
+
"loss_rank_avg": 0.2031068354845047,
|
| 1581 |
+
"step": 715,
|
| 1582 |
+
"valid_targets_mean": 6072.4,
|
| 1583 |
+
"valid_targets_min": 1032
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 2.3008,
|
| 1587 |
+
"grad_norm": 0.38008293980011126,
|
| 1588 |
+
"learning_rate": 2.623124663067034e-05,
|
| 1589 |
+
"loss": 0.3578,
|
| 1590 |
+
"loss_nan_ranks": 0,
|
| 1591 |
+
"loss_rank_avg": 0.17572256922721863,
|
| 1592 |
+
"step": 720,
|
| 1593 |
+
"valid_targets_mean": 4194.4,
|
| 1594 |
+
"valid_targets_min": 875
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 2.3168,
|
| 1598 |
+
"grad_norm": 0.4090255604288786,
|
| 1599 |
+
"learning_rate": 2.6018844640705448e-05,
|
| 1600 |
+
"loss": 0.3539,
|
| 1601 |
+
"loss_nan_ranks": 0,
|
| 1602 |
+
"loss_rank_avg": 0.18574786186218262,
|
| 1603 |
+
"step": 725,
|
| 1604 |
+
"valid_targets_mean": 5218.9,
|
| 1605 |
+
"valid_targets_min": 1056
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 2.3327999999999998,
|
| 1609 |
+
"grad_norm": 0.3695725965918268,
|
| 1610 |
+
"learning_rate": 2.580569354508925e-05,
|
| 1611 |
+
"loss": 0.3543,
|
| 1612 |
+
"loss_nan_ranks": 0,
|
| 1613 |
+
"loss_rank_avg": 0.19336247444152832,
|
| 1614 |
+
"step": 730,
|
| 1615 |
+
"valid_targets_mean": 3839.9,
|
| 1616 |
+
"valid_targets_min": 734
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 2.3487999999999998,
|
| 1620 |
+
"grad_norm": 0.4037004371230533,
|
| 1621 |
+
"learning_rate": 2.5591819872615856e-05,
|
| 1622 |
+
"loss": 0.344,
|
| 1623 |
+
"loss_nan_ranks": 0,
|
| 1624 |
+
"loss_rank_avg": 0.1890317052602768,
|
| 1625 |
+
"step": 735,
|
| 1626 |
+
"valid_targets_mean": 3146.3,
|
| 1627 |
+
"valid_targets_min": 777
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 2.3648,
|
| 1631 |
+
"grad_norm": 0.42034002047772934,
|
| 1632 |
+
"learning_rate": 2.5377250242011338e-05,
|
| 1633 |
+
"loss": 0.3671,
|
| 1634 |
+
"loss_nan_ranks": 0,
|
| 1635 |
+
"loss_rank_avg": 0.21791379153728485,
|
| 1636 |
+
"step": 740,
|
| 1637 |
+
"valid_targets_mean": 4096.5,
|
| 1638 |
+
"valid_targets_min": 744
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 2.3808,
|
| 1642 |
+
"grad_norm": 0.42506199410875983,
|
| 1643 |
+
"learning_rate": 2.516201135862073e-05,
|
| 1644 |
+
"loss": 0.3339,
|
| 1645 |
+
"loss_nan_ranks": 0,
|
| 1646 |
+
"loss_rank_avg": 0.18126055598258972,
|
| 1647 |
+
"step": 745,
|
| 1648 |
+
"valid_targets_mean": 4013.9,
|
| 1649 |
+
"valid_targets_min": 1198
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 2.3968,
|
| 1653 |
+
"grad_norm": 0.34527703907607993,
|
| 1654 |
+
"learning_rate": 2.494613001108431e-05,
|
| 1655 |
+
"loss": 0.3543,
|
| 1656 |
+
"loss_nan_ranks": 0,
|
| 1657 |
+
"loss_rank_avg": 0.12390469014644623,
|
| 1658 |
+
"step": 750,
|
| 1659 |
+
"valid_targets_mean": 3295.7,
|
| 1660 |
+
"valid_targets_min": 752
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 2.4128,
|
| 1664 |
+
"grad_norm": 0.34904761415566266,
|
| 1665 |
+
"learning_rate": 2.4729633068003466e-05,
|
| 1666 |
+
"loss": 0.3474,
|
| 1667 |
+
"loss_nan_ranks": 0,
|
| 1668 |
+
"loss_rank_avg": 0.1934797316789627,
|
| 1669 |
+
"step": 755,
|
| 1670 |
+
"valid_targets_mean": 4930.4,
|
| 1671 |
+
"valid_targets_min": 899
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 2.4288,
|
| 1675 |
+
"grad_norm": 0.38001649074269783,
|
| 1676 |
+
"learning_rate": 2.4512547474596624e-05,
|
| 1677 |
+
"loss": 0.3413,
|
| 1678 |
+
"loss_nan_ranks": 0,
|
| 1679 |
+
"loss_rank_avg": 0.2003239393234253,
|
| 1680 |
+
"step": 760,
|
| 1681 |
+
"valid_targets_mean": 3611.4,
|
| 1682 |
+
"valid_targets_min": 741
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.4448,
|
| 1686 |
+
"grad_norm": 0.3518099712444648,
|
| 1687 |
+
"learning_rate": 2.429490024934566e-05,
|
| 1688 |
+
"loss": 0.3511,
|
| 1689 |
+
"loss_nan_ranks": 0,
|
| 1690 |
+
"loss_rank_avg": 0.15470391511917114,
|
| 1691 |
+
"step": 765,
|
| 1692 |
+
"valid_targets_mean": 3514.5,
|
| 1693 |
+
"valid_targets_min": 972
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 2.4608,
|
| 1697 |
+
"grad_norm": 0.3448887159515848,
|
| 1698 |
+
"learning_rate": 2.4076718480633178e-05,
|
| 1699 |
+
"loss": 0.358,
|
| 1700 |
+
"loss_nan_ranks": 0,
|
| 1701 |
+
"loss_rank_avg": 0.16293978691101074,
|
| 1702 |
+
"step": 770,
|
| 1703 |
+
"valid_targets_mean": 3817.4,
|
| 1704 |
+
"valid_targets_min": 848
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 2.4768,
|
| 1708 |
+
"grad_norm": 0.3636861883493942,
|
| 1709 |
+
"learning_rate": 2.3858029323371067e-05,
|
| 1710 |
+
"loss": 0.3526,
|
| 1711 |
+
"loss_nan_ranks": 0,
|
| 1712 |
+
"loss_rank_avg": 0.16884376108646393,
|
| 1713 |
+
"step": 775,
|
| 1714 |
+
"valid_targets_mean": 3655.5,
|
| 1715 |
+
"valid_targets_min": 687
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 2.4928,
|
| 1719 |
+
"grad_norm": 0.36593321086782454,
|
| 1720 |
+
"learning_rate": 2.363885999562084e-05,
|
| 1721 |
+
"loss": 0.3737,
|
| 1722 |
+
"loss_nan_ranks": 0,
|
| 1723 |
+
"loss_rank_avg": 0.12894733250141144,
|
| 1724 |
+
"step": 780,
|
| 1725 |
+
"valid_targets_mean": 2824.6,
|
| 1726 |
+
"valid_targets_min": 758
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 2.5088,
|
| 1730 |
+
"grad_norm": 0.48954320211825053,
|
| 1731 |
+
"learning_rate": 2.3419237775206026e-05,
|
| 1732 |
+
"loss": 0.3356,
|
| 1733 |
+
"loss_nan_ranks": 0,
|
| 1734 |
+
"loss_rank_avg": 0.1467287838459015,
|
| 1735 |
+
"step": 785,
|
| 1736 |
+
"valid_targets_mean": 4619.7,
|
| 1737 |
+
"valid_targets_min": 613
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 2.5248,
|
| 1741 |
+
"grad_norm": 0.38004730454836627,
|
| 1742 |
+
"learning_rate": 2.3199189996317205e-05,
|
| 1743 |
+
"loss": 0.3398,
|
| 1744 |
+
"loss_nan_ranks": 0,
|
| 1745 |
+
"loss_rank_avg": 0.1634030044078827,
|
| 1746 |
+
"step": 790,
|
| 1747 |
+
"valid_targets_mean": 2998.8,
|
| 1748 |
+
"valid_targets_min": 767
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 2.5408,
|
| 1752 |
+
"grad_norm": 0.4280336673825541,
|
| 1753 |
+
"learning_rate": 2.297874404610998e-05,
|
| 1754 |
+
"loss": 0.3531,
|
| 1755 |
+
"loss_nan_ranks": 0,
|
| 1756 |
+
"loss_rank_avg": 0.15717938542366028,
|
| 1757 |
+
"step": 795,
|
| 1758 |
+
"valid_targets_mean": 2775.2,
|
| 1759 |
+
"valid_targets_min": 679
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 2.5568,
|
| 1763 |
+
"grad_norm": 0.3977995070636329,
|
| 1764 |
+
"learning_rate": 2.2757927361296376e-05,
|
| 1765 |
+
"loss": 0.3555,
|
| 1766 |
+
"loss_nan_ranks": 0,
|
| 1767 |
+
"loss_rank_avg": 0.17521412670612335,
|
| 1768 |
+
"step": 800,
|
| 1769 |
+
"valid_targets_mean": 4207.9,
|
| 1770 |
+
"valid_targets_min": 920
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.5728,
|
| 1774 |
+
"grad_norm": 0.3767564140590417,
|
| 1775 |
+
"learning_rate": 2.2536767424730052e-05,
|
| 1776 |
+
"loss": 0.3813,
|
| 1777 |
+
"loss_nan_ranks": 0,
|
| 1778 |
+
"loss_rank_avg": 0.18163838982582092,
|
| 1779 |
+
"step": 805,
|
| 1780 |
+
"valid_targets_mean": 4502.3,
|
| 1781 |
+
"valid_targets_min": 676
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.5888,
|
| 1785 |
+
"grad_norm": 0.33598923045311496,
|
| 1786 |
+
"learning_rate": 2.2315291761985803e-05,
|
| 1787 |
+
"loss": 0.3588,
|
| 1788 |
+
"loss_nan_ranks": 0,
|
| 1789 |
+
"loss_rank_avg": 0.19112834334373474,
|
| 1790 |
+
"step": 810,
|
| 1791 |
+
"valid_targets_mean": 4212.1,
|
| 1792 |
+
"valid_targets_min": 701
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 2.6048,
|
| 1796 |
+
"grad_norm": 0.3321302659711612,
|
| 1797 |
+
"learning_rate": 2.2093527937933716e-05,
|
| 1798 |
+
"loss": 0.3554,
|
| 1799 |
+
"loss_nan_ranks": 0,
|
| 1800 |
+
"loss_rank_avg": 0.18555206060409546,
|
| 1801 |
+
"step": 815,
|
| 1802 |
+
"valid_targets_mean": 5138.4,
|
| 1803 |
+
"valid_targets_min": 903
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 2.6208,
|
| 1807 |
+
"grad_norm": 0.4332190176537359,
|
| 1808 |
+
"learning_rate": 2.1871503553308447e-05,
|
| 1809 |
+
"loss": 0.3716,
|
| 1810 |
+
"loss_nan_ranks": 0,
|
| 1811 |
+
"loss_rank_avg": 0.24525032937526703,
|
| 1812 |
+
"step": 820,
|
| 1813 |
+
"valid_targets_mean": 3339.2,
|
| 1814 |
+
"valid_targets_min": 530
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 2.6368,
|
| 1818 |
+
"grad_norm": 0.3356538304748196,
|
| 1819 |
+
"learning_rate": 2.164924624127403e-05,
|
| 1820 |
+
"loss": 0.3841,
|
| 1821 |
+
"loss_nan_ranks": 0,
|
| 1822 |
+
"loss_rank_avg": 0.1583971232175827,
|
| 1823 |
+
"step": 825,
|
| 1824 |
+
"valid_targets_mean": 4937.4,
|
| 1825 |
+
"valid_targets_min": 780
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 2.6528,
|
| 1829 |
+
"grad_norm": 0.43608081245236874,
|
| 1830 |
+
"learning_rate": 2.1426783663984648e-05,
|
| 1831 |
+
"loss": 0.3584,
|
| 1832 |
+
"loss_nan_ranks": 0,
|
| 1833 |
+
"loss_rank_avg": 0.20730876922607422,
|
| 1834 |
+
"step": 830,
|
| 1835 |
+
"valid_targets_mean": 3397.2,
|
| 1836 |
+
"valid_targets_min": 572
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.6688,
|
| 1840 |
+
"grad_norm": 0.44955532042396595,
|
| 1841 |
+
"learning_rate": 2.1204143509141818e-05,
|
| 1842 |
+
"loss": 0.3553,
|
| 1843 |
+
"loss_nan_ranks": 0,
|
| 1844 |
+
"loss_rank_avg": 0.15707382559776306,
|
| 1845 |
+
"step": 835,
|
| 1846 |
+
"valid_targets_mean": 4469.5,
|
| 1847 |
+
"valid_targets_min": 716
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 2.6848,
|
| 1851 |
+
"grad_norm": 0.5050914114602801,
|
| 1852 |
+
"learning_rate": 2.0981353486548363e-05,
|
| 1853 |
+
"loss": 0.3745,
|
| 1854 |
+
"loss_nan_ranks": 0,
|
| 1855 |
+
"loss_rank_avg": 0.20182707905769348,
|
| 1856 |
+
"step": 840,
|
| 1857 |
+
"valid_targets_mean": 3843.1,
|
| 1858 |
+
"valid_targets_min": 1064
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 2.7008,
|
| 1862 |
+
"grad_norm": 0.39149794144314437,
|
| 1863 |
+
"learning_rate": 2.075844132465964e-05,
|
| 1864 |
+
"loss": 0.3687,
|
| 1865 |
+
"loss_nan_ranks": 0,
|
| 1866 |
+
"loss_rank_avg": 0.1607154756784439,
|
| 1867 |
+
"step": 845,
|
| 1868 |
+
"valid_targets_mean": 3015.9,
|
| 1869 |
+
"valid_targets_min": 787
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.7168,
|
| 1873 |
+
"grad_norm": 0.38997342582901406,
|
| 1874 |
+
"learning_rate": 2.0535434767132495e-05,
|
| 1875 |
+
"loss": 0.361,
|
| 1876 |
+
"loss_nan_ranks": 0,
|
| 1877 |
+
"loss_rank_avg": 0.2090197205543518,
|
| 1878 |
+
"step": 850,
|
| 1879 |
+
"valid_targets_mean": 3698.9,
|
| 1880 |
+
"valid_targets_min": 523
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 2.7328,
|
| 1884 |
+
"grad_norm": 0.3588635051303132,
|
| 1885 |
+
"learning_rate": 2.0312361569372215e-05,
|
| 1886 |
+
"loss": 0.372,
|
| 1887 |
+
"loss_nan_ranks": 0,
|
| 1888 |
+
"loss_rank_avg": 0.20288428664207458,
|
| 1889 |
+
"step": 855,
|
| 1890 |
+
"valid_targets_mean": 4794.2,
|
| 1891 |
+
"valid_targets_min": 624
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 2.7488,
|
| 1895 |
+
"grad_norm": 0.36454271223668117,
|
| 1896 |
+
"learning_rate": 2.0089249495078186e-05,
|
| 1897 |
+
"loss": 0.3525,
|
| 1898 |
+
"loss_nan_ranks": 0,
|
| 1899 |
+
"loss_rank_avg": 0.14510402083396912,
|
| 1900 |
+
"step": 860,
|
| 1901 |
+
"valid_targets_mean": 3878.5,
|
| 1902 |
+
"valid_targets_min": 647
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 2.7648,
|
| 1906 |
+
"grad_norm": 0.3980778917765405,
|
| 1907 |
+
"learning_rate": 1.9866126312788333e-05,
|
| 1908 |
+
"loss": 0.3439,
|
| 1909 |
+
"loss_nan_ranks": 0,
|
| 1910 |
+
"loss_rank_avg": 0.17586301267147064,
|
| 1911 |
+
"step": 865,
|
| 1912 |
+
"valid_targets_mean": 3193.8,
|
| 1913 |
+
"valid_targets_min": 734
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 2.7808,
|
| 1917 |
+
"grad_norm": 0.38107961900162757,
|
| 1918 |
+
"learning_rate": 1.964301979242308e-05,
|
| 1919 |
+
"loss": 0.3571,
|
| 1920 |
+
"loss_nan_ranks": 0,
|
| 1921 |
+
"loss_rank_avg": 0.1435280442237854,
|
| 1922 |
+
"step": 870,
|
| 1923 |
+
"valid_targets_mean": 3100.1,
|
| 1924 |
+
"valid_targets_min": 757
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 2.7968,
|
| 1928 |
+
"grad_norm": 0.3421598363206481,
|
| 1929 |
+
"learning_rate": 1.9419957701829138e-05,
|
| 1930 |
+
"loss": 0.3594,
|
| 1931 |
+
"loss_nan_ranks": 0,
|
| 1932 |
+
"loss_rank_avg": 0.18592098355293274,
|
| 1933 |
+
"step": 875,
|
| 1934 |
+
"valid_targets_mean": 5316.6,
|
| 1935 |
+
"valid_targets_min": 665
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 2.8128,
|
| 1939 |
+
"grad_norm": 0.3625020704663103,
|
| 1940 |
+
"learning_rate": 1.9196967803323464e-05,
|
| 1941 |
+
"loss": 0.36,
|
| 1942 |
+
"loss_nan_ranks": 0,
|
| 1943 |
+
"loss_rank_avg": 0.1660204827785492,
|
| 1944 |
+
"step": 880,
|
| 1945 |
+
"valid_targets_mean": 3241.3,
|
| 1946 |
+
"valid_targets_min": 739
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 2.8288,
|
| 1950 |
+
"grad_norm": 0.3844458210877275,
|
| 1951 |
+
"learning_rate": 1.8974077850237983e-05,
|
| 1952 |
+
"loss": 0.3501,
|
| 1953 |
+
"loss_nan_ranks": 0,
|
| 1954 |
+
"loss_rank_avg": 0.2059369534254074,
|
| 1955 |
+
"step": 885,
|
| 1956 |
+
"valid_targets_mean": 4584.8,
|
| 1957 |
+
"valid_targets_min": 816
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"epoch": 2.8448,
|
| 1961 |
+
"grad_norm": 0.3539100216907783,
|
| 1962 |
+
"learning_rate": 1.875131558346542e-05,
|
| 1963 |
+
"loss": 0.3435,
|
| 1964 |
+
"loss_nan_ranks": 0,
|
| 1965 |
+
"loss_rank_avg": 0.13612182438373566,
|
| 1966 |
+
"step": 890,
|
| 1967 |
+
"valid_targets_mean": 2774.1,
|
| 1968 |
+
"valid_targets_min": 729
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 2.8608000000000002,
|
| 1972 |
+
"grad_norm": 0.40583212495073406,
|
| 1973 |
+
"learning_rate": 1.8528708728006654e-05,
|
| 1974 |
+
"loss": 0.3559,
|
| 1975 |
+
"loss_nan_ranks": 0,
|
| 1976 |
+
"loss_rank_avg": 0.1654030978679657,
|
| 1977 |
+
"step": 895,
|
| 1978 |
+
"valid_targets_mean": 3030.8,
|
| 1979 |
+
"valid_targets_min": 803
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 2.8768000000000002,
|
| 1983 |
+
"grad_norm": 0.45124674084690286,
|
| 1984 |
+
"learning_rate": 1.8306284989520055e-05,
|
| 1985 |
+
"loss": 0.3575,
|
| 1986 |
+
"loss_nan_ranks": 0,
|
| 1987 |
+
"loss_rank_avg": 0.22975601255893707,
|
| 1988 |
+
"step": 900,
|
| 1989 |
+
"valid_targets_mean": 5792.0,
|
| 1990 |
+
"valid_targets_min": 713
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 2.8928000000000003,
|
| 1994 |
+
"grad_norm": 0.374865102674942,
|
| 1995 |
+
"learning_rate": 1.8084072050873265e-05,
|
| 1996 |
+
"loss": 0.3526,
|
| 1997 |
+
"loss_nan_ranks": 0,
|
| 1998 |
+
"loss_rank_avg": 0.1920439600944519,
|
| 1999 |
+
"step": 905,
|
| 2000 |
+
"valid_targets_mean": 4009.6,
|
| 2001 |
+
"valid_targets_min": 404
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 2.9088000000000003,
|
| 2005 |
+
"grad_norm": 0.33227134778852785,
|
| 2006 |
+
"learning_rate": 1.786209756869775e-05,
|
| 2007 |
+
"loss": 0.3495,
|
| 2008 |
+
"loss_nan_ranks": 0,
|
| 2009 |
+
"loss_rank_avg": 0.20190231502056122,
|
| 2010 |
+
"step": 910,
|
| 2011 |
+
"valid_targets_mean": 6347.8,
|
| 2012 |
+
"valid_targets_min": 1377
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 2.9248,
|
| 2016 |
+
"grad_norm": 0.35861025661918494,
|
| 2017 |
+
"learning_rate": 1.764038916994669e-05,
|
| 2018 |
+
"loss": 0.3448,
|
| 2019 |
+
"loss_nan_ranks": 0,
|
| 2020 |
+
"loss_rank_avg": 0.1896182894706726,
|
| 2021 |
+
"step": 915,
|
| 2022 |
+
"valid_targets_mean": 4381.7,
|
| 2023 |
+
"valid_targets_min": 859
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 2.9408,
|
| 2027 |
+
"grad_norm": 0.35561939433082074,
|
| 2028 |
+
"learning_rate": 1.741897444845649e-05,
|
| 2029 |
+
"loss": 0.3404,
|
| 2030 |
+
"loss_nan_ranks": 0,
|
| 2031 |
+
"loss_rank_avg": 0.22202759981155396,
|
| 2032 |
+
"step": 920,
|
| 2033 |
+
"valid_targets_mean": 4680.8,
|
| 2034 |
+
"valid_targets_min": 1115
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.9568,
|
| 2038 |
+
"grad_norm": 0.34917086301788913,
|
| 2039 |
+
"learning_rate": 1.7197880961512498e-05,
|
| 2040 |
+
"loss": 0.3537,
|
| 2041 |
+
"loss_nan_ranks": 0,
|
| 2042 |
+
"loss_rank_avg": 0.1740960329771042,
|
| 2043 |
+
"step": 925,
|
| 2044 |
+
"valid_targets_mean": 4159.9,
|
| 2045 |
+
"valid_targets_min": 916
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 2.9728,
|
| 2049 |
+
"grad_norm": 0.36394569930759646,
|
| 2050 |
+
"learning_rate": 1.6977136226419187e-05,
|
| 2051 |
+
"loss": 0.3671,
|
| 2052 |
+
"loss_nan_ranks": 0,
|
| 2053 |
+
"loss_rank_avg": 0.19525155425071716,
|
| 2054 |
+
"step": 930,
|
| 2055 |
+
"valid_targets_mean": 4545.5,
|
| 2056 |
+
"valid_targets_min": 1010
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 2.9888,
|
| 2060 |
+
"grad_norm": 0.38020671979315085,
|
| 2061 |
+
"learning_rate": 1.6756767717075354e-05,
|
| 2062 |
+
"loss": 0.3801,
|
| 2063 |
+
"loss_nan_ranks": 0,
|
| 2064 |
+
"loss_rank_avg": 0.20520594716072083,
|
| 2065 |
+
"step": 935,
|
| 2066 |
+
"valid_targets_mean": 3499.2,
|
| 2067 |
+
"valid_targets_min": 1092
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 3.0032,
|
| 2071 |
+
"grad_norm": 0.37605046406373455,
|
| 2072 |
+
"learning_rate": 1.6536802860554723e-05,
|
| 2073 |
+
"loss": 0.3727,
|
| 2074 |
+
"loss_nan_ranks": 0,
|
| 2075 |
+
"loss_rank_avg": 0.1897515505552292,
|
| 2076 |
+
"step": 940,
|
| 2077 |
+
"valid_targets_mean": 3898.6,
|
| 2078 |
+
"valid_targets_min": 794
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 3.0192,
|
| 2082 |
+
"grad_norm": 0.335356176522162,
|
| 2083 |
+
"learning_rate": 1.631726903369238e-05,
|
| 2084 |
+
"loss": 0.3438,
|
| 2085 |
+
"loss_nan_ranks": 0,
|
| 2086 |
+
"loss_rank_avg": 0.20420154929161072,
|
| 2087 |
+
"step": 945,
|
| 2088 |
+
"valid_targets_mean": 4534.9,
|
| 2089 |
+
"valid_targets_min": 1332
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 3.0352,
|
| 2093 |
+
"grad_norm": 0.4094788051075967,
|
| 2094 |
+
"learning_rate": 1.609819355967744e-05,
|
| 2095 |
+
"loss": 0.3595,
|
| 2096 |
+
"loss_nan_ranks": 0,
|
| 2097 |
+
"loss_rank_avg": 0.21997550129890442,
|
| 2098 |
+
"step": 950,
|
| 2099 |
+
"valid_targets_mean": 5079.9,
|
| 2100 |
+
"valid_targets_min": 899
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 3.0512,
|
| 2104 |
+
"grad_norm": 0.3776128013373243,
|
| 2105 |
+
"learning_rate": 1.587960370465239e-05,
|
| 2106 |
+
"loss": 0.3418,
|
| 2107 |
+
"loss_nan_ranks": 0,
|
| 2108 |
+
"loss_rank_avg": 0.1931406706571579,
|
| 2109 |
+
"step": 955,
|
| 2110 |
+
"valid_targets_mean": 4443.5,
|
| 2111 |
+
"valid_targets_min": 942
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 3.0672,
|
| 2115 |
+
"grad_norm": 0.3360958142209563,
|
| 2116 |
+
"learning_rate": 1.5661526674319582e-05,
|
| 2117 |
+
"loss": 0.3244,
|
| 2118 |
+
"loss_nan_ranks": 0,
|
| 2119 |
+
"loss_rank_avg": 0.13528843224048615,
|
| 2120 |
+
"step": 960,
|
| 2121 |
+
"valid_targets_mean": 4246.4,
|
| 2122 |
+
"valid_targets_min": 829
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 3.0832,
|
| 2126 |
+
"grad_norm": 0.4239322119841672,
|
| 2127 |
+
"learning_rate": 1.544398961055516e-05,
|
| 2128 |
+
"loss": 0.3404,
|
| 2129 |
+
"loss_nan_ranks": 0,
|
| 2130 |
+
"loss_rank_avg": 0.1699754148721695,
|
| 2131 |
+
"step": 965,
|
| 2132 |
+
"valid_targets_mean": 2840.8,
|
| 2133 |
+
"valid_targets_min": 784
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 3.0992,
|
| 2137 |
+
"grad_norm": 0.3993307352764048,
|
| 2138 |
+
"learning_rate": 1.5227019588031035e-05,
|
| 2139 |
+
"loss": 0.3537,
|
| 2140 |
+
"loss_nan_ranks": 0,
|
| 2141 |
+
"loss_rank_avg": 0.14093229174613953,
|
| 2142 |
+
"step": 970,
|
| 2143 |
+
"valid_targets_mean": 2824.7,
|
| 2144 |
+
"valid_targets_min": 512
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 3.1152,
|
| 2148 |
+
"grad_norm": 0.44328943732647597,
|
| 2149 |
+
"learning_rate": 1.501064361084511e-05,
|
| 2150 |
+
"loss": 0.3441,
|
| 2151 |
+
"loss_nan_ranks": 0,
|
| 2152 |
+
"loss_rank_avg": 0.20833003520965576,
|
| 2153 |
+
"step": 975,
|
| 2154 |
+
"valid_targets_mean": 3540.4,
|
| 2155 |
+
"valid_targets_min": 676
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 3.1312,
|
| 2159 |
+
"grad_norm": 0.3454439076872715,
|
| 2160 |
+
"learning_rate": 1.47948886091604e-05,
|
| 2161 |
+
"loss": 0.3388,
|
| 2162 |
+
"loss_nan_ranks": 0,
|
| 2163 |
+
"loss_rank_avg": 0.13593928515911102,
|
| 2164 |
+
"step": 980,
|
| 2165 |
+
"valid_targets_mean": 4455.2,
|
| 2166 |
+
"valid_targets_min": 1037
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 3.1471999999999998,
|
| 2170 |
+
"grad_norm": 0.38888253914726173,
|
| 2171 |
+
"learning_rate": 1.4579781435853289e-05,
|
| 2172 |
+
"loss": 0.3407,
|
| 2173 |
+
"loss_nan_ranks": 0,
|
| 2174 |
+
"loss_rank_avg": 0.2271447479724884,
|
| 2175 |
+
"step": 985,
|
| 2176 |
+
"valid_targets_mean": 4185.3,
|
| 2177 |
+
"valid_targets_min": 865
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 3.1632,
|
| 2181 |
+
"grad_norm": 0.34563917020171253,
|
| 2182 |
+
"learning_rate": 1.4365348863171406e-05,
|
| 2183 |
+
"loss": 0.3485,
|
| 2184 |
+
"loss_nan_ranks": 0,
|
| 2185 |
+
"loss_rank_avg": 0.09505438804626465,
|
| 2186 |
+
"step": 990,
|
| 2187 |
+
"valid_targets_mean": 3164.1,
|
| 2188 |
+
"valid_targets_min": 691
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 3.1792,
|
| 2192 |
+
"grad_norm": 0.31352592534427964,
|
| 2193 |
+
"learning_rate": 1.4151617579401551e-05,
|
| 2194 |
+
"loss": 0.3483,
|
| 2195 |
+
"loss_nan_ranks": 0,
|
| 2196 |
+
"loss_rank_avg": 0.14270518720149994,
|
| 2197 |
+
"step": 995,
|
| 2198 |
+
"valid_targets_mean": 4691.3,
|
| 2199 |
+
"valid_targets_min": 672
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 3.1952,
|
| 2203 |
+
"grad_norm": 0.43447105744832676,
|
| 2204 |
+
"learning_rate": 1.3938614185548094e-05,
|
| 2205 |
+
"loss": 0.3432,
|
| 2206 |
+
"loss_nan_ranks": 0,
|
| 2207 |
+
"loss_rank_avg": 0.13177445530891418,
|
| 2208 |
+
"step": 1000,
|
| 2209 |
+
"valid_targets_mean": 2326.1,
|
| 2210 |
+
"valid_targets_min": 672
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 3.2112,
|
| 2214 |
+
"grad_norm": 0.3563119894321619,
|
| 2215 |
+
"learning_rate": 1.3726365192022173e-05,
|
| 2216 |
+
"loss": 0.3404,
|
| 2217 |
+
"loss_nan_ranks": 0,
|
| 2218 |
+
"loss_rank_avg": 0.1878964751958847,
|
| 2219 |
+
"step": 1005,
|
| 2220 |
+
"valid_targets_mean": 4940.8,
|
| 2221 |
+
"valid_targets_min": 762
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 3.2272,
|
| 2225 |
+
"grad_norm": 0.35363752093967743,
|
| 2226 |
+
"learning_rate": 1.3514897015342257e-05,
|
| 2227 |
+
"loss": 0.3253,
|
| 2228 |
+
"loss_nan_ranks": 0,
|
| 2229 |
+
"loss_rank_avg": 0.13989496231079102,
|
| 2230 |
+
"step": 1010,
|
| 2231 |
+
"valid_targets_mean": 4354.2,
|
| 2232 |
+
"valid_targets_min": 523
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 3.2432,
|
| 2236 |
+
"grad_norm": 0.4319224916079313,
|
| 2237 |
+
"learning_rate": 1.3304235974846295e-05,
|
| 2238 |
+
"loss": 0.3299,
|
| 2239 |
+
"loss_nan_ranks": 0,
|
| 2240 |
+
"loss_rank_avg": 0.1998392939567566,
|
| 2241 |
+
"step": 1015,
|
| 2242 |
+
"valid_targets_mean": 3906.0,
|
| 2243 |
+
"valid_targets_min": 679
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 3.2592,
|
| 2247 |
+
"grad_norm": 0.3495855473672183,
|
| 2248 |
+
"learning_rate": 1.3094408289416052e-05,
|
| 2249 |
+
"loss": 0.3475,
|
| 2250 |
+
"loss_nan_ranks": 0,
|
| 2251 |
+
"loss_rank_avg": 0.20489974319934845,
|
| 2252 |
+
"step": 1020,
|
| 2253 |
+
"valid_targets_mean": 5238.0,
|
| 2254 |
+
"valid_targets_min": 668
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 3.2752,
|
| 2258 |
+
"grad_norm": 0.4060553154313819,
|
| 2259 |
+
"learning_rate": 1.2885440074213877e-05,
|
| 2260 |
+
"loss": 0.3325,
|
| 2261 |
+
"loss_nan_ranks": 0,
|
| 2262 |
+
"loss_rank_avg": 0.22139576077461243,
|
| 2263 |
+
"step": 1025,
|
| 2264 |
+
"valid_targets_mean": 4605.4,
|
| 2265 |
+
"valid_targets_min": 796
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 3.2912,
|
| 2269 |
+
"grad_norm": 0.38852642508297336,
|
| 2270 |
+
"learning_rate": 1.267735733743242e-05,
|
| 2271 |
+
"loss": 0.3319,
|
| 2272 |
+
"loss_nan_ranks": 0,
|
| 2273 |
+
"loss_rank_avg": 0.19194181263446808,
|
| 2274 |
+
"step": 1030,
|
| 2275 |
+
"valid_targets_mean": 4457.9,
|
| 2276 |
+
"valid_targets_min": 1327
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 3.3072,
|
| 2280 |
+
"grad_norm": 0.36166900656776446,
|
| 2281 |
+
"learning_rate": 1.2470185977057643e-05,
|
| 2282 |
+
"loss": 0.3286,
|
| 2283 |
+
"loss_nan_ranks": 0,
|
| 2284 |
+
"loss_rank_avg": 0.1793379783630371,
|
| 2285 |
+
"step": 1035,
|
| 2286 |
+
"valid_targets_mean": 4544.1,
|
| 2287 |
+
"valid_targets_min": 1022
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 3.3232,
|
| 2291 |
+
"grad_norm": 0.36293178811713217,
|
| 2292 |
+
"learning_rate": 1.2263951777645588e-05,
|
| 2293 |
+
"loss": 0.3282,
|
| 2294 |
+
"loss_nan_ranks": 0,
|
| 2295 |
+
"loss_rank_avg": 0.10857829451560974,
|
| 2296 |
+
"step": 1040,
|
| 2297 |
+
"valid_targets_mean": 1993.7,
|
| 2298 |
+
"valid_targets_min": 911
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.3392,
|
| 2302 |
+
"grad_norm": 0.5084148493340929,
|
| 2303 |
+
"learning_rate": 1.2058680407113176e-05,
|
| 2304 |
+
"loss": 0.3532,
|
| 2305 |
+
"loss_nan_ranks": 0,
|
| 2306 |
+
"loss_rank_avg": 0.1812812238931656,
|
| 2307 |
+
"step": 1045,
|
| 2308 |
+
"valid_targets_mean": 2746.2,
|
| 2309 |
+
"valid_targets_min": 667
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 3.3552,
|
| 2313 |
+
"grad_norm": 0.3430189590754264,
|
| 2314 |
+
"learning_rate": 1.1854397413543626e-05,
|
| 2315 |
+
"loss": 0.3523,
|
| 2316 |
+
"loss_nan_ranks": 0,
|
| 2317 |
+
"loss_rank_avg": 0.13636204600334167,
|
| 2318 |
+
"step": 1050,
|
| 2319 |
+
"valid_targets_mean": 4036.2,
|
| 2320 |
+
"valid_targets_min": 811
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 3.3712,
|
| 2324 |
+
"grad_norm": 0.331734814176711,
|
| 2325 |
+
"learning_rate": 1.1651128222006713e-05,
|
| 2326 |
+
"loss": 0.3327,
|
| 2327 |
+
"loss_nan_ranks": 0,
|
| 2328 |
+
"loss_rank_avg": 0.15516330301761627,
|
| 2329 |
+
"step": 1055,
|
| 2330 |
+
"valid_targets_mean": 4857.7,
|
| 2331 |
+
"valid_targets_min": 729
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 3.3872,
|
| 2335 |
+
"grad_norm": 0.3484206028478157,
|
| 2336 |
+
"learning_rate": 1.1448898131394364e-05,
|
| 2337 |
+
"loss": 0.3261,
|
| 2338 |
+
"loss_nan_ranks": 0,
|
| 2339 |
+
"loss_rank_avg": 0.153816819190979,
|
| 2340 |
+
"step": 1060,
|
| 2341 |
+
"valid_targets_mean": 4114.8,
|
| 2342 |
+
"valid_targets_min": 569
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 3.4032,
|
| 2346 |
+
"grad_norm": 0.3944658789774959,
|
| 2347 |
+
"learning_rate": 1.124773231127196e-05,
|
| 2348 |
+
"loss": 0.36,
|
| 2349 |
+
"loss_nan_ranks": 0,
|
| 2350 |
+
"loss_rank_avg": 0.16333676874637604,
|
| 2351 |
+
"step": 1065,
|
| 2352 |
+
"valid_targets_mean": 4517.0,
|
| 2353 |
+
"valid_targets_min": 933
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 3.4192,
|
| 2357 |
+
"grad_norm": 0.44319906945749254,
|
| 2358 |
+
"learning_rate": 1.1047655798745752e-05,
|
| 2359 |
+
"loss": 0.3546,
|
| 2360 |
+
"loss_nan_ranks": 0,
|
| 2361 |
+
"loss_rank_avg": 0.1139206737279892,
|
| 2362 |
+
"step": 1070,
|
| 2363 |
+
"valid_targets_mean": 2351.4,
|
| 2364 |
+
"valid_targets_min": 654
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 3.4352,
|
| 2368 |
+
"grad_norm": 0.3562865395677016,
|
| 2369 |
+
"learning_rate": 1.084869349534671e-05,
|
| 2370 |
+
"loss": 0.3428,
|
| 2371 |
+
"loss_nan_ranks": 0,
|
| 2372 |
+
"loss_rank_avg": 0.19184201955795288,
|
| 2373 |
+
"step": 1075,
|
| 2374 |
+
"valid_targets_mean": 4532.2,
|
| 2375 |
+
"valid_targets_min": 722
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 3.4512,
|
| 2379 |
+
"grad_norm": 0.3779331561005202,
|
| 2380 |
+
"learning_rate": 1.0650870163931275e-05,
|
| 2381 |
+
"loss": 0.358,
|
| 2382 |
+
"loss_nan_ranks": 0,
|
| 2383 |
+
"loss_rank_avg": 0.15721634030342102,
|
| 2384 |
+
"step": 1080,
|
| 2385 |
+
"valid_targets_mean": 3805.3,
|
| 2386 |
+
"valid_targets_min": 964
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 3.4672,
|
| 2390 |
+
"grad_norm": 0.3606978223574716,
|
| 2391 |
+
"learning_rate": 1.0454210425599426e-05,
|
| 2392 |
+
"loss": 0.3508,
|
| 2393 |
+
"loss_nan_ranks": 0,
|
| 2394 |
+
"loss_rank_avg": 0.16364584863185883,
|
| 2395 |
+
"step": 1085,
|
| 2396 |
+
"valid_targets_mean": 3938.2,
|
| 2397 |
+
"valid_targets_min": 697
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 3.4832,
|
| 2401 |
+
"grad_norm": 0.37686671437971875,
|
| 2402 |
+
"learning_rate": 1.0258738756630255e-05,
|
| 2403 |
+
"loss": 0.3474,
|
| 2404 |
+
"loss_nan_ranks": 0,
|
| 2405 |
+
"loss_rank_avg": 0.17473503947257996,
|
| 2406 |
+
"step": 1090,
|
| 2407 |
+
"valid_targets_mean": 3853.2,
|
| 2408 |
+
"valid_targets_min": 639
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 3.4992,
|
| 2412 |
+
"grad_norm": 0.4429048555200298,
|
| 2413 |
+
"learning_rate": 1.0064479485435737e-05,
|
| 2414 |
+
"loss": 0.3574,
|
| 2415 |
+
"loss_nan_ranks": 0,
|
| 2416 |
+
"loss_rank_avg": 0.15065409243106842,
|
| 2417 |
+
"step": 1095,
|
| 2418 |
+
"valid_targets_mean": 2704.8,
|
| 2419 |
+
"valid_targets_min": 762
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 3.5152,
|
| 2423 |
+
"grad_norm": 0.39157782240601363,
|
| 2424 |
+
"learning_rate": 9.871456789532736e-06,
|
| 2425 |
+
"loss": 0.3448,
|
| 2426 |
+
"loss_nan_ranks": 0,
|
| 2427 |
+
"loss_rank_avg": 0.170841783285141,
|
| 2428 |
+
"step": 1100,
|
| 2429 |
+
"valid_targets_mean": 3648.5,
|
| 2430 |
+
"valid_targets_min": 714
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 3.5312,
|
| 2434 |
+
"grad_norm": 0.4032833854947422,
|
| 2435 |
+
"learning_rate": 9.679694692533909e-06,
|
| 2436 |
+
"loss": 0.3401,
|
| 2437 |
+
"loss_nan_ranks": 0,
|
| 2438 |
+
"loss_rank_avg": 0.17664435505867004,
|
| 2439 |
+
"step": 1105,
|
| 2440 |
+
"valid_targets_mean": 3603.8,
|
| 2441 |
+
"valid_targets_min": 426
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 3.5472,
|
| 2445 |
+
"grad_norm": 0.3759820875959348,
|
| 2446 |
+
"learning_rate": 9.489217061157744e-06,
|
| 2447 |
+
"loss": 0.3525,
|
| 2448 |
+
"loss_nan_ranks": 0,
|
| 2449 |
+
"loss_rank_avg": 0.19742423295974731,
|
| 2450 |
+
"step": 1110,
|
| 2451 |
+
"valid_targets_mean": 3592.0,
|
| 2452 |
+
"valid_targets_min": 958
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 3.5632,
|
| 2456 |
+
"grad_norm": 0.44117131213661204,
|
| 2457 |
+
"learning_rate": 9.30004760225806e-06,
|
| 2458 |
+
"loss": 0.3412,
|
| 2459 |
+
"loss_nan_ranks": 0,
|
| 2460 |
+
"loss_rank_avg": 0.13590778410434723,
|
| 2461 |
+
"step": 1115,
|
| 2462 |
+
"valid_targets_mean": 2444.6,
|
| 2463 |
+
"valid_targets_min": 490
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 3.5792,
|
| 2467 |
+
"grad_norm": 0.35876633764128313,
|
| 2468 |
+
"learning_rate": 9.112209859873479e-06,
|
| 2469 |
+
"loss": 0.3414,
|
| 2470 |
+
"loss_nan_ranks": 0,
|
| 2471 |
+
"loss_rank_avg": 0.15319585800170898,
|
| 2472 |
+
"step": 1120,
|
| 2473 |
+
"valid_targets_mean": 3815.9,
|
| 2474 |
+
"valid_targets_min": 704
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 3.5952,
|
| 2478 |
+
"grad_norm": 0.34990276526484193,
|
| 2479 |
+
"learning_rate": 8.925727212297154e-06,
|
| 2480 |
+
"loss": 0.3407,
|
| 2481 |
+
"loss_nan_ranks": 0,
|
| 2482 |
+
"loss_rank_avg": 0.11832721531391144,
|
| 2483 |
+
"step": 1125,
|
| 2484 |
+
"valid_targets_mean": 3515.6,
|
| 2485 |
+
"valid_targets_min": 796
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 3.6112,
|
| 2489 |
+
"grad_norm": 0.4211378285934783,
|
| 2490 |
+
"learning_rate": 8.74062286916705e-06,
|
| 2491 |
+
"loss": 0.3524,
|
| 2492 |
+
"loss_nan_ranks": 0,
|
| 2493 |
+
"loss_rank_avg": 0.19017146527767181,
|
| 2494 |
+
"step": 1130,
|
| 2495 |
+
"valid_targets_mean": 3507.1,
|
| 2496 |
+
"valid_targets_min": 685
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 3.6272,
|
| 2500 |
+
"grad_norm": 0.522703439122451,
|
| 2501 |
+
"learning_rate": 8.55691986857733e-06,
|
| 2502 |
+
"loss": 0.3489,
|
| 2503 |
+
"loss_nan_ranks": 0,
|
| 2504 |
+
"loss_rank_avg": 0.18660058081150055,
|
| 2505 |
+
"step": 1135,
|
| 2506 |
+
"valid_targets_mean": 2299.6,
|
| 2507 |
+
"valid_targets_min": 660
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 3.6432,
|
| 2511 |
+
"grad_norm": 0.3408305954246411,
|
| 2512 |
+
"learning_rate": 8.374641074210979e-06,
|
| 2513 |
+
"loss": 0.3171,
|
| 2514 |
+
"loss_nan_ranks": 0,
|
| 2515 |
+
"loss_rank_avg": 0.18012124300003052,
|
| 2516 |
+
"step": 1140,
|
| 2517 |
+
"valid_targets_mean": 5196.1,
|
| 2518 |
+
"valid_targets_min": 713
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 3.6592000000000002,
|
| 2522 |
+
"grad_norm": 0.3788128493752177,
|
| 2523 |
+
"learning_rate": 8.193809172494249e-06,
|
| 2524 |
+
"loss": 0.3547,
|
| 2525 |
+
"loss_nan_ranks": 0,
|
| 2526 |
+
"loss_rank_avg": 0.1889040768146515,
|
| 2527 |
+
"step": 1145,
|
| 2528 |
+
"valid_targets_mean": 3868.0,
|
| 2529 |
+
"valid_targets_min": 563
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 3.6752000000000002,
|
| 2533 |
+
"grad_norm": 0.3195327070645042,
|
| 2534 |
+
"learning_rate": 8.014446669773061e-06,
|
| 2535 |
+
"loss": 0.3508,
|
| 2536 |
+
"loss_nan_ranks": 0,
|
| 2537 |
+
"loss_rank_avg": 0.17038068175315857,
|
| 2538 |
+
"step": 1150,
|
| 2539 |
+
"valid_targets_mean": 5676.6,
|
| 2540 |
+
"valid_targets_min": 1112
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 3.6912000000000003,
|
| 2544 |
+
"grad_norm": 0.3292140039943816,
|
| 2545 |
+
"learning_rate": 7.83657588951187e-06,
|
| 2546 |
+
"loss": 0.3547,
|
| 2547 |
+
"loss_nan_ranks": 0,
|
| 2548 |
+
"loss_rank_avg": 0.18771226704120636,
|
| 2549 |
+
"step": 1155,
|
| 2550 |
+
"valid_targets_mean": 6315.8,
|
| 2551 |
+
"valid_targets_min": 702
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 3.7072000000000003,
|
| 2555 |
+
"grad_norm": 0.36796060727007385,
|
| 2556 |
+
"learning_rate": 7.66021896951529e-06,
|
| 2557 |
+
"loss": 0.314,
|
| 2558 |
+
"loss_nan_ranks": 0,
|
| 2559 |
+
"loss_rank_avg": 0.13613468408584595,
|
| 2560 |
+
"step": 1160,
|
| 2561 |
+
"valid_targets_mean": 4006.8,
|
| 2562 |
+
"valid_targets_min": 777
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 3.7232,
|
| 2566 |
+
"grad_norm": 0.4094060697982775,
|
| 2567 |
+
"learning_rate": 7.485397859172841e-06,
|
| 2568 |
+
"loss": 0.3557,
|
| 2569 |
+
"loss_nan_ranks": 0,
|
| 2570 |
+
"loss_rank_avg": 0.16176709532737732,
|
| 2571 |
+
"step": 1165,
|
| 2572 |
+
"valid_targets_mean": 3108.9,
|
| 2573 |
+
"valid_targets_min": 922
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 3.7392,
|
| 2577 |
+
"grad_norm": 0.34905958959203853,
|
| 2578 |
+
"learning_rate": 7.312134316727093e-06,
|
| 2579 |
+
"loss": 0.3372,
|
| 2580 |
+
"loss_nan_ranks": 0,
|
| 2581 |
+
"loss_rank_avg": 0.16702544689178467,
|
| 2582 |
+
"step": 1170,
|
| 2583 |
+
"valid_targets_mean": 4760.9,
|
| 2584 |
+
"valid_targets_min": 1183
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 3.7552,
|
| 2588 |
+
"grad_norm": 0.35661452032445096,
|
| 2589 |
+
"learning_rate": 7.140449906565656e-06,
|
| 2590 |
+
"loss": 0.3375,
|
| 2591 |
+
"loss_nan_ranks": 0,
|
| 2592 |
+
"loss_rank_avg": 0.15472249686717987,
|
| 2593 |
+
"step": 1175,
|
| 2594 |
+
"valid_targets_mean": 3650.1,
|
| 2595 |
+
"valid_targets_min": 656
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 3.7712,
|
| 2599 |
+
"grad_norm": 0.37358322798596966,
|
| 2600 |
+
"learning_rate": 6.970365996537285e-06,
|
| 2601 |
+
"loss": 0.3413,
|
| 2602 |
+
"loss_nan_ranks": 0,
|
| 2603 |
+
"loss_rank_avg": 0.16219046711921692,
|
| 2604 |
+
"step": 1180,
|
| 2605 |
+
"valid_targets_mean": 4001.8,
|
| 2606 |
+
"valid_targets_min": 862
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 3.7872,
|
| 2610 |
+
"grad_norm": 0.3749806820714067,
|
| 2611 |
+
"learning_rate": 6.801903755292403e-06,
|
| 2612 |
+
"loss": 0.3533,
|
| 2613 |
+
"loss_nan_ranks": 0,
|
| 2614 |
+
"loss_rank_avg": 0.1868247240781784,
|
| 2615 |
+
"step": 1185,
|
| 2616 |
+
"valid_targets_mean": 4775.1,
|
| 2617 |
+
"valid_targets_min": 578
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 3.8032,
|
| 2621 |
+
"grad_norm": 0.38429670420945505,
|
| 2622 |
+
"learning_rate": 6.635084149648481e-06,
|
| 2623 |
+
"loss": 0.3661,
|
| 2624 |
+
"loss_nan_ranks": 0,
|
| 2625 |
+
"loss_rank_avg": 0.12616121768951416,
|
| 2626 |
+
"step": 1190,
|
| 2627 |
+
"valid_targets_mean": 2768.1,
|
| 2628 |
+
"valid_targets_min": 563
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 3.8192,
|
| 2632 |
+
"grad_norm": 0.36030429317287976,
|
| 2633 |
+
"learning_rate": 6.469927941980483e-06,
|
| 2634 |
+
"loss": 0.346,
|
| 2635 |
+
"loss_nan_ranks": 0,
|
| 2636 |
+
"loss_rank_avg": 0.186874657869339,
|
| 2637 |
+
"step": 1195,
|
| 2638 |
+
"valid_targets_mean": 4400.2,
|
| 2639 |
+
"valid_targets_min": 843
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 3.8352,
|
| 2643 |
+
"grad_norm": 0.35202133068417446,
|
| 2644 |
+
"learning_rate": 6.30645568763681e-06,
|
| 2645 |
+
"loss": 0.339,
|
| 2646 |
+
"loss_nan_ranks": 0,
|
| 2647 |
+
"loss_rank_avg": 0.18812918663024902,
|
| 2648 |
+
"step": 1200,
|
| 2649 |
+
"valid_targets_mean": 4493.2,
|
| 2650 |
+
"valid_targets_min": 730
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 3.8512,
|
| 2654 |
+
"grad_norm": 0.32683827554231104,
|
| 2655 |
+
"learning_rate": 6.144687732380963e-06,
|
| 2656 |
+
"loss": 0.3287,
|
| 2657 |
+
"loss_nan_ranks": 0,
|
| 2658 |
+
"loss_rank_avg": 0.14614273607730865,
|
| 2659 |
+
"step": 1205,
|
| 2660 |
+
"valid_targets_mean": 4499.8,
|
| 2661 |
+
"valid_targets_min": 820
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 3.8672,
|
| 2665 |
+
"grad_norm": 0.41211985482110514,
|
| 2666 |
+
"learning_rate": 5.9846442098592895e-06,
|
| 2667 |
+
"loss": 0.3321,
|
| 2668 |
+
"loss_nan_ranks": 0,
|
| 2669 |
+
"loss_rank_avg": 0.1370638906955719,
|
| 2670 |
+
"step": 1210,
|
| 2671 |
+
"valid_targets_mean": 2420.4,
|
| 2672 |
+
"valid_targets_min": 619
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 3.8832,
|
| 2676 |
+
"grad_norm": 0.34359938713276317,
|
| 2677 |
+
"learning_rate": 5.826345039095178e-06,
|
| 2678 |
+
"loss": 0.3399,
|
| 2679 |
+
"loss_nan_ranks": 0,
|
| 2680 |
+
"loss_rank_avg": 0.16373825073242188,
|
| 2681 |
+
"step": 1215,
|
| 2682 |
+
"valid_targets_mean": 4538.1,
|
| 2683 |
+
"valid_targets_min": 695
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 3.8992,
|
| 2687 |
+
"grad_norm": 0.35681929563367065,
|
| 2688 |
+
"learning_rate": 5.669809922009937e-06,
|
| 2689 |
+
"loss": 0.3239,
|
| 2690 |
+
"loss_nan_ranks": 0,
|
| 2691 |
+
"loss_rank_avg": 0.1745469570159912,
|
| 2692 |
+
"step": 1220,
|
| 2693 |
+
"valid_targets_mean": 4123.0,
|
| 2694 |
+
"valid_targets_min": 960
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 3.9152,
|
| 2698 |
+
"grad_norm": 0.3230534112138346,
|
| 2699 |
+
"learning_rate": 5.515058340970665e-06,
|
| 2700 |
+
"loss": 0.3512,
|
| 2701 |
+
"loss_nan_ranks": 0,
|
| 2702 |
+
"loss_rank_avg": 0.16375651955604553,
|
| 2703 |
+
"step": 1225,
|
| 2704 |
+
"valid_targets_mean": 4488.4,
|
| 2705 |
+
"valid_targets_min": 1016
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 3.9312,
|
| 2709 |
+
"grad_norm": 0.3788271465767256,
|
| 2710 |
+
"learning_rate": 5.362109556365496e-06,
|
| 2711 |
+
"loss": 0.3407,
|
| 2712 |
+
"loss_nan_ranks": 0,
|
| 2713 |
+
"loss_rank_avg": 0.140263631939888,
|
| 2714 |
+
"step": 1230,
|
| 2715 |
+
"valid_targets_mean": 3269.2,
|
| 2716 |
+
"valid_targets_min": 752
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 3.9472,
|
| 2720 |
+
"grad_norm": 0.3908033970184151,
|
| 2721 |
+
"learning_rate": 5.2109826042064445e-06,
|
| 2722 |
+
"loss": 0.3464,
|
| 2723 |
+
"loss_nan_ranks": 0,
|
| 2724 |
+
"loss_rank_avg": 0.15427762269973755,
|
| 2725 |
+
"step": 1235,
|
| 2726 |
+
"valid_targets_mean": 2903.1,
|
| 2727 |
+
"valid_targets_min": 687
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 3.9632,
|
| 2731 |
+
"grad_norm": 0.31824838313855275,
|
| 2732 |
+
"learning_rate": 5.0616962937601945e-06,
|
| 2733 |
+
"loss": 0.3375,
|
| 2734 |
+
"loss_nan_ranks": 0,
|
| 2735 |
+
"loss_rank_avg": 0.17652173340320587,
|
| 2736 |
+
"step": 1240,
|
| 2737 |
+
"valid_targets_mean": 6038.0,
|
| 2738 |
+
"valid_targets_min": 711
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 3.9792,
|
| 2742 |
+
"grad_norm": 0.41835790896429187,
|
| 2743 |
+
"learning_rate": 4.914269205207076e-06,
|
| 2744 |
+
"loss": 0.34,
|
| 2745 |
+
"loss_nan_ranks": 0,
|
| 2746 |
+
"loss_rank_avg": 0.16578412055969238,
|
| 2747 |
+
"step": 1245,
|
| 2748 |
+
"valid_targets_mean": 3098.8,
|
| 2749 |
+
"valid_targets_min": 1084
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 3.9952,
|
| 2753 |
+
"grad_norm": 0.3239771649167043,
|
| 2754 |
+
"learning_rate": 4.76871968732858e-06,
|
| 2755 |
+
"loss": 0.3606,
|
| 2756 |
+
"loss_nan_ranks": 0,
|
| 2757 |
+
"loss_rank_avg": 0.15589842200279236,
|
| 2758 |
+
"step": 1250,
|
| 2759 |
+
"valid_targets_mean": 5305.8,
|
| 2760 |
+
"valid_targets_min": 1187
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 4.0096,
|
| 2764 |
+
"grad_norm": 0.35238292143174227,
|
| 2765 |
+
"learning_rate": 4.625065855223689e-06,
|
| 2766 |
+
"loss": 0.3435,
|
| 2767 |
+
"loss_nan_ranks": 0,
|
| 2768 |
+
"loss_rank_avg": 0.1814814805984497,
|
| 2769 |
+
"step": 1255,
|
| 2770 |
+
"valid_targets_mean": 4887.1,
|
| 2771 |
+
"valid_targets_min": 847
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 4.0256,
|
| 2775 |
+
"grad_norm": 0.45008459085343927,
|
| 2776 |
+
"learning_rate": 4.483325588054259e-06,
|
| 2777 |
+
"loss": 0.3363,
|
| 2778 |
+
"loss_nan_ranks": 0,
|
| 2779 |
+
"loss_rank_avg": 0.17144045233726501,
|
| 2780 |
+
"step": 1260,
|
| 2781 |
+
"valid_targets_mean": 3204.2,
|
| 2782 |
+
"valid_targets_min": 489
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 4.0416,
|
| 2786 |
+
"grad_norm": 0.38402873507610885,
|
| 2787 |
+
"learning_rate": 4.343516526819755e-06,
|
| 2788 |
+
"loss": 0.3283,
|
| 2789 |
+
"loss_nan_ranks": 0,
|
| 2790 |
+
"loss_rank_avg": 0.16901005804538727,
|
| 2791 |
+
"step": 1265,
|
| 2792 |
+
"valid_targets_mean": 3899.9,
|
| 2793 |
+
"valid_targets_min": 731
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 4.0576,
|
| 2797 |
+
"grad_norm": 0.3788020921136641,
|
| 2798 |
+
"learning_rate": 4.205656072161681e-06,
|
| 2799 |
+
"loss": 0.315,
|
| 2800 |
+
"loss_nan_ranks": 0,
|
| 2801 |
+
"loss_rank_avg": 0.2033902406692505,
|
| 2802 |
+
"step": 1270,
|
| 2803 |
+
"valid_targets_mean": 5402.2,
|
| 2804 |
+
"valid_targets_min": 767
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 4.0736,
|
| 2808 |
+
"grad_norm": 0.3924558007635478,
|
| 2809 |
+
"learning_rate": 4.069761382197901e-06,
|
| 2810 |
+
"loss": 0.3473,
|
| 2811 |
+
"loss_nan_ranks": 0,
|
| 2812 |
+
"loss_rank_avg": 0.14622555673122406,
|
| 2813 |
+
"step": 1275,
|
| 2814 |
+
"valid_targets_mean": 3266.4,
|
| 2815 |
+
"valid_targets_min": 472
|
| 2816 |
+
},
|
| 2817 |
+
{
|
| 2818 |
+
"epoch": 4.0896,
|
| 2819 |
+
"grad_norm": 0.30034356754743674,
|
| 2820 |
+
"learning_rate": 3.935849370387104e-06,
|
| 2821 |
+
"loss": 0.3218,
|
| 2822 |
+
"loss_nan_ranks": 0,
|
| 2823 |
+
"loss_rank_avg": 0.15228131413459778,
|
| 2824 |
+
"step": 1280,
|
| 2825 |
+
"valid_targets_mean": 5391.9,
|
| 2826 |
+
"valid_targets_min": 847
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 4.1056,
|
| 2830 |
+
"grad_norm": 0.3250917202336007,
|
| 2831 |
+
"learning_rate": 3.803936703423783e-06,
|
| 2832 |
+
"loss": 0.3253,
|
| 2833 |
+
"loss_nan_ranks": 0,
|
| 2834 |
+
"loss_rank_avg": 0.14912036061286926,
|
| 2835 |
+
"step": 1285,
|
| 2836 |
+
"valid_targets_mean": 5266.2,
|
| 2837 |
+
"valid_targets_min": 914
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 4.1216,
|
| 2841 |
+
"grad_norm": 0.31424788351984845,
|
| 2842 |
+
"learning_rate": 3.6740397991638864e-06,
|
| 2843 |
+
"loss": 0.3152,
|
| 2844 |
+
"loss_nan_ranks": 0,
|
| 2845 |
+
"loss_rank_avg": 0.14228126406669617,
|
| 2846 |
+
"step": 1290,
|
| 2847 |
+
"valid_targets_mean": 4604.3,
|
| 2848 |
+
"valid_targets_min": 936
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 4.1376,
|
| 2852 |
+
"grad_norm": 0.36643651361028096,
|
| 2853 |
+
"learning_rate": 3.5461748245814633e-06,
|
| 2854 |
+
"loss": 0.3371,
|
| 2855 |
+
"loss_nan_ranks": 0,
|
| 2856 |
+
"loss_rank_avg": 0.1840534508228302,
|
| 2857 |
+
"step": 1295,
|
| 2858 |
+
"valid_targets_mean": 4418.9,
|
| 2859 |
+
"valid_targets_min": 713
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 4.1536,
|
| 2863 |
+
"grad_norm": 0.4546431971157884,
|
| 2864 |
+
"learning_rate": 3.420357693756502e-06,
|
| 2865 |
+
"loss": 0.3471,
|
| 2866 |
+
"loss_nan_ranks": 0,
|
| 2867 |
+
"loss_rank_avg": 0.17221775650978088,
|
| 2868 |
+
"step": 1300,
|
| 2869 |
+
"valid_targets_mean": 3512.8,
|
| 2870 |
+
"valid_targets_min": 728
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 4.1696,
|
| 2874 |
+
"grad_norm": 0.3805593027549495,
|
| 2875 |
+
"learning_rate": 3.2966040658942666e-06,
|
| 2876 |
+
"loss": 0.3371,
|
| 2877 |
+
"loss_nan_ranks": 0,
|
| 2878 |
+
"loss_rank_avg": 0.1598142385482788,
|
| 2879 |
+
"step": 1305,
|
| 2880 |
+
"valid_targets_mean": 3979.7,
|
| 2881 |
+
"valid_targets_min": 857
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 4.1856,
|
| 2885 |
+
"grad_norm": 0.3335980469652557,
|
| 2886 |
+
"learning_rate": 3.174929343376374e-06,
|
| 2887 |
+
"loss": 0.3536,
|
| 2888 |
+
"loss_nan_ranks": 0,
|
| 2889 |
+
"loss_rank_avg": 0.20903745293617249,
|
| 2890 |
+
"step": 1310,
|
| 2891 |
+
"valid_targets_mean": 6657.6,
|
| 2892 |
+
"valid_targets_min": 569
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 4.2016,
|
| 2896 |
+
"grad_norm": 0.3604800331212591,
|
| 2897 |
+
"learning_rate": 3.055348669843794e-06,
|
| 2898 |
+
"loss": 0.3187,
|
| 2899 |
+
"loss_nan_ranks": 0,
|
| 2900 |
+
"loss_rank_avg": 0.11375663429498672,
|
| 2901 |
+
"step": 1315,
|
| 2902 |
+
"valid_targets_mean": 3103.1,
|
| 2903 |
+
"valid_targets_min": 448
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 4.2176,
|
| 2907 |
+
"grad_norm": 0.28671003001950635,
|
| 2908 |
+
"learning_rate": 2.937876928312062e-06,
|
| 2909 |
+
"loss": 0.3431,
|
| 2910 |
+
"loss_nan_ranks": 0,
|
| 2911 |
+
"loss_rank_avg": 0.11587075144052505,
|
| 2912 |
+
"step": 1320,
|
| 2913 |
+
"valid_targets_mean": 4968.6,
|
| 2914 |
+
"valid_targets_min": 861
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 4.2336,
|
| 2918 |
+
"grad_norm": 0.4017656623883969,
|
| 2919 |
+
"learning_rate": 2.8225287393189547e-06,
|
| 2920 |
+
"loss": 0.3269,
|
| 2921 |
+
"loss_nan_ranks": 0,
|
| 2922 |
+
"loss_rank_avg": 0.17351335287094116,
|
| 2923 |
+
"step": 1325,
|
| 2924 |
+
"valid_targets_mean": 3881.8,
|
| 2925 |
+
"valid_targets_min": 1049
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 4.2496,
|
| 2929 |
+
"grad_norm": 0.373357304461226,
|
| 2930 |
+
"learning_rate": 2.709318459104815e-06,
|
| 2931 |
+
"loss": 0.3235,
|
| 2932 |
+
"loss_nan_ranks": 0,
|
| 2933 |
+
"loss_rank_avg": 0.174965038895607,
|
| 2934 |
+
"step": 1330,
|
| 2935 |
+
"valid_targets_mean": 4631.8,
|
| 2936 |
+
"valid_targets_min": 535
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 4.2656,
|
| 2940 |
+
"grad_norm": 0.4122717520251167,
|
| 2941 |
+
"learning_rate": 2.5982601778257733e-06,
|
| 2942 |
+
"loss": 0.3388,
|
| 2943 |
+
"loss_nan_ranks": 0,
|
| 2944 |
+
"loss_rank_avg": 0.2125864326953888,
|
| 2945 |
+
"step": 1335,
|
| 2946 |
+
"valid_targets_mean": 3899.5,
|
| 2947 |
+
"valid_targets_min": 950
|
| 2948 |
+
},
|
| 2949 |
+
{
|
| 2950 |
+
"epoch": 4.2816,
|
| 2951 |
+
"grad_norm": 0.3495807169302053,
|
| 2952 |
+
"learning_rate": 2.4893677178000797e-06,
|
| 2953 |
+
"loss": 0.3223,
|
| 2954 |
+
"loss_nan_ranks": 0,
|
| 2955 |
+
"loss_rank_avg": 0.21441903710365295,
|
| 2956 |
+
"step": 1340,
|
| 2957 |
+
"valid_targets_mean": 5717.9,
|
| 2958 |
+
"valid_targets_min": 711
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 4.2976,
|
| 2962 |
+
"grad_norm": 0.4388515044874441,
|
| 2963 |
+
"learning_rate": 2.3826546317877795e-06,
|
| 2964 |
+
"loss": 0.3761,
|
| 2965 |
+
"loss_nan_ranks": 0,
|
| 2966 |
+
"loss_rank_avg": 0.1932186335325241,
|
| 2967 |
+
"step": 1345,
|
| 2968 |
+
"valid_targets_mean": 3286.0,
|
| 2969 |
+
"valid_targets_min": 567
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 4.3136,
|
| 2973 |
+
"grad_norm": 0.34611942555321196,
|
| 2974 |
+
"learning_rate": 2.278134201303952e-06,
|
| 2975 |
+
"loss": 0.3265,
|
| 2976 |
+
"loss_nan_ranks": 0,
|
| 2977 |
+
"loss_rank_avg": 0.21352379024028778,
|
| 2978 |
+
"step": 1350,
|
| 2979 |
+
"valid_targets_mean": 5044.2,
|
| 2980 |
+
"valid_targets_min": 558
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"epoch": 4.3296,
|
| 2984 |
+
"grad_norm": 0.3551898027835839,
|
| 2985 |
+
"learning_rate": 2.1758194349656624e-06,
|
| 2986 |
+
"loss": 0.357,
|
| 2987 |
+
"loss_nan_ranks": 0,
|
| 2988 |
+
"loss_rank_avg": 0.16915194690227509,
|
| 2989 |
+
"step": 1355,
|
| 2990 |
+
"valid_targets_mean": 4792.9,
|
| 2991 |
+
"valid_targets_min": 655
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 4.3456,
|
| 2995 |
+
"grad_norm": 0.39252254953522203,
|
| 2996 |
+
"learning_rate": 2.075723066872939e-06,
|
| 2997 |
+
"loss": 0.3279,
|
| 2998 |
+
"loss_nan_ranks": 0,
|
| 2999 |
+
"loss_rank_avg": 0.13038407266139984,
|
| 3000 |
+
"step": 1360,
|
| 3001 |
+
"valid_targets_mean": 3598.0,
|
| 3002 |
+
"valid_targets_min": 729
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 4.3616,
|
| 3006 |
+
"grad_norm": 0.33078484844023676,
|
| 3007 |
+
"learning_rate": 1.977857555023854e-06,
|
| 3008 |
+
"loss": 0.3422,
|
| 3009 |
+
"loss_nan_ranks": 0,
|
| 3010 |
+
"loss_rank_avg": 0.09628881514072418,
|
| 3011 |
+
"step": 1365,
|
| 3012 |
+
"valid_targets_mean": 3414.8,
|
| 3013 |
+
"valid_targets_min": 499
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 4.3776,
|
| 3017 |
+
"grad_norm": 0.3784326265909576,
|
| 3018 |
+
"learning_rate": 1.8822350797640543e-06,
|
| 3019 |
+
"loss": 0.3449,
|
| 3020 |
+
"loss_nan_ranks": 0,
|
| 3021 |
+
"loss_rank_avg": 0.15875539183616638,
|
| 3022 |
+
"step": 1370,
|
| 3023 |
+
"valid_targets_mean": 3690.6,
|
| 3024 |
+
"valid_targets_min": 721
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 4.3936,
|
| 3028 |
+
"grad_norm": 0.35061428660208377,
|
| 3029 |
+
"learning_rate": 1.788867542270729e-06,
|
| 3030 |
+
"loss": 0.3443,
|
| 3031 |
+
"loss_nan_ranks": 0,
|
| 3032 |
+
"loss_rank_avg": 0.17543794214725494,
|
| 3033 |
+
"step": 1375,
|
| 3034 |
+
"valid_targets_mean": 4049.7,
|
| 3035 |
+
"valid_targets_min": 427
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 4.4096,
|
| 3039 |
+
"grad_norm": 0.38301901048882014,
|
| 3040 |
+
"learning_rate": 1.6977665630714345e-06,
|
| 3041 |
+
"loss": 0.3392,
|
| 3042 |
+
"loss_nan_ranks": 0,
|
| 3043 |
+
"loss_rank_avg": 0.1260443776845932,
|
| 3044 |
+
"step": 1380,
|
| 3045 |
+
"valid_targets_mean": 2746.2,
|
| 3046 |
+
"valid_targets_min": 734
|
| 3047 |
+
},
|
| 3048 |
+
{
|
| 3049 |
+
"epoch": 4.4256,
|
| 3050 |
+
"grad_norm": 0.39167736257806884,
|
| 3051 |
+
"learning_rate": 1.6089434805977799e-06,
|
| 3052 |
+
"loss": 0.3486,
|
| 3053 |
+
"loss_nan_ranks": 0,
|
| 3054 |
+
"loss_rank_avg": 0.19264991581439972,
|
| 3055 |
+
"step": 1385,
|
| 3056 |
+
"valid_targets_mean": 4090.4,
|
| 3057 |
+
"valid_targets_min": 732
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 4.4416,
|
| 3061 |
+
"grad_norm": 0.3724692418976962,
|
| 3062 |
+
"learning_rate": 1.5224093497742654e-06,
|
| 3063 |
+
"loss": 0.3473,
|
| 3064 |
+
"loss_nan_ranks": 0,
|
| 3065 |
+
"loss_rank_avg": 0.19777175784111023,
|
| 3066 |
+
"step": 1390,
|
| 3067 |
+
"valid_targets_mean": 4353.8,
|
| 3068 |
+
"valid_targets_min": 682
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 4.4576,
|
| 3072 |
+
"grad_norm": 2.915445545430493,
|
| 3073 |
+
"learning_rate": 1.4381749406423695e-06,
|
| 3074 |
+
"loss": 0.3411,
|
| 3075 |
+
"loss_nan_ranks": 0,
|
| 3076 |
+
"loss_rank_avg": 0.19258227944374084,
|
| 3077 |
+
"step": 1395,
|
| 3078 |
+
"valid_targets_mean": 4315.4,
|
| 3079 |
+
"valid_targets_min": 855
|
| 3080 |
+
},
|
| 3081 |
+
{
|
| 3082 |
+
"epoch": 4.4736,
|
| 3083 |
+
"grad_norm": 0.3805102691027801,
|
| 3084 |
+
"learning_rate": 1.3562507370201062e-06,
|
| 3085 |
+
"loss": 0.3402,
|
| 3086 |
+
"loss_nan_ranks": 0,
|
| 3087 |
+
"loss_rank_avg": 0.18339085578918457,
|
| 3088 |
+
"step": 1400,
|
| 3089 |
+
"valid_targets_mean": 3903.1,
|
| 3090 |
+
"valid_targets_min": 764
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 4.4896,
|
| 3094 |
+
"grad_norm": 0.37129014250841613,
|
| 3095 |
+
"learning_rate": 1.2766469351972345e-06,
|
| 3096 |
+
"loss": 0.3156,
|
| 3097 |
+
"loss_nan_ranks": 0,
|
| 3098 |
+
"loss_rank_avg": 0.15153226256370544,
|
| 3099 |
+
"step": 1405,
|
| 3100 |
+
"valid_targets_mean": 3064.1,
|
| 3101 |
+
"valid_targets_min": 699
|
| 3102 |
+
},
|
| 3103 |
+
{
|
| 3104 |
+
"epoch": 4.5056,
|
| 3105 |
+
"grad_norm": 0.3726092816106853,
|
| 3106 |
+
"learning_rate": 1.1993734426661985e-06,
|
| 3107 |
+
"loss": 0.3506,
|
| 3108 |
+
"loss_nan_ranks": 0,
|
| 3109 |
+
"loss_rank_avg": 0.1535291075706482,
|
| 3110 |
+
"step": 1410,
|
| 3111 |
+
"valid_targets_mean": 3952.9,
|
| 3112 |
+
"valid_targets_min": 543
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 4.5216,
|
| 3116 |
+
"grad_norm": 0.3591276212261323,
|
| 3117 |
+
"learning_rate": 1.1244398768890496e-06,
|
| 3118 |
+
"loss": 0.3161,
|
| 3119 |
+
"loss_nan_ranks": 0,
|
| 3120 |
+
"loss_rank_avg": 0.1446693241596222,
|
| 3121 |
+
"step": 1415,
|
| 3122 |
+
"valid_targets_mean": 3641.4,
|
| 3123 |
+
"valid_targets_min": 892
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 4.5376,
|
| 3127 |
+
"grad_norm": 0.3912764396308065,
|
| 3128 |
+
"learning_rate": 1.0518555641004613e-06,
|
| 3129 |
+
"loss": 0.3421,
|
| 3130 |
+
"loss_nan_ranks": 0,
|
| 3131 |
+
"loss_rank_avg": 0.10798218846321106,
|
| 3132 |
+
"step": 1420,
|
| 3133 |
+
"valid_targets_mean": 2451.6,
|
| 3134 |
+
"valid_targets_min": 699
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 4.5536,
|
| 3138 |
+
"grad_norm": 0.3411240761105926,
|
| 3139 |
+
"learning_rate": 9.816295381469954e-07,
|
| 3140 |
+
"loss": 0.3401,
|
| 3141 |
+
"loss_nan_ranks": 0,
|
| 3142 |
+
"loss_rank_avg": 0.19344064593315125,
|
| 3143 |
+
"step": 1425,
|
| 3144 |
+
"valid_targets_mean": 5260.7,
|
| 3145 |
+
"valid_targets_min": 430
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 4.5696,
|
| 3149 |
+
"grad_norm": 0.3991311146571275,
|
| 3150 |
+
"learning_rate": 9.137705393627239e-07,
|
| 3151 |
+
"loss": 0.3458,
|
| 3152 |
+
"loss_nan_ranks": 0,
|
| 3153 |
+
"loss_rank_avg": 0.17564912140369415,
|
| 3154 |
+
"step": 1430,
|
| 3155 |
+
"valid_targets_mean": 3772.7,
|
| 3156 |
+
"valid_targets_min": 771
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 4.5856,
|
| 3160 |
+
"grad_norm": 0.4907284404134834,
|
| 3161 |
+
"learning_rate": 8.482870134814214e-07,
|
| 3162 |
+
"loss": 0.3632,
|
| 3163 |
+
"loss_nan_ranks": 0,
|
| 3164 |
+
"loss_rank_avg": 0.1778586059808731,
|
| 3165 |
+
"step": 1435,
|
| 3166 |
+
"valid_targets_mean": 2586.2,
|
| 3167 |
+
"valid_targets_min": 1050
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 4.6016,
|
| 3171 |
+
"grad_norm": 0.3983087034279942,
|
| 3172 |
+
"learning_rate": 7.851871105854125e-07,
|
| 3173 |
+
"loss": 0.3218,
|
| 3174 |
+
"loss_nan_ranks": 0,
|
| 3175 |
+
"loss_rank_avg": 0.19727054238319397,
|
| 3176 |
+
"step": 1440,
|
| 3177 |
+
"valid_targets_mean": 4347.9,
|
| 3178 |
+
"valid_targets_min": 649
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 4.6176,
|
| 3182 |
+
"grad_norm": 0.3428497731326709,
|
| 3183 |
+
"learning_rate": 7.244786840912033e-07,
|
| 3184 |
+
"loss": 0.3274,
|
| 3185 |
+
"loss_nan_ranks": 0,
|
| 3186 |
+
"loss_rank_avg": 0.17086507380008698,
|
| 3187 |
+
"step": 1445,
|
| 3188 |
+
"valid_targets_mean": 4404.5,
|
| 3189 |
+
"valid_targets_min": 953
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 4.6336,
|
| 3193 |
+
"grad_norm": 0.40290620049486076,
|
| 3194 |
+
"learning_rate": 6.661692897720517e-07,
|
| 3195 |
+
"loss": 0.3457,
|
| 3196 |
+
"loss_nan_ranks": 0,
|
| 3197 |
+
"loss_rank_avg": 0.20860999822616577,
|
| 3198 |
+
"step": 1450,
|
| 3199 |
+
"valid_targets_mean": 4176.6,
|
| 3200 |
+
"valid_targets_min": 840
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 4.6495999999999995,
|
| 3204 |
+
"grad_norm": 0.3960734519984612,
|
| 3205 |
+
"learning_rate": 6.10266184817565e-07,
|
| 3206 |
+
"loss": 0.3496,
|
| 3207 |
+
"loss_nan_ranks": 0,
|
| 3208 |
+
"loss_rank_avg": 0.12197977304458618,
|
| 3209 |
+
"step": 1455,
|
| 3210 |
+
"valid_targets_mean": 3047.1,
|
| 3211 |
+
"valid_targets_min": 706
|
| 3212 |
+
},
|
| 3213 |
+
{
|
| 3214 |
+
"epoch": 4.6655999999999995,
|
| 3215 |
+
"grad_norm": 0.3969892477990081,
|
| 3216 |
+
"learning_rate": 5.567763269304927e-07,
|
| 3217 |
+
"loss": 0.3343,
|
| 3218 |
+
"loss_nan_ranks": 0,
|
| 3219 |
+
"loss_rank_avg": 0.19510459899902344,
|
| 3220 |
+
"step": 1460,
|
| 3221 |
+
"valid_targets_mean": 4116.6,
|
| 3222 |
+
"valid_targets_min": 707
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 4.6815999999999995,
|
| 3226 |
+
"grad_norm": 0.4355543500009561,
|
| 3227 |
+
"learning_rate": 5.057063734607392e-07,
|
| 3228 |
+
"loss": 0.3518,
|
| 3229 |
+
"loss_nan_ranks": 0,
|
| 3230 |
+
"loss_rank_avg": 0.14991804957389832,
|
| 3231 |
+
"step": 1465,
|
| 3232 |
+
"valid_targets_mean": 3097.3,
|
| 3233 |
+
"valid_targets_min": 725
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 4.6975999999999996,
|
| 3237 |
+
"grad_norm": 0.35587447872188005,
|
| 3238 |
+
"learning_rate": 4.570626805768119e-07,
|
| 3239 |
+
"loss": 0.3197,
|
| 3240 |
+
"loss_nan_ranks": 0,
|
| 3241 |
+
"loss_rank_avg": 0.17723935842514038,
|
| 3242 |
+
"step": 1470,
|
| 3243 |
+
"valid_targets_mean": 5329.2,
|
| 3244 |
+
"valid_targets_min": 706
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 4.7136,
|
| 3248 |
+
"grad_norm": 0.39446481317079257,
|
| 3249 |
+
"learning_rate": 4.1085130247472625e-07,
|
| 3250 |
+
"loss": 0.324,
|
| 3251 |
+
"loss_nan_ranks": 0,
|
| 3252 |
+
"loss_rank_avg": 0.14012877643108368,
|
| 3253 |
+
"step": 1475,
|
| 3254 |
+
"valid_targets_mean": 3485.1,
|
| 3255 |
+
"valid_targets_min": 681
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 4.7296,
|
| 3259 |
+
"grad_norm": 0.4072062654615722,
|
| 3260 |
+
"learning_rate": 3.670779906244981e-07,
|
| 3261 |
+
"loss": 0.3271,
|
| 3262 |
+
"loss_nan_ranks": 0,
|
| 3263 |
+
"loss_rank_avg": 0.19580386579036713,
|
| 3264 |
+
"step": 1480,
|
| 3265 |
+
"valid_targets_mean": 4416.1,
|
| 3266 |
+
"valid_targets_min": 776
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 4.7456,
|
| 3270 |
+
"grad_norm": 0.3084493418868167,
|
| 3271 |
+
"learning_rate": 3.2574819305432713e-07,
|
| 3272 |
+
"loss": 0.3248,
|
| 3273 |
+
"loss_nan_ranks": 0,
|
| 3274 |
+
"loss_rank_avg": 0.16733351349830627,
|
| 3275 |
+
"step": 1485,
|
| 3276 |
+
"valid_targets_mean": 4452.8,
|
| 3277 |
+
"valid_targets_min": 698
|
| 3278 |
+
},
|
| 3279 |
+
{
|
| 3280 |
+
"epoch": 4.7616,
|
| 3281 |
+
"grad_norm": 0.3598510249644468,
|
| 3282 |
+
"learning_rate": 2.8686705367250824e-07,
|
| 3283 |
+
"loss": 0.3355,
|
| 3284 |
+
"loss_nan_ranks": 0,
|
| 3285 |
+
"loss_rank_avg": 0.14451885223388672,
|
| 3286 |
+
"step": 1490,
|
| 3287 |
+
"valid_targets_mean": 2834.7,
|
| 3288 |
+
"valid_targets_min": 632
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 4.7776,
|
| 3292 |
+
"grad_norm": 0.4461356647281616,
|
| 3293 |
+
"learning_rate": 2.504394116272502e-07,
|
| 3294 |
+
"loss": 0.3403,
|
| 3295 |
+
"loss_nan_ranks": 0,
|
| 3296 |
+
"loss_rank_avg": 0.16207420825958252,
|
| 3297 |
+
"step": 1495,
|
| 3298 |
+
"valid_targets_mean": 2442.2,
|
| 3299 |
+
"valid_targets_min": 846
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 4.7936,
|
| 3303 |
+
"grad_norm": 0.3939489927901413,
|
| 3304 |
+
"learning_rate": 2.1646980070437973e-07,
|
| 3305 |
+
"loss": 0.323,
|
| 3306 |
+
"loss_nan_ranks": 0,
|
| 3307 |
+
"loss_rank_avg": 0.22261710464954376,
|
| 3308 |
+
"step": 1500,
|
| 3309 |
+
"valid_targets_mean": 4795.7,
|
| 3310 |
+
"valid_targets_min": 944
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 4.8096,
|
| 3314 |
+
"grad_norm": 0.3578185640710523,
|
| 3315 |
+
"learning_rate": 1.8496244876306858e-07,
|
| 3316 |
+
"loss": 0.3259,
|
| 3317 |
+
"loss_nan_ranks": 0,
|
| 3318 |
+
"loss_rank_avg": 0.12147246301174164,
|
| 3319 |
+
"step": 1505,
|
| 3320 |
+
"valid_targets_mean": 2807.9,
|
| 3321 |
+
"valid_targets_min": 922
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 4.8256,
|
| 3325 |
+
"grad_norm": 0.3669760845147576,
|
| 3326 |
+
"learning_rate": 1.559212772096319e-07,
|
| 3327 |
+
"loss": 0.331,
|
| 3328 |
+
"loss_nan_ranks": 0,
|
| 3329 |
+
"loss_rank_avg": 0.13725429773330688,
|
| 3330 |
+
"step": 1510,
|
| 3331 |
+
"valid_targets_mean": 3579.9,
|
| 3332 |
+
"valid_targets_min": 1080
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 4.8416,
|
| 3336 |
+
"grad_norm": 0.37618766080255067,
|
| 3337 |
+
"learning_rate": 1.2934990050947228e-07,
|
| 3338 |
+
"loss": 0.3334,
|
| 3339 |
+
"loss_nan_ranks": 0,
|
| 3340 |
+
"loss_rank_avg": 0.15945936739444733,
|
| 3341 |
+
"step": 1515,
|
| 3342 |
+
"valid_targets_mean": 4211.1,
|
| 3343 |
+
"valid_targets_min": 660
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 4.8576,
|
| 3347 |
+
"grad_norm": 0.498977453048548,
|
| 3348 |
+
"learning_rate": 1.0525162573723269e-07,
|
| 3349 |
+
"loss": 0.3489,
|
| 3350 |
+
"loss_nan_ranks": 0,
|
| 3351 |
+
"loss_rank_avg": 0.15219208598136902,
|
| 3352 |
+
"step": 1520,
|
| 3353 |
+
"valid_targets_mean": 2302.8,
|
| 3354 |
+
"valid_targets_min": 673
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 4.8736,
|
| 3358 |
+
"grad_norm": 0.3680209079970337,
|
| 3359 |
+
"learning_rate": 8.362945216517704e-08,
|
| 3360 |
+
"loss": 0.3326,
|
| 3361 |
+
"loss_nan_ranks": 0,
|
| 3362 |
+
"loss_rank_avg": 0.19316397607326508,
|
| 3363 |
+
"step": 1525,
|
| 3364 |
+
"valid_targets_mean": 4751.6,
|
| 3365 |
+
"valid_targets_min": 829
|
| 3366 |
+
},
|
| 3367 |
+
{
|
| 3368 |
+
"epoch": 4.8896,
|
| 3369 |
+
"grad_norm": 0.3308997729831502,
|
| 3370 |
+
"learning_rate": 6.448607088991532e-08,
|
| 3371 |
+
"loss": 0.3455,
|
| 3372 |
+
"loss_nan_ranks": 0,
|
| 3373 |
+
"loss_rank_avg": 0.1902647465467453,
|
| 3374 |
+
"step": 1530,
|
| 3375 |
+
"valid_targets_mean": 6998.1,
|
| 3376 |
+
"valid_targets_min": 1277
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 4.9056,
|
| 3380 |
+
"grad_norm": 0.3690095050432829,
|
| 3381 |
+
"learning_rate": 4.782386449746934e-08,
|
| 3382 |
+
"loss": 0.3253,
|
| 3383 |
+
"loss_nan_ranks": 0,
|
| 3384 |
+
"loss_rank_avg": 0.128545343875885,
|
| 3385 |
+
"step": 1535,
|
| 3386 |
+
"valid_targets_mean": 3642.2,
|
| 3387 |
+
"valid_targets_min": 677
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 4.9216,
|
| 3391 |
+
"grad_norm": 0.4085506322401526,
|
| 3392 |
+
"learning_rate": 3.3644906766734374e-08,
|
| 3393 |
+
"loss": 0.334,
|
| 3394 |
+
"loss_nan_ranks": 0,
|
| 3395 |
+
"loss_rank_avg": 0.20792028307914734,
|
| 3396 |
+
"step": 1540,
|
| 3397 |
+
"valid_targets_mean": 4389.9,
|
| 3398 |
+
"valid_targets_min": 275
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 4.9376,
|
| 3402 |
+
"grad_norm": 0.41617364004087926,
|
| 3403 |
+
"learning_rate": 2.1950962411367848e-08,
|
| 3404 |
+
"loss": 0.3283,
|
| 3405 |
+
"loss_nan_ranks": 0,
|
| 3406 |
+
"loss_rank_avg": 0.20928677916526794,
|
| 3407 |
+
"step": 1545,
|
| 3408 |
+
"valid_targets_mean": 4790.5,
|
| 3409 |
+
"valid_targets_min": 1106
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 4.9536,
|
| 3413 |
+
"grad_norm": 0.3449113707606809,
|
| 3414 |
+
"learning_rate": 1.2743486860165022e-08,
|
| 3415 |
+
"loss": 0.3304,
|
| 3416 |
+
"loss_nan_ranks": 0,
|
| 3417 |
+
"loss_rank_avg": 0.18289443850517273,
|
| 3418 |
+
"step": 1550,
|
| 3419 |
+
"valid_targets_mean": 4802.8,
|
| 3420 |
+
"valid_targets_min": 823
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 4.9696,
|
| 3424 |
+
"grad_norm": 0.39524375236617265,
|
| 3425 |
+
"learning_rate": 6.023626075915001e-09,
|
| 3426 |
+
"loss": 0.3408,
|
| 3427 |
+
"loss_nan_ranks": 0,
|
| 3428 |
+
"loss_rank_avg": 0.18949632346630096,
|
| 3429 |
+
"step": 1555,
|
| 3430 |
+
"valid_targets_mean": 3546.4,
|
| 3431 |
+
"valid_targets_min": 907
|
| 3432 |
+
},
|
| 3433 |
+
{
|
| 3434 |
+
"epoch": 4.9856,
|
| 3435 |
+
"grad_norm": 0.3612907490017341,
|
| 3436 |
+
"learning_rate": 1.7922164127659457e-09,
|
| 3437 |
+
"loss": 0.3266,
|
| 3438 |
+
"loss_nan_ranks": 0,
|
| 3439 |
+
"loss_rank_avg": 0.21476075053215027,
|
| 3440 |
+
"step": 1560,
|
| 3441 |
+
"valid_targets_mean": 5423.8,
|
| 3442 |
+
"valid_targets_min": 711
|
| 3443 |
+
},
|
| 3444 |
+
{
|
| 3445 |
+
"epoch": 5.0,
|
| 3446 |
+
"grad_norm": 0.5832122688927835,
|
| 3447 |
+
"learning_rate": 4.978451213499824e-11,
|
| 3448 |
+
"loss": 0.337,
|
| 3449 |
+
"loss_nan_ranks": 0,
|
| 3450 |
+
"loss_rank_avg": 0.3492066562175751,
|
| 3451 |
+
"step": 1565,
|
| 3452 |
+
"valid_targets_mean": 3627.8,
|
| 3453 |
+
"valid_targets_min": 954
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 5.0,
|
| 3457 |
+
"loss_nan_ranks": 0,
|
| 3458 |
+
"loss_rank_avg": 0.3492066562175751,
|
| 3459 |
+
"step": 1565,
|
| 3460 |
+
"total_flos": 7.336290139934556e+17,
|
| 3461 |
+
"train_loss": 0.37341142744301986,
|
| 3462 |
+
"train_runtime": 17388.125,
|
| 3463 |
+
"train_samples_per_second": 2.876,
|
| 3464 |
+
"train_steps_per_second": 0.09,
|
| 3465 |
+
"valid_targets_mean": 3627.8,
|
| 3466 |
+
"valid_targets_min": 954
|
| 3467 |
+
}
|
| 3468 |
+
],
|
| 3469 |
+
"logging_steps": 5,
|
| 3470 |
+
"max_steps": 1565,
|
| 3471 |
+
"num_input_tokens_seen": 0,
|
| 3472 |
+
"num_train_epochs": 5,
|
| 3473 |
+
"save_steps": 500,
|
| 3474 |
+
"stateful_callbacks": {
|
| 3475 |
+
"TrainerControl": {
|
| 3476 |
+
"args": {
|
| 3477 |
+
"should_epoch_stop": false,
|
| 3478 |
+
"should_evaluate": false,
|
| 3479 |
+
"should_log": false,
|
| 3480 |
+
"should_save": false,
|
| 3481 |
+
"should_training_stop": false
|
| 3482 |
+
},
|
| 3483 |
+
"attributes": {}
|
| 3484 |
+
}
|
| 3485 |
+
},
|
| 3486 |
+
"total_flos": 7.336290139934556e+17,
|
| 3487 |
+
"train_batch_size": 1,
|
| 3488 |
+
"trial_name": null,
|
| 3489 |
+
"trial_params": null
|
| 3490 |
+
}
|
training_loss.png
ADDED
|