Super-squash branch 'main' using huggingface_hub

.gitattributes

@@ -0,0 +1,37 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+val_progress.png filter=lfs diff=lfs merge=lfs -text
+final/run.log filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,16 @@
+# fruit-phase1-ckpt — live training checkpoints
+
+Rolling checkpoints of the **GLM-5.2-SIQ-Fruit** Phase-1 pretrain
+(5.04B-param GLM-5.2 architectural mimic; see
+[proxy-fruit](https://github.com/malaiwah/proxy-fruit) for the full
+program, trainer, and measured findings).
+
+![training progress](val_progress.png)
+
+- `checkpoints/fruit_v1_main_ckpt.pt` — latest full state (model +
+  optimizer + step + tokens_seen + RNG), pushed every 600 steps.
+- `logs/train_metrics.log` — the durable metrics ledger (val sweeps +
+  step lines), merged across nodes so history survives spot preemption
+  and tier changes.
+- Plot regenerates ~3-hourly during training; final stage weights land
+  under `final/`.

checkpoints/fruit_v1_final_ckpt.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:308393ff31dd41c01965948b2bc7c40a5e8917c6b9b796417ec56f03b79a70e0
+size 20319774863

checkpoints/fruit_v1_instruct_ckpt.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffa9eea5d3bb02a4bd4b343ad1fa83282129f01b80f521610d3a9ba2c451f17d
+size 20195740548

checkpoints/fruit_v1_long_ckpt.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd0f43cb93677c8f31d1dd62c156cdf7a5a0f5dd041829782494911d4e049a22
+size 20195736168

checkpoints/fruit_v1_main_ckpt.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7b9a1bded5defc8f4968a1fa8979d3906c8e384034a9442ecc3cc8a19595169
+size 20195726070

final/fruit_v1_annealed.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98ac7cb4f7799194424782b505d622069fecf4dbca5f5acb2658f2a66c3631f6
+size 10080855259

final/fruit_v1_final.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:670a760a26fc7f335512e8d151ed7dc16c04d65d8e25089f1fd0639904036638
+size 10080854344

final/fruit_v1_instruct.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32dbf82d40b88a92b8dccd563c593b5971be358cf11895eb150f18644ff93c27
+size 10080855259

final/fruit_v1_instruct_ckpt.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9f043b223280004096395f0efb1992751eb60b0b649ba82c55398739970cf9b
+size 20195740164

final/fruit_v1_long.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9ed21645b5a2f803bd376ded1a1934959362ea71ac47735361863ddad68c3f8
+size 10080854039

final/fruit_v1_main.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09a5d8b54397b461d5bcf0353e5b95d9b61c9b4127eaf80eb52abff7dea55625
+size 10080854039

final/run.log

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:058fd4a79723e791ddc6fb6f8b325215c52492b6ad612095af6f6c4560921c04
+size 28066786

logs/train_metrics.log

@@ -0,0 +1,715 @@
+[4750/46793] loss 2.939 mtp 3.227 aux 1.42 gnorm 0.29 lr 2.9e-04 18.12s/step
+[4950/46793] loss 2.104 mtp 2.344 aux 0.98 gnorm 0.25 lr 2.9e-04 1.90s/step
+[val 4999] global=3.1314  code=2.036  fineweb_edu=4.010  glm_magpie=2.384  glm_regen=3.577  reap_calib=3.238  spdx=2.933  tinystories=1.513  wiki_en=3.775  wiki_zh=4.716
+[5000/46793] loss 1.935 mtp 2.134 aux 1.00 gnorm 0.31 lr 2.9e-04 1.96s/step
+[5050/46793] loss 3.337 mtp 3.646 aux 1.09 gnorm 0.28 lr 2.9e-04 1.92s/step
+[5100/46793] loss 1.789 mtp 2.050 aux 1.10 gnorm 0.31 lr 2.9e-04 1.89s/step
+[5150/46793] loss 1.857 mtp 2.170 aux 1.43 gnorm 0.27 lr 2.9e-04 1.87s/step
+[5200/46793] loss 3.039 mtp 3.490 aux 1.47 gnorm 0.27 lr 2.9e-04 1.85s/step
+[5250/46793] loss 2.921 mtp 3.272 aux 1.38 gnorm 0.26 lr 2.9e-04 1.88s/step
+[5300/46793] loss 2.172 mtp 2.445 aux 1.13 gnorm 0.28 lr 2.9e-04 1.87s/step
+[5350/46793] loss 3.104 mtp 3.456 aux 1.41 gnorm 0.29 lr 2.9e-04 1.86s/step
+[5500/46793] loss 3.069 mtp 3.392 aux 1.08 gnorm 0.26 lr 2.9e-04 1.89s/step
+[5550/46793] loss 1.894 mtp 2.179 aux 1.36 gnorm 0.28 lr 2.9e-04 1.88s/step
+[5600/46793] loss 2.326 mtp 2.538 aux 1.16 gnorm 0.28 lr 2.9e-04 1.87s/step
+[5650/46793] loss 2.081 mtp 2.448 aux 1.33 gnorm 0.27 lr 2.9e-04 1.87s/step
+[5700/46793] loss 2.696 mtp 3.108 aux 1.61 gnorm 0.28 lr 2.9e-04 1.86s/step
+[5750/46793] loss 2.564 mtp 2.818 aux 0.88 gnorm 0.24 lr 2.9e-04 1.88s/step
+[5800/46793] loss 1.617 mtp 1.830 aux 1.02 gnorm 0.25 lr 2.9e-04 1.87s/step
+[5850/46793] loss 1.654 mtp 1.915 aux 1.21 gnorm 0.26 lr 2.9e-04 1.86s/step
+[5900/46793] loss 2.326 mtp 2.638 aux 1.27 gnorm 0.26 lr 2.9e-04 1.86s/step
+[5950/46793] loss 2.594 mtp 2.958 aux 1.36 gnorm 0.28 lr 2.9e-04 1.85s/step
+[6100/46793] loss 2.319 mtp 2.634 aux 0.94 gnorm 0.28 lr 2.9e-04 1.87s/step
+[6150/46793] loss 3.222 mtp 3.515 aux 1.05 gnorm 0.26 lr 2.9e-04 1.87s/step
+[6200/46793] loss 2.167 mtp 2.412 aux 0.94 gnorm 0.25 lr 2.9e-04 1.87s/step
+[6250/46793] loss 2.682 mtp 2.986 aux 1.06 gnorm 0.26 lr 2.9e-04 1.88s/step
+[6300/46793] loss 3.297 mtp 3.786 aux 1.36 gnorm 0.28 lr 2.9e-04 1.87s/step
+[6350/46793] loss 3.009 mtp 3.348 aux 1.08 gnorm 0.26 lr 2.9e-04 1.87s/step
+[6400/46793] loss 2.655 mtp 2.955 aux 0.90 gnorm 0.29 lr 2.9e-04 1.86s/step
+[6450/46793] loss 3.273 mtp 3.581 aux 0.91 gnorm 0.27 lr 2.9e-04 1.86s/step
+[6500/46793] loss 3.390 mtp 3.708 aux 0.96 gnorm 0.25 lr 2.9e-04 1.87s/step
+[6550/46793] loss 2.326 mtp 2.607 aux 1.00 gnorm 0.24 lr 2.9e-04 1.87s/step
+[6700/46793] loss 2.539 mtp 2.782 aux 0.81 gnorm 5.31 lr 2.9e-04 1.87s/step
+[6750/46793] loss 2.122 mtp 2.302 aux 0.83 gnorm 0.25 lr 2.8e-04 1.88s/step
+[6800/46793] loss 2.007 mtp 2.289 aux 1.07 gnorm 0.26 lr 2.8e-04 1.88s/step
+[6850/46793] loss 2.055 mtp 2.303 aux 0.92 gnorm 0.26 lr 2.8e-04 1.87s/step
+[6900/46793] loss 2.823 mtp 3.154 aux 0.94 gnorm 0.27 lr 2.8e-04 1.87s/step
+[6950/46793] loss 3.147 mtp 3.429 aux 0.96 gnorm 0.26 lr 2.8e-04 1.87s/step
+[val 6999] global=3.0086  code=1.841  fineweb_edu=3.868  glm_magpie=2.263  glm_regen=3.467  reap_calib=3.113  spdx=2.997  tinystories=1.408  wiki_en=3.626  wiki_zh=4.494
+[7000/46793] loss 2.592 mtp 2.842 aux 0.79 gnorm 0.29 lr 2.8e-04 1.87s/step
+[7050/46793] loss 3.146 mtp 3.533 aux 1.12 gnorm 0.27 lr 2.8e-04 1.87s/step
+[7100/46793] loss 2.418 mtp 2.799 aux 1.20 gnorm 0.26 lr 2.8e-04 1.87s/step
+[7150/46793] loss 2.169 mtp 2.418 aux 0.81 gnorm 0.23 lr 2.8e-04 1.87s/step
+[7400/46793] loss 2.616 mtp 3.045 aux 1.19 gnorm 0.24 lr 2.8e-04 1.87s/step
+[7450/46793] loss 2.327 mtp 2.736 aux 1.02 gnorm 0.33 lr 2.8e-04 1.87s/step
+[7500/46793] loss 2.014 mtp 2.252 aux 0.82 gnorm 0.26 lr 2.8e-04 1.88s/step
+[7550/46793] loss 2.197 mtp 2.467 aux 0.85 gnorm 0.24 lr 2.8e-04 1.87s/step
+[7600/46793] loss 1.812 mtp 2.076 aux 0.96 gnorm 0.27 lr 2.8e-04 1.87s/step
+[7650/46793] loss 1.505 mtp 1.783 aux 1.06 gnorm 0.26 lr 2.8e-04 1.87s/step
+[7700/46793] loss 2.808 mtp 3.127 aux 0.97 gnorm 0.26 lr 2.8e-04 1.87s/step
+[7750/46793] loss 1.700 mtp 2.027 aux 0.99 gnorm 0.29 lr 2.8e-04 1.87s/step
+[7900/46793] loss 2.420 mtp 2.845 aux 1.45 gnorm 0.28 lr 2.8e-04 1.88s/step
+[7950/46793] loss 2.271 mtp 2.557 aux 0.83 gnorm 0.29 lr 2.8e-04 1.87s/step
+[val 7999] global=2.9493  code=1.705  fineweb_edu=3.812  glm_magpie=2.213  glm_regen=3.395  reap_calib=3.115  spdx=2.960  tinystories=1.365  wiki_en=3.576  wiki_zh=4.404
+[8000/46793] loss 1.824 mtp 2.059 aux 0.77 gnorm 0.25 lr 2.8e-04 1.88s/step
+[8050/46793] loss 2.590 mtp 2.887 aux 0.88 gnorm 0.28 lr 2.8e-04 1.88s/step
+[8100/46793] loss 1.777 mtp 2.007 aux 0.72 gnorm 0.26 lr 2.8e-04 1.88s/step
+[8150/46793] loss 2.523 mtp 2.877 aux 1.15 gnorm 0.90 lr 2.8e-04 1.87s/step
+[8200/46793] loss 1.604 mtp 1.914 aux 1.00 gnorm 0.27 lr 2.8e-04 1.87s/step
+[8250/46793] loss 1.789 mtp 2.115 aux 1.11 gnorm 0.22 lr 2.8e-04 1.88s/step
+[8300/46793] loss 2.672 mtp 3.038 aux 1.05 gnorm 0.27 lr 2.8e-04 1.88s/step
+[8350/46793] loss 2.375 mtp 2.619 aux 1.05 gnorm 0.28 lr 2.8e-04 1.87s/step
+[8500/46793] loss 1.750 mtp 1.987 aux 0.91 gnorm 0.25 lr 2.8e-04 1.88s/step
+[8550/46793] loss 2.370 mtp 2.762 aux 1.30 gnorm 0.29 lr 2.8e-04 1.88s/step
+[8600/46793] loss 2.573 mtp 3.041 aux 1.49 gnorm 0.27 lr 2.8e-04 1.88s/step
+[8650/46793] loss 2.625 mtp 2.926 aux 0.89 gnorm 0.27 lr 2.8e-04 1.88s/step
+[8700/46793] loss 1.924 mtp 2.241 aux 0.92 gnorm 0.26 lr 2.8e-04 1.88s/step
+[8750/46793] loss 2.755 mtp 3.036 aux 0.95 gnorm 0.26 lr 2.7e-04 1.88s/step
+[8800/46793] loss 2.592 mtp 2.859 aux 0.80 gnorm 0.28 lr 2.7e-04 1.88s/step
+[8850/46793] loss 2.435 mtp 2.848 aux 1.05 gnorm 0.27 lr 2.7e-04 1.88s/step
+[8900/46793] loss 2.595 mtp 2.955 aux 1.06 gnorm 0.28 lr 2.7e-04 1.87s/step
+[8950/46793] loss 2.114 mtp 2.594 aux 1.40 gnorm 0.27 lr 2.7e-04 1.87s/step
+[9100/46793] loss 2.119 mtp 2.364 aux 0.61 gnorm 0.26 lr 2.7e-04 1.88s/step
+[9150/46793] loss 2.307 mtp 2.547 aux 0.74 gnorm 0.24 lr 2.7e-04 1.88s/step
+[9200/46793] loss 2.446 mtp 2.719 aux 0.84 gnorm 0.25 lr 2.7e-04 1.88s/step
+[9250/46793] loss 2.897 mtp 3.208 aux 0.85 gnorm 0.28 lr 2.7e-04 1.88s/step
+[9300/46793] loss 2.016 mtp 2.261 aux 0.71 gnorm 0.24 lr 2.7e-04 1.88s/step
+[9350/46793] loss 2.295 mtp 2.600 aux 0.84 gnorm 0.29 lr 2.7e-04 1.88s/step
+[9400/46793] loss 1.496 mtp 1.964 aux 1.48 gnorm 0.26 lr 2.7e-04 1.88s/step
+[9450/46793] loss 2.358 mtp 2.664 aux 0.80 gnorm 0.27 lr 2.7e-04 1.88s/step
+[9500/46793] loss 2.619 mtp 2.923 aux 0.83 gnorm 0.27 lr 2.7e-04 1.88s/step
+[9550/46793] loss 2.198 mtp 2.565 aux 1.07 gnorm 0.33 lr 2.7e-04 1.88s/step
+[9700/46793] loss 2.574 mtp 2.925 aux 1.05 gnorm 0.26 lr 2.7e-04 1.88s/step
+[9750/46793] loss 2.602 mtp 2.887 aux 0.73 gnorm 0.25 lr 2.7e-04 1.88s/step
+[9800/46793] loss 3.472 mtp 3.731 aux 0.68 gnorm 0.26 lr 2.7e-04 1.88s/step
+[9850/46793] loss 1.678 mtp 2.019 aux 1.01 gnorm 0.25 lr 2.7e-04 1.88s/step
+[9900/46793] loss 1.854 mtp 2.344 aux 1.19 gnorm 0.27 lr 2.7e-04 1.88s/step
+[9950/46793] loss 2.167 mtp 2.382 aux 0.57 gnorm 0.24 lr 2.7e-04 1.88s/step
+[val 9999] global=2.8791  code=1.519  fineweb_edu=3.743  glm_magpie=2.149  glm_regen=3.301  reap_calib=3.140  spdx=3.005  tinystories=1.301  wiki_en=3.490  wiki_zh=4.264
+[10000/46793] loss 2.481 mtp 2.701 aux 0.56 gnorm 0.23 lr 2.7e-04 1.88s/step
+[10050/46793] loss 1.568 mtp 1.771 aux 0.56 gnorm 0.29 lr 2.7e-04 1.88s/step
+[10100/46793] loss 2.387 mtp 2.733 aux 0.86 gnorm 0.27 lr 2.7e-04 1.88s/step
+[10150/46793] loss 1.980 mtp 2.381 aux 1.14 gnorm 0.25 lr 2.7e-04 1.88s/step
+[10300/46793] loss 2.935 mtp 3.235 aux 0.67 gnorm 0.26 lr 2.7e-04 1.89s/step
+[10350/46793] loss 1.535 mtp 1.737 aux 0.55 gnorm 0.26 lr 2.7e-04 1.89s/step
+[10400/46793] loss 1.837 mtp 2.187 aux 0.95 gnorm 0.27 lr 2.6e-04 1.89s/step
+[10450/46793] loss 2.610 mtp 2.884 aux 0.60 gnorm 0.24 lr 2.6e-04 1.89s/step
+[10500/46793] loss 3.026 mtp 3.410 aux 0.87 gnorm 0.26 lr 2.6e-04 1.89s/step
+[10550/46793] loss 2.986 mtp 3.291 aux 0.70 gnorm 0.25 lr 2.6e-04 1.89s/step
+[10600/46793] loss 2.434 mtp 2.721 aux 0.66 gnorm 0.24 lr 2.6e-04 1.89s/step
+[10650/46793] loss 2.606 mtp 3.037 aux 1.20 gnorm 0.24 lr 2.6e-04 1.89s/step
+[10700/46793] loss 2.374 mtp 2.677 aux 0.71 gnorm 0.28 lr 2.6e-04 1.89s/step
+[10750/46793] loss 1.795 mtp 2.008 aux 0.60 gnorm 0.33 lr 2.6e-04 1.89s/step
+[10900/46793] loss 3.017 mtp 3.347 aux 0.69 gnorm 0.26 lr 2.6e-04 1.89s/step
+[10950/46793] loss 2.558 mtp 2.951 aux 0.94 gnorm 0.25 lr 2.6e-04 1.89s/step
+[val 10999] global=2.8581  code=1.557  fineweb_edu=3.722  glm_magpie=2.126  glm_regen=3.259  reap_calib=3.126  spdx=2.965  tinystories=1.292  wiki_en=3.446  wiki_zh=4.230
+[11000/46793] loss 2.534 mtp 2.823 aux 0.61 gnorm 0.26 lr 2.6e-04 1.90s/step
+[11050/46793] loss 1.887 mtp 2.065 aux 0.45 gnorm 0.25 lr 2.6e-04 1.90s/step
+[11100/46793] loss 2.989 mtp 3.330 aux 0.64 gnorm 0.25 lr 2.6e-04 1.90s/step
+[11150/46793] loss 1.999 mtp 2.279 aux 0.54 gnorm 0.26 lr 2.6e-04 1.90s/step
+[11200/46793] loss 1.847 mtp 2.152 aux 0.83 gnorm 0.26 lr 2.6e-04 1.90s/step
+[11250/46793] loss 1.668 mtp 1.863 aux 0.54 gnorm 0.25 lr 2.6e-04 1.90s/step
+[11300/46793] loss 2.061 mtp 2.424 aux 0.80 gnorm 0.26 lr 2.6e-04 1.90s/step
+[11350/46793] loss 1.800 mtp 2.110 aux 0.84 gnorm 0.28 lr 2.6e-04 1.90s/step
+[11500/46793] loss 2.777 mtp 3.124 aux 0.88 gnorm 0.25 lr 2.6e-04 1.90s/step
+[11550/46793] loss 2.740 mtp 2.988 aux 0.56 gnorm 0.23 lr 2.6e-04 1.90s/step
+[11600/46793] loss 2.993 mtp 3.415 aux 0.99 gnorm 0.26 lr 2.6e-04 1.90s/step
+[11650/46793] loss 2.529 mtp 2.994 aux 1.18 gnorm 0.25 lr 2.6e-04 1.90s/step
+[11700/46793] loss 1.817 mtp 2.293 aux 1.09 gnorm 0.28 lr 2.6e-04 1.90s/step
+[11750/46793] loss 2.611 mtp 2.883 aux 0.57 gnorm 0.24 lr 2.6e-04 1.91s/step
+[11800/46793] loss 2.135 mtp 2.370 aux 0.49 gnorm 0.27 lr 2.6e-04 1.90s/step
+[11850/46793] loss 2.536 mtp 2.804 aux 0.58 gnorm 0.25 lr 2.5e-04 1.90s/step
+[11900/46793] loss 2.212 mtp 2.454 aux 0.58 gnorm 0.24 lr 2.5e-04 1.90s/step
+[11950/46793] loss 2.890 mtp 3.320 aux 0.87 gnorm 0.26 lr 2.5e-04 1.90s/step
+[12100/46793] loss 1.965 mtp 2.171 aux 0.50 gnorm 0.24 lr 2.5e-04 1.91s/step
+[12150/46793] loss 1.733 mtp 1.944 aux 0.48 gnorm 0.33 lr 2.5e-04 1.91s/step
+[12200/46793] loss 2.630 mtp 2.921 aux 0.56 gnorm 0.25 lr 2.5e-04 1.91s/step
+[12250/46793] loss 2.037 mtp 2.236 aux 0.47 gnorm 0.28 lr 2.5e-04 1.91s/step
+[12300/46793] loss 1.782 mtp 2.310 aux 1.21 gnorm 0.27 lr 2.5e-04 1.91s/step
+[12350/46793] loss 2.067 mtp 2.539 aux 1.04 gnorm 0.29 lr 2.5e-04 1.91s/step
+[12400/46793] loss 2.358 mtp 2.746 aux 0.79 gnorm 0.25 lr 2.5e-04 1.91s/step
+[12450/46793] loss 1.731 mtp 1.944 aux 0.48 gnorm 0.33 lr 2.5e-04 1.91s/step
+[12500/46793] loss 1.877 mtp 2.185 aux 0.64 gnorm 0.25 lr 2.5e-04 1.91s/step
+[12550/46793] loss 2.822 mtp 3.153 aux 0.58 gnorm 0.39 lr 2.5e-04 1.91s/step
+[12700/46793] loss 2.691 mtp 2.947 aux 0.56 gnorm 0.27 lr 2.5e-04 1.91s/step
+[12750/46793] loss 2.430 mtp 2.740 aux 0.60 gnorm 0.25 lr 2.5e-04 1.92s/step
+[12800/46793] loss 2.583 mtp 3.061 aux 0.91 gnorm 0.23 lr 2.5e-04 1.92s/step
+[12850/46793] loss 2.598 mtp 2.908 aux 0.55 gnorm 0.26 lr 2.5e-04 1.92s/step
+[12900/46793] loss 2.308 mtp 2.835 aux 1.10 gnorm 0.27 lr 2.5e-04 1.92s/step
+[12950/46793] loss 1.875 mtp 2.095 aux 0.56 gnorm 0.38 lr 2.5e-04 1.92s/step
+[val 12999] global=2.7944  code=1.437  fineweb_edu=3.657  glm_magpie=2.066  glm_regen=3.201  reap_calib=3.157  spdx=2.868  tinystories=1.251  wiki_en=3.382  wiki_zh=4.131
+[13000/46793] loss 2.041 mtp 2.399 aux 0.85 gnorm 0.25 lr 2.5e-04 1.92s/step
+[13050/46793] loss 2.762 mtp 3.169 aux 0.87 gnorm 0.29 lr 2.5e-04 1.92s/step
+[13100/46793] loss 2.624 mtp 3.009 aux 0.93 gnorm 0.25 lr 2.5e-04 1.92s/step
+[13150/46793] loss 2.350 mtp 2.666 aux 0.58 gnorm 0.26 lr 2.5e-04 1.92s/step
+[13300/46793] loss 1.960 mtp 2.251 aux 0.63 gnorm 0.25 lr 2.4e-04 1.92s/step
+[13350/46793] loss 2.067 mtp 2.321 aux 0.52 gnorm 0.26 lr 2.4e-04 1.92s/step
+[13400/46793] loss 2.517 mtp 2.823 aux 0.54 gnorm 0.26 lr 2.4e-04 1.92s/step
+[13450/46793] loss 1.347 mtp 1.539 aux 0.45 gnorm 0.23 lr 2.4e-04 1.92s/step
+[13500/46793] loss 2.446 mtp 2.728 aux 0.52 gnorm 0.26 lr 2.4e-04 1.93s/step
+[13550/46793] loss 2.494 mtp 2.861 aux 0.70 gnorm 0.26 lr 2.4e-04 1.92s/step
+[13600/46793] loss 2.368 mtp 2.669 aux 0.57 gnorm 0.25 lr 2.4e-04 1.92s/step
+[13650/46793] loss 1.831 mtp 2.172 aux 0.71 gnorm 0.23 lr 2.4e-04 1.92s/step
+[13700/46793] loss 2.508 mtp 2.772 aux 0.64 gnorm 0.26 lr 2.4e-04 1.92s/step
+[13750/46793] loss 2.134 mtp 2.547 aux 0.71 gnorm 0.25 lr 2.4e-04 1.93s/step
+[13950/46793] loss 2.175 mtp 2.590 aux 0.92 gnorm 0.26 lr 2.4e-04 1.93s/step
+[val 13999] global=2.7977  code=1.471  fineweb_edu=3.644  glm_magpie=2.057  glm_regen=3.175  reap_calib=3.240  spdx=2.899  tinystories=1.224  wiki_en=3.370  wiki_zh=4.100
+[14000/46793] loss 3.483 mtp 2.432 aux 2.66 gnorm 2.94 lr 2.4e-04 1.93s/step
+[14050/46793] loss 2.758 mtp 3.016 aux 0.50 gnorm 0.28 lr 2.4e-04 1.93s/step
+[14100/46793] loss 2.678 mtp 2.956 aux 0.49 gnorm 0.27 lr 2.4e-04 1.93s/step
+[14150/46793] loss 2.439 mtp 2.721 aux 0.52 gnorm 0.25 lr 2.4e-04 1.93s/step
+[14200/46793] loss 1.747 mtp 2.044 aux 0.56 gnorm 0.26 lr 2.4e-04 1.93s/step
+[14250/46793] loss 1.890 mtp 2.342 aux 0.90 gnorm 0.25 lr 2.4e-04 1.93s/step
+[14300/46793] loss 2.564 mtp 2.820 aux 0.47 gnorm 0.27 lr 2.4e-04 1.93s/step
+[14350/46793] loss 2.806 mtp 3.083 aux 0.51 gnorm 0.28 lr 2.4e-04 1.93s/step
+[14500/46793] loss 1.756 mtp 2.177 aux 0.89 gnorm 0.23 lr 2.3e-04 1.94s/step
+[14550/46793] loss 2.186 mtp 2.399 aux 0.46 gnorm 0.26 lr 2.3e-04 1.94s/step
+[14600/46793] loss 3.076 mtp 3.393 aux 0.65 gnorm 0.27 lr 2.3e-04 1.94s/step
+[14650/46793] loss 1.612 mtp 1.845 aux 0.51 gnorm 0.24 lr 2.3e-04 1.94s/step
+[14700/46793] loss 3.067 mtp 3.389 aux 0.56 gnorm 0.24 lr 2.3e-04 1.94s/step
+[14750/46793] loss 1.400 mtp 1.556 aux 0.40 gnorm 0.28 lr 2.3e-04 1.94s/step
+[14800/46793] loss 1.743 mtp 1.941 aux 0.46 gnorm 0.24 lr 2.3e-04 1.94s/step
+[14850/46793] loss 3.006 mtp 3.334 aux 0.65 gnorm 0.24 lr 2.3e-04 1.94s/step
+[14900/46793] loss 1.146 mtp 1.367 aux 0.63 gnorm 0.23 lr 2.3e-04 1.94s/step
+[14950/46793] loss 2.131 mtp 2.562 aux 0.69 gnorm 0.26 lr 2.3e-04 1.94s/step
+[15100/46793] loss 2.183 mtp 2.533 aux 0.76 gnorm 0.24 lr 2.3e-04 1.94s/step
+[15150/46793] loss 1.771 mtp 1.989 aux 0.36 gnorm 0.25 lr 2.3e-04 1.94s/step
+[15200/46793] loss 2.757 mtp 3.032 aux 0.48 gnorm 0.25 lr 2.3e-04 1.94s/step
+[15250/46793] loss 2.009 mtp 2.368 aux 0.72 gnorm 0.24 lr 2.3e-04 1.94s/step
+[15300/46793] loss 1.948 mtp 2.460 aux 0.88 gnorm 0.27 lr 2.3e-04 1.94s/step
+[15350/46793] loss 1.406 mtp 1.602 aux 0.43 gnorm 0.23 lr 2.3e-04 1.94s/step
+[15400/46793] loss 2.329 mtp 2.725 aux 0.71 gnorm 0.25 lr 2.3e-04 1.94s/step
+[15450/46793] loss 2.284 mtp 2.550 aux 0.46 gnorm 0.26 lr 2.3e-04 1.94s/step
+[15500/46793] loss 3.761 mtp 4.049 aux 0.55 gnorm 0.26 lr 2.3e-04 1.94s/step
+[15550/46793] loss 3.054 mtp 3.350 aux 0.54 gnorm 0.28 lr 2.3e-04 1.94s/step
+[15700/46793] loss 2.139 mtp 2.462 aux 0.73 gnorm 0.23 lr 2.2e-04 1.95s/step
+[15750/46793] loss 1.691 mtp 1.891 aux 0.38 gnorm 0.23 lr 2.2e-04 1.95s/step
+[15800/46793] loss 3.004 mtp 3.305 aux 0.49 gnorm 0.28 lr 2.2e-04 1.95s/step
+[15850/46793] loss 1.390 mtp 1.585 aux 0.40 gnorm 0.23 lr 2.2e-04 1.95s/step
+[15900/46793] loss 1.935 mtp 2.216 aux 0.46 gnorm 0.22 lr 2.2e-04 1.95s/step
+[15950/46793] loss 1.537 mtp 1.734 aux 0.39 gnorm 0.24 lr 2.2e-04 1.95s/step
+[val 15999] global=2.7661  code=1.369  fineweb_edu=3.599  glm_magpie=2.025  glm_regen=3.134  reap_calib=3.305  spdx=2.893  tinystories=1.190  wiki_en=3.324  wiki_zh=4.056
+[16000/46793] loss 2.188 mtp 2.485 aux 0.52 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16050/46793] loss 1.734 mtp 2.247 aux 0.93 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16100/46793] loss 2.679 mtp 2.894 aux 0.45 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16150/46793] loss 1.895 mtp 2.104 aux 0.46 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16300/46793] loss 2.355 mtp 2.656 aux 0.47 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16350/46793] loss 1.685 mtp 2.025 aux 0.66 gnorm 0.27 lr 2.2e-04 1.95s/step
+[16400/46793] loss 2.409 mtp 2.644 aux 0.36 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16450/46793] loss 1.988 mtp 2.474 aux 0.85 gnorm 0.26 lr 2.2e-04 1.95s/step
+[16500/46793] loss 2.627 mtp 2.907 aux 0.44 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16550/46793] loss 2.285 mtp 2.655 aux 0.67 gnorm 0.28 lr 2.2e-04 1.95s/step
+[16600/46793] loss 2.112 mtp 2.368 aux 0.44 gnorm 0.24 lr 2.2e-04 1.95s/step
+[16650/46793] loss 2.600 mtp 3.032 aux 0.75 gnorm 0.27 lr 2.2e-04 1.95s/step
+[16700/46793] loss 1.484 mtp 1.903 aux 0.68 gnorm 0.25 lr 2.2e-04 1.95s/step
+[16750/46793] loss 1.916 mtp 2.264 aux 0.56 gnorm 0.26 lr 2.1e-04 1.96s/step
+[16900/46793] loss 2.752 mtp 3.053 aux 0.46 gnorm 0.26 lr 2.1e-04 1.96s/step
+[16950/46793] loss 1.976 mtp 2.189 aux 0.31 gnorm 0.25 lr 2.1e-04 1.96s/step
+[val 16999] global=2.7679  code=1.375  fineweb_edu=3.588  glm_magpie=2.001  glm_regen=3.118  reap_calib=3.415  spdx=2.953  tinystories=1.167  wiki_en=3.294  wiki_zh=4.001
+[17000/46793] loss 1.636 mtp 1.828 aux 0.33 gnorm 0.27 lr 2.1e-04 1.96s/step
+[17050/46793] loss 1.306 mtp 1.451 aux 0.30 gnorm 0.23 lr 2.1e-04 1.96s/step
+[17100/46793] loss 2.613 mtp 2.951 aux 0.46 gnorm 0.25 lr 2.1e-04 1.96s/step
+[17150/46793] loss 2.258 mtp 2.518 aux 0.47 gnorm 0.31 lr 2.1e-04 1.96s/step
+[17200/46793] loss 2.705 mtp 3.053 aux 0.50 gnorm 0.25 lr 2.1e-04 1.96s/step
+[17250/46793] loss 2.025 mtp 2.472 aux 0.66 gnorm 0.25 lr 2.1e-04 1.96s/step
+[17300/46793] loss 2.196 mtp 2.543 aux 0.60 gnorm 0.26 lr 2.1e-04 1.96s/step
+[17350/46793] loss 1.954 mtp 2.192 aux 0.34 gnorm 0.25 lr 2.1e-04 1.96s/step
+[17550/46793] loss 2.572 mtp 2.856 aux 0.41 gnorm 0.25 lr 2.1e-04 1.96s/step
+[17600/46793] loss 2.468 mtp 2.770 aux 0.43 gnorm 0.27 lr 2.1e-04 1.96s/step
+[17650/46793] loss 2.549 mtp 2.892 aux 0.65 gnorm 0.23 lr 2.1e-04 1.96s/step
+[17700/46793] loss 2.070 mtp 2.321 aux 0.36 gnorm 0.25 lr 2.1e-04 1.96s/step
+[17750/46793] loss 0.682 mtp 1.116 aux 0.76 gnorm 0.26 lr 2.1e-04 1.96s/step
+[17800/46793] loss 1.011 mtp 1.374 aux 0.81 gnorm 0.27 lr 2.1e-04 1.96s/step
+[17850/46793] loss 1.581 mtp 1.998 aux 0.69 gnorm 0.26 lr 2.0e-04 1.96s/step
+[17900/46793] loss 1.532 mtp 1.726 aux 0.29 gnorm 0.26 lr 2.0e-04 1.96s/step
+[17950/46793] loss 2.215 mtp 2.474 aux 0.45 gnorm 0.27 lr 2.0e-04 1.96s/step
+[18100/46793] loss 2.109 mtp 2.360 aux 0.42 gnorm 0.22 lr 2.0e-04 1.97s/step
+[18150/46793] loss 2.592 mtp 2.933 aux 0.45 gnorm 0.24 lr 2.0e-04 1.97s/step
+[18200/46793] loss 2.543 mtp 2.836 aux 0.51 gnorm 0.29 lr 2.0e-04 1.97s/step
+[18250/46793] loss 2.384 mtp 2.630 aux 0.36 gnorm 0.27 lr 2.0e-04 1.97s/step
+[18300/46793] loss 1.975 mtp 2.473 aux 0.90 gnorm 0.24 lr 2.0e-04 1.97s/step
+[18350/46793] loss 1.683 mtp 2.304 aux 1.04 gnorm 0.39 lr 2.0e-04 1.97s/step
+[18400/46793] loss 1.867 mtp 2.148 aux 0.54 gnorm 0.26 lr 2.0e-04 1.97s/step
+[18450/46793] loss 1.361 mtp 1.563 aux 0.37 gnorm 0.24 lr 2.0e-04 1.97s/step
+[18500/46793] loss 2.047 mtp 2.288 aux 0.38 gnorm 0.25 lr 2.0e-04 1.97s/step
+[18550/46793] loss 1.727 mtp 1.942 aux 0.35 gnorm 0.25 lr 2.0e-04 1.97s/step
+[18700/46793] loss 2.857 mtp 3.174 aux 0.44 gnorm 0.27 lr 2.0e-04 1.97s/step
+[18750/46793] loss 2.339 mtp 2.588 aux 0.35 gnorm 0.25 lr 2.0e-04 1.97s/step
+[18800/46793] loss 2.549 mtp 2.830 aux 0.37 gnorm 0.25 lr 2.0e-04 1.97s/step
+[18850/46793] loss 2.905 mtp 3.328 aux 0.51 gnorm 0.26 lr 2.0e-04 1.97s/step
+[18900/46793] loss 1.653 mtp 2.094 aux 0.81 gnorm 0.29 lr 1.9e-04 1.97s/step
+[18950/46793] loss 2.086 mtp 2.354 aux 0.34 gnorm 0.25 lr 1.9e-04 1.97s/step
+[val 18999] global=2.7310  code=1.337  fineweb_edu=3.555  glm_magpie=1.976  glm_regen=3.074  reap_calib=3.488  spdx=2.755  tinystories=1.153  wiki_en=3.275  wiki_zh=3.967
+[19000/46793] loss 1.852 mtp 2.070 aux 0.32 gnorm 0.25 lr 1.9e-04 1.97s/step
+[19050/46793] loss 2.382 mtp 2.681 aux 0.40 gnorm 0.25 lr 1.9e-04 1.97s/step
+[19100/46793] loss 2.188 mtp 2.530 aux 0.44 gnorm 0.27 lr 1.9e-04 1.97s/step
+[19150/46793] loss 1.129 mtp 1.301 aux 0.41 gnorm 0.23 lr 1.9e-04 1.97s/step
+[19300/46793] loss 1.443 mtp 1.622 aux 0.27 gnorm 0.24 lr 1.9e-04 1.98s/step
+[19350/46793] loss 2.320 mtp 2.566 aux 0.37 gnorm 0.24 lr 1.9e-04 1.98s/step
+[19400/46793] loss 2.292 mtp 2.560 aux 0.38 gnorm 0.24 lr 1.9e-04 1.98s/step
+[19450/46793] loss 1.773 mtp 2.068 aux 0.40 gnorm 0.28 lr 1.9e-04 1.98s/step
+[19500/46793] loss 2.160 mtp 2.474 aux 0.50 gnorm 0.24 lr 1.9e-04 1.98s/step
+[19550/46793] loss 2.387 mtp 2.679 aux 0.41 gnorm 2.62 lr 1.9e-04 1.98s/step
+[19600/46793] loss 2.624 mtp 2.904 aux 0.40 gnorm 0.23 lr 1.9e-04 1.98s/step
+[19650/46793] loss 2.554 mtp 2.861 aux 0.38 gnorm 0.24 lr 1.9e-04 1.98s/step
+[19700/46793] loss 2.408 mtp 2.687 aux 0.39 gnorm 0.23 lr 1.9e-04 1.98s/step
+[19750/46793] loss 1.021 mtp 1.300 aux 0.49 gnorm 0.22 lr 1.9e-04 1.98s/step
+[19900/46793] loss 2.030 mtp 2.313 aux 0.37 gnorm 0.26 lr 1.8e-04 1.98s/step
+[19950/46793] loss 1.610 mtp 1.921 aux 0.49 gnorm 0.23 lr 1.8e-04 1.98s/step
+[val 19999] global=2.7170  code=1.334  fineweb_edu=3.544  glm_magpie=1.968  glm_regen=3.053  reap_calib=3.472  spdx=2.762  tinystories=1.138  wiki_en=3.254  wiki_zh=3.929
+[20000/46793] loss 1.617 mtp 1.793 aux 0.25 gnorm 0.25 lr 1.8e-04 1.98s/step
+[20050/46793] loss 2.529 mtp 2.935 aux 0.55 gnorm 0.25 lr 1.8e-04 1.98s/step
+[20100/46793] loss 2.522 mtp 2.848 aux 0.45 gnorm 0.27 lr 1.8e-04 1.98s/step
+[20150/46793] loss 2.507 mtp 2.802 aux 0.37 gnorm 0.25 lr 1.8e-04 1.98s/step
+[20200/46793] loss 1.618 mtp 1.959 aux 0.48 gnorm 0.24 lr 1.8e-04 1.98s/step
+[20250/46793] loss 1.756 mtp 2.157 aux 0.75 gnorm 0.23 lr 1.8e-04 1.98s/step
+[20300/46793] loss 2.585 mtp 2.856 aux 0.33 gnorm 0.24 lr 1.8e-04 1.98s/step
+[20350/46793] loss 1.737 mtp 2.242 aux 0.78 gnorm 0.30 lr 1.8e-04 1.98s/step
+[20550/46793] loss 2.189 mtp 2.484 aux 0.50 gnorm 0.23 lr 1.8e-04 1.98s/step
+[20600/46793] loss 2.724 mtp 3.013 aux 0.35 gnorm 0.25 lr 1.8e-04 1.98s/step
+[20650/46793] loss 1.759 mtp 2.081 aux 0.49 gnorm 0.24 lr 1.8e-04 1.98s/step
+[20700/46793] loss 2.130 mtp 2.401 aux 0.31 gnorm 0.25 lr 1.8e-04 1.98s/step
+[20750/46793] loss 2.119 mtp 2.437 aux 0.45 gnorm 0.23 lr 1.8e-04 1.98s/step
+[20800/46793] loss 1.764 mtp 2.065 aux 0.57 gnorm 0.24 lr 1.8e-04 1.98s/step
+[20850/46793] loss 2.507 mtp 2.811 aux 0.39 gnorm 0.26 lr 1.8e-04 1.98s/step
+[20900/46793] loss 2.753 mtp 3.064 aux 0.36 gnorm 0.28 lr 1.8e-04 1.98s/step
+[20950/46793] loss 1.801 mtp 2.057 aux 0.34 gnorm 0.25 lr 1.7e-04 1.98s/step
+[21100/46793] loss 1.721 mtp 1.966 aux 0.32 gnorm 0.24 lr 1.7e-04 1.99s/step
+[21150/46793] loss 2.402 mtp 2.691 aux 0.35 gnorm 0.26 lr 1.7e-04 1.99s/step
+[21200/46793] loss 1.874 mtp 2.207 aux 0.49 gnorm 0.25 lr 1.7e-04 1.99s/step
+[21250/46793] loss 2.892 mtp 3.202 aux 0.38 gnorm 0.26 lr 1.7e-04 1.99s/step
+[21300/46793] loss 2.683 mtp 2.968 aux 0.36 gnorm 0.26 lr 1.7e-04 1.99s/step
+[21350/46793] loss 2.190 mtp 2.481 aux 0.37 gnorm 0.26 lr 1.7e-04 1.99s/step
+[21400/46793] loss 2.623 mtp 2.922 aux 0.36 gnorm 0.27 lr 1.7e-04 1.99s/step
+[21450/46793] loss 2.348 mtp 2.636 aux 0.36 gnorm 0.24 lr 1.7e-04 1.99s/step
+[21500/46793] loss 2.517 mtp 2.801 aux 0.34 gnorm 0.25 lr 1.7e-04 1.99s/step
+[21550/46793] loss 1.862 mtp 2.249 aux 0.55 gnorm 0.24 lr 1.7e-04 1.99s/step
+[21700/46793] loss 2.569 mtp 2.843 aux 0.33 gnorm 0.25 lr 1.7e-04 1.99s/step
+[21750/46793] loss 1.780 mtp 2.122 aux 0.50 gnorm 0.23 lr 1.7e-04 1.99s/step
+[21800/46793] loss 2.003 mtp 2.249 aux 0.31 gnorm 0.23 lr 1.7e-04 1.99s/step
+[21850/46793] loss 2.417 mtp 2.669 aux 0.33 gnorm 0.24 lr 1.7e-04 1.99s/step
+[21900/46793] loss 1.493 mtp 1.947 aux 0.72 gnorm 0.26 lr 1.7e-04 1.99s/step
+[21950/46793] loss 2.249 mtp 2.721 aux 0.67 gnorm 0.25 lr 1.6e-04 1.99s/step
+[val 21999] global=2.7240  code=1.314  fineweb_edu=3.529  glm_magpie=1.946  glm_regen=3.014  reap_calib=3.562  spdx=2.905  tinystories=1.118  wiki_en=3.225  wiki_zh=3.902
+[22000/46793] loss 1.742 mtp 1.989 aux 0.28 gnorm 0.26 lr 1.6e-04 1.99s/step
+[22050/46793] loss 1.356 mtp 1.544 aux 0.25 gnorm 0.24 lr 1.6e-04 1.99s/step
+[22100/46793] loss 2.570 mtp 2.863 aux 0.33 gnorm 0.24 lr 1.6e-04 1.99s/step
+[22150/46793] loss 1.524 mtp 1.739 aux 0.27 gnorm 0.25 lr 1.6e-04 1.99s/step
+[22300/46793] loss 2.105 mtp 2.368 aux 0.28 gnorm 0.42 lr 1.6e-04 1.99s/step
+[22350/46793] loss 2.280 mtp 2.582 aux 0.32 gnorm 0.27 lr 1.6e-04 1.99s/step
+[22400/46793] loss 1.852 mtp 2.111 aux 0.44 gnorm 0.30 lr 1.6e-04 1.99s/step
+[22450/46793] loss 1.140 mtp 1.297 aux 0.21 gnorm 0.23 lr 1.6e-04 1.99s/step
+[22500/46793] loss 1.256 mtp 1.432 aux 0.26 gnorm 0.23 lr 1.6e-04 2.00s/step
+[22550/46793] loss 2.481 mtp 2.868 aux 0.47 gnorm 0.26 lr 1.6e-04 2.00s/step
+[22600/46793] loss 2.670 mtp 2.954 aux 0.32 gnorm 0.25 lr 1.6e-04 2.00s/step
+[22650/46793] loss 2.557 mtp 2.876 aux 0.33 gnorm 0.25 lr 1.6e-04 2.00s/step
+[22700/46793] loss 2.334 mtp 2.577 aux 0.25 gnorm 0.24 lr 1.6e-04 2.00s/step
+[22750/46793] loss 2.064 mtp 2.288 aux 0.25 gnorm 0.23 lr 1.6e-04 2.00s/step
+[22950/46793] loss 2.777 mtp 3.087 aux 0.33 gnorm 0.24 lr 1.5e-04 2.00s/step
+[val 22999] global=2.6997  code=1.290  fineweb_edu=3.502  glm_magpie=1.934  glm_regen=3.009  reap_calib=3.570  spdx=2.773  tinystories=1.119  wiki_en=3.210  wiki_zh=3.890
+[23000/46793] loss 1.504 mtp 1.943 aux 0.47 gnorm 0.27 lr 1.5e-04 2.00s/step
+[23050/46793] loss 1.290 mtp 1.616 aux 0.42 gnorm 0.24 lr 1.5e-04 2.00s/step
+[23100/46793] loss 1.811 mtp 2.020 aux 0.27 gnorm 0.23 lr 1.5e-04 2.00s/step
+[23150/46793] loss 1.013 mtp 1.342 aux 0.57 gnorm 0.23 lr 1.5e-04 2.00s/step
+[23200/46793] loss 2.088 mtp 2.464 aux 0.48 gnorm 0.25 lr 1.5e-04 2.00s/step
+[23250/46793] loss 0.970 mtp 1.122 aux 0.19 gnorm 0.23 lr 1.5e-04 2.00s/step
+[23300/46793] loss 2.196 mtp 2.543 aux 0.43 gnorm 0.28 lr 1.5e-04 2.00s/step
+[23350/46793] loss 2.197 mtp 2.434 aux 0.26 gnorm 0.24 lr 1.5e-04 2.00s/step
+[23500/46793] loss 2.122 mtp 2.609 aux 0.57 gnorm 0.24 lr 1.5e-04 2.00s/step
+[23550/46793] loss 1.573 mtp 1.769 aux 0.22 gnorm 0.25 lr 1.5e-04 2.00s/step
+[23600/46793] loss 2.069 mtp 2.428 aux 0.41 gnorm 0.24 lr 1.5e-04 2.00s/step
+[23650/46793] loss 1.623 mtp 1.850 aux 0.29 gnorm 0.27 lr 1.5e-04 2.00s/step
+[23700/46793] loss 2.092 mtp 2.366 aux 0.30 gnorm 0.32 lr 1.5e-04 2.00s/step
+[23750/46793] loss 2.202 mtp 2.526 aux 0.40 gnorm 0.25 lr 1.5e-04 2.00s/step
+[23800/46793] loss 1.279 mtp 1.554 aux 0.36 gnorm 0.24 lr 1.5e-04 2.00s/step
+[23850/46793] loss 2.199 mtp 2.435 aux 0.23 gnorm 0.23 lr 1.5e-04 2.00s/step
+[23900/46793] loss 1.339 mtp 1.848 aux 0.67 gnorm 0.25 lr 1.4e-04 2.00s/step
+[23950/46793] loss 2.351 mtp 2.674 aux 0.30 gnorm 0.24 lr 1.4e-04 2.00s/step
+[24100/46793] loss 1.491 mtp 1.716 aux 0.35 gnorm 0.23 lr 1.4e-04 2.00s/step
+[24150/46793] loss 2.036 mtp 2.328 aux 0.30 gnorm 0.25 lr 1.4e-04 2.00s/step
+[24200/46793] loss 2.258 mtp 2.545 aux 0.27 gnorm 0.26 lr 1.4e-04 2.00s/step
+[24250/46793] loss 2.347 mtp 2.590 aux 0.24 gnorm 0.21 lr 1.4e-04 2.00s/step
+[24300/46793] loss 1.542 mtp 1.773 aux 0.37 gnorm 0.22 lr 1.4e-04 2.00s/step
+[24350/46793] loss 1.592 mtp 1.814 aux 0.26 gnorm 0.25 lr 1.4e-04 2.00s/step
+[24400/46793] loss 1.940 mtp 2.211 aux 0.29 gnorm 0.25 lr 1.4e-04 2.00s/step
+[24450/46793] loss 1.181 mtp 1.340 aux 0.18 gnorm 0.23 lr 1.4e-04 2.00s/step
+[24500/46793] loss 2.091 mtp 2.329 aux 0.26 gnorm 0.25 lr 1.4e-04 2.01s/step
+[24550/46793] loss 2.092 mtp 2.419 aux 0.37 gnorm 0.24 lr 1.4e-04 2.01s/step
+[24850/46793] loss 2.513 mtp 2.840 aux 0.29 gnorm 0.26 lr 1.4e-04 2.01s/step
+[24900/46793] loss 2.396 mtp 2.752 aux 0.34 gnorm 0.27 lr 1.3e-04 2.01s/step
+[24950/46793] loss 2.128 mtp 2.410 aux 0.24 gnorm 0.24 lr 1.3e-04 2.01s/step
+[val 24999] global=2.6951  code=1.256  fineweb_edu=3.489  glm_magpie=1.921  glm_regen=2.977  reap_calib=3.579  spdx=2.927  tinystories=1.056  wiki_en=3.191  wiki_zh=3.860
+[25000/46793] loss 2.019 mtp 2.264 aux 0.25 gnorm 0.28 lr 1.3e-04 2.01s/step
+[25050/46793] loss 1.906 mtp 2.168 aux 0.26 gnorm 0.23 lr 1.3e-04 2.01s/step
+[25100/46793] loss 2.856 mtp 3.201 aux 0.29 gnorm 0.24 lr 1.3e-04 2.01s/step
+[25150/46793] loss 2.036 mtp 2.359 aux 0.40 gnorm 0.25 lr 1.3e-04 2.01s/step
+[25300/46793] loss 2.451 mtp 2.795 aux 0.29 gnorm 0.30 lr 1.3e-04 2.01s/step
+[25350/46793] loss 1.765 mtp 2.007 aux 0.27 gnorm 0.25 lr 1.3e-04 2.01s/step
+[25400/46793] loss 1.820 mtp 2.079 aux 0.38 gnorm 0.31 lr 1.3e-04 2.01s/step
+[25450/46793] loss 1.751 mtp 2.036 aux 0.36 gnorm 0.25 lr 1.3e-04 2.01s/step
+[25500/46793] loss 1.790 mtp 2.242 aux 0.45 gnorm 0.27 lr 1.3e-04 2.01s/step
+[25550/46793] loss 1.960 mtp 2.235 aux 0.24 gnorm 0.25 lr 1.3e-04 2.01s/step
+[25600/46793] loss 1.226 mtp 1.397 aux 0.20 gnorm 0.25 lr 1.3e-04 2.01s/step
+[25650/46793] loss 2.032 mtp 2.333 aux 0.29 gnorm 0.24 lr 1.3e-04 2.01s/step
+[25700/46793] loss 0.790 mtp 1.294 aux 0.83 gnorm 0.24 lr 1.3e-04 2.01s/step
+[25750/46793] loss 1.681 mtp 1.974 aux 0.34 gnorm 0.23 lr 1.3e-04 2.01s/step
+[25900/46793] loss 2.055 mtp 2.353 aux 0.27 gnorm 0.25 lr 1.2e-04 2.01s/step
+[25950/46793] loss 2.393 mtp 2.667 aux 0.25 gnorm 0.23 lr 1.2e-04 2.01s/step
+[val 25999] global=2.6938  code=1.259  fineweb_edu=3.483  glm_magpie=1.915  glm_regen=2.970  reap_calib=3.630  spdx=2.890  tinystories=1.059  wiki_en=3.183  wiki_zh=3.856
+[26000/46793] loss 1.537 mtp 1.754 aux 0.20 gnorm 0.22 lr 1.2e-04 2.01s/step
+[26050/46793] loss 2.315 mtp 2.586 aux 0.23 gnorm 0.25 lr 1.2e-04 2.01s/step
+[26100/46793] loss 1.195 mtp 1.440 aux 0.31 gnorm 0.22 lr 1.2e-04 2.01s/step
+[26150/46793] loss 1.670 mtp 1.942 aux 0.25 gnorm 0.24 lr 1.2e-04 2.01s/step
+[26200/46793] loss 2.462 mtp 2.717 aux 0.24 gnorm 0.25 lr 1.2e-04 2.01s/step
+[26250/46793] loss 2.215 mtp 2.480 aux 0.25 gnorm 0.23 lr 1.2e-04 2.01s/step
+[26300/46793] loss 2.296 mtp 2.598 aux 0.26 gnorm 0.26 lr 1.2e-04 2.01s/step
+[26350/46793] loss 2.373 mtp 2.666 aux 0.25 gnorm 0.25 lr 1.2e-04 2.01s/step
+[26500/46793] loss 2.163 mtp 2.419 aux 0.21 gnorm 0.23 lr 1.2e-04 2.02s/step
+[26550/46793] loss 1.743 mtp 1.962 aux 0.20 gnorm 0.24 lr 1.2e-04 2.02s/step
+[26600/46793] loss 2.441 mtp 2.698 aux 0.22 gnorm 0.23 lr 1.2e-04 2.02s/step
+[26650/46793] loss 1.739 mtp 1.964 aux 0.21 gnorm 0.30 lr 1.2e-04 2.02s/step
+[26700/46793] loss 1.453 mtp 1.844 aux 0.46 gnorm 0.23 lr 1.2e-04 2.02s/step
+[26750/46793] loss 1.412 mtp 1.775 aux 0.54 gnorm 0.28 lr 1.2e-04 2.02s/step
+[26800/46793] loss 2.252 mtp 2.611 aux 0.35 gnorm 0.25 lr 1.2e-04 2.02s/step
+[26850/46793] loss 2.714 mtp 3.050 aux 0.27 gnorm 0.26 lr 1.2e-04 2.02s/step
+[26900/46793] loss 2.043 mtp 2.381 aux 0.34 gnorm 0.24 lr 1.2e-04 2.02s/step
+[26950/46793] loss 2.820 mtp 3.176 aux 0.26 gnorm 0.25 lr 1.1e-04 2.02s/step
+[27100/46793] loss 2.652 mtp 2.952 aux 0.23 gnorm 0.24 lr 1.1e-04 2.02s/step
+[27150/46793] loss 1.365 mtp 1.692 aux 0.33 gnorm 0.23 lr 1.1e-04 2.02s/step
+[27200/46793] loss 1.321 mtp 1.516 aux 0.22 gnorm 0.27 lr 1.1e-04 2.02s/step
+[27250/46793] loss 2.742 mtp 3.071 aux 0.25 gnorm 0.26 lr 1.1e-04 2.02s/step
+[27300/46793] loss 1.664 mtp 1.927 aux 0.27 gnorm 0.23 lr 1.1e-04 2.02s/step
+[27350/46793] loss 1.301 mtp 1.618 aux 0.32 gnorm 0.22 lr 1.1e-04 2.02s/step
+[27400/46793] loss 2.384 mtp 2.678 aux 0.23 gnorm 0.26 lr 1.1e-04 2.02s/step
+[27450/46793] loss 2.419 mtp 2.725 aux 0.24 gnorm 0.25 lr 1.1e-04 2.02s/step
+[27500/46793] loss 2.080 mtp 2.322 aux 0.20 gnorm 0.23 lr 1.1e-04 2.02s/step
+[27550/46793] loss 2.053 mtp 2.306 aux 0.19 gnorm 0.25 lr 1.1e-04 2.02s/step
+[27700/46793] loss 1.605 mtp 1.768 aux 0.15 gnorm 0.24 lr 1.1e-04 2.02s/step
+[27750/46793] loss 2.639 mtp 2.930 aux 0.23 gnorm 0.23 lr 1.1e-04 2.02s/step
+[27800/46793] loss 2.101 mtp 2.338 aux 0.19 gnorm 0.49 lr 1.1e-04 2.02s/step
+[27850/46793] loss 1.132 mtp 1.476 aux 0.46 gnorm 0.24 lr 1.1e-04 2.02s/step
+[27900/46793] loss 1.609 mtp 1.861 aux 0.29 gnorm 0.25 lr 1.1e-04 2.02s/step
+[27950/46793] loss 1.933 mtp 2.202 aux 0.23 gnorm 0.25 lr 1.0e-04 2.02s/step
+[val 27999] global=2.6838  code=1.231  fineweb_edu=3.471  glm_magpie=1.905  glm_regen=2.953  reap_calib=3.678  spdx=2.855  tinystories=1.047  wiki_en=3.176  wiki_zh=3.839
+[28000/46793] loss 2.148 mtp 2.416 aux 0.22 gnorm 0.23 lr 1.0e-04 2.02s/step
+[28050/46793] loss 2.002 mtp 2.260 aux 0.20 gnorm 0.25 lr 1.0e-04 2.02s/step
+[28100/46793] loss 1.408 mtp 1.587 aux 0.15 gnorm 0.24 lr 1.0e-04 2.02s/step
+[28150/46793] loss 2.062 mtp 2.309 aux 0.19 gnorm 0.26 lr 1.0e-04 2.02s/step
+[28300/46793] loss 2.445 mtp 2.721 aux 0.20 gnorm 0.25 lr 1.0e-04 2.02s/step
+[28350/46793] loss 1.988 mtp 2.245 aux 0.20 gnorm 0.22 lr 1.0e-04 2.02s/step
+[28400/46793] loss 0.906 mtp 1.143 aux 0.25 gnorm 0.22 lr 1.0e-04 2.02s/step
+[28450/46793] loss 1.639 mtp 1.857 aux 0.20 gnorm 0.24 lr 1.0e-04 2.02s/step
+[28500/46793] loss 2.650 mtp 2.951 aux 0.24 gnorm 0.25 lr 1.0e-04 2.02s/step
+[28550/46793] loss 2.168 mtp 2.451 aux 0.21 gnorm 0.24 lr 9.9e-05 2.02s/step
+[28600/46793] loss 2.710 mtp 3.071 aux 0.23 gnorm 0.25 lr 9.9e-05 2.02s/step
+[28650/46793] loss 1.641 mtp 1.848 aux 0.16 gnorm 0.29 lr 9.8e-05 2.02s/step
+[28700/46793] loss 2.166 mtp 2.558 aux 0.32 gnorm 0.25 lr 9.8e-05 2.02s/step
+[28750/46793] loss 2.364 mtp 2.638 aux 0.22 gnorm 0.26 lr 9.7e-05 2.03s/step
+[28900/46793] loss 1.758 mtp 2.040 aux 0.28 gnorm 0.24 lr 9.6e-05 2.03s/step
+[28950/46793] loss 1.452 mtp 1.847 aux 0.43 gnorm 0.23 lr 9.5e-05 2.03s/step
+[val 28999] global=2.6733  code=1.222  fineweb_edu=3.464  glm_magpie=1.896  glm_regen=2.949  reap_calib=3.666  spdx=2.825  tinystories=1.041  wiki_en=3.166  wiki_zh=3.831
+[29000/46793] loss 1.227 mtp 1.524 aux 0.30 gnorm 0.23 lr 9.5e-05 2.03s/step
+[29050/46793] loss 2.286 mtp 2.553 aux 0.21 gnorm 0.24 lr 9.4e-05 2.03s/step
+[29100/46793] loss 1.548 mtp 1.763 aux 0.17 gnorm 0.22 lr 9.4e-05 2.03s/step
+[29150/46793] loss 2.186 mtp 2.451 aux 0.21 gnorm 0.24 lr 9.3e-05 2.03s/step
+[29200/46793] loss 1.901 mtp 2.134 aux 0.19 gnorm 0.24 lr 9.3e-05 2.03s/step
+[29250/46793] loss 1.985 mtp 2.252 aux 0.20 gnorm 0.24 lr 9.3e-05 2.03s/step
+[29300/46793] loss 1.989 mtp 2.314 aux 0.26 gnorm 0.26 lr 9.2e-05 2.03s/step
+[29350/46793] loss 1.904 mtp 2.146 aux 0.16 gnorm 0.28 lr 9.2e-05 2.03s/step
+[29500/46793] loss 2.338 mtp 2.623 aux 0.19 gnorm 0.25 lr 9.0e-05 2.03s/step
+[29550/46793] loss 1.210 mtp 1.478 aux 0.27 gnorm 0.23 lr 9.0e-05 2.03s/step
+[29600/46793] loss 2.491 mtp 2.818 aux 0.21 gnorm 0.26 lr 8.9e-05 2.03s/step
+[29650/46793] loss 2.424 mtp 2.752 aux 0.21 gnorm 0.22 lr 8.9e-05 2.03s/step
+[29700/46793] loss 2.488 mtp 2.790 aux 0.25 gnorm 0.22 lr 8.8e-05 2.03s/step
+[29750/46793] loss 2.082 mtp 2.335 aux 0.17 gnorm 0.24 lr 8.8e-05 2.03s/step
+[29800/46793] loss 2.510 mtp 2.835 aux 0.21 gnorm 0.27 lr 8.7e-05 2.03s/step
+[29850/46793] loss 1.829 mtp 2.064 aux 0.19 gnorm 0.24 lr 8.7e-05 2.03s/step
+[29900/46793] loss 1.625 mtp 1.886 aux 0.26 gnorm 0.22 lr 8.7e-05 2.03s/step
+[29950/46793] loss 1.638 mtp 1.871 aux 0.18 gnorm 0.23 lr 8.6e-05 2.03s/step
+[30100/46793] loss 1.027 mtp 1.139 aux 0.09 gnorm 0.22 lr 8.5e-05 2.03s/step
+[30150/46793] loss 1.996 mtp 2.277 aux 0.23 gnorm 0.25 lr 8.4e-05 2.03s/step
+[30200/46793] loss 1.790 mtp 1.994 aux 0.15 gnorm 0.22 lr 8.4e-05 2.03s/step
+[30250/46793] loss 2.318 mtp 2.577 aux 0.19 gnorm 0.24 lr 8.3e-05 2.03s/step
+[30300/46793] loss 2.131 mtp 2.416 aux 0.19 gnorm 0.26 lr 8.3e-05 2.03s/step
+[30350/46793] loss 1.317 mtp 1.602 aux 0.28 gnorm 0.20 lr 8.2e-05 2.03s/step
+[30400/46793] loss 1.753 mtp 2.012 aux 0.24 gnorm 0.23 lr 8.2e-05 2.03s/step
+[30450/46793] loss 1.924 mtp 2.193 aux 0.19 gnorm 0.23 lr 8.2e-05 2.03s/step
+[30500/46793] loss 1.235 mtp 1.448 aux 0.25 gnorm 0.22 lr 8.1e-05 2.03s/step
+[30550/46793] loss 1.477 mtp 1.825 aux 0.30 gnorm 0.25 lr 8.1e-05 2.03s/step
+[30700/46793] loss 2.175 mtp 2.522 aux 0.30 gnorm 0.25 lr 7.9e-05 2.03s/step
+[30750/46793] loss 1.825 mtp 2.145 aux 0.30 gnorm 0.25 lr 7.9e-05 2.03s/step
+[30800/46793] loss 1.262 mtp 1.556 aux 0.33 gnorm 0.23 lr 7.8e-05 2.03s/step
+[30850/46793] loss 1.846 mtp 2.177 aux 0.31 gnorm 0.23 lr 7.8e-05 2.03s/step
+[30900/46793] loss 2.314 mtp 2.599 aux 0.21 gnorm 0.23 lr 7.8e-05 2.03s/step
+[30950/46793] loss 2.155 mtp 2.411 aux 0.17 gnorm 0.23 lr 7.7e-05 2.03s/step
+[val 30999] global=2.6693  code=1.209  fineweb_edu=3.460  glm_magpie=1.891  glm_regen=2.938  reap_calib=3.709  spdx=2.808  tinystories=1.038  wiki_en=3.154  wiki_zh=3.816
+[31000/46793] loss 1.486 mtp 1.717 aux 0.17 gnorm 0.24 lr 7.7e-05 2.03s/step
+[31050/46793] loss 2.623 mtp 2.970 aux 0.21 gnorm 0.24 lr 7.6e-05 2.03s/step
+[31100/46793] loss 2.078 mtp 2.354 aux 0.18 gnorm 0.26 lr 7.6e-05 2.03s/step
+[31150/46793] loss 3.135 mtp 3.494 aux 0.21 gnorm 0.23 lr 7.5e-05 2.03s/step
+[31300/46793] loss 2.309 mtp 2.647 aux 0.21 gnorm 0.25 lr 7.4e-05 2.04s/step
+[31350/46793] loss 2.749 mtp 3.066 aux 0.21 gnorm 0.25 lr 7.4e-05 2.04s/step
+[31400/46793] loss 1.783 mtp 2.033 aux 0.17 gnorm 0.25 lr 7.3e-05 2.04s/step
+[31450/46793] loss 1.510 mtp 1.742 aux 0.16 gnorm 0.23 lr 7.3e-05 2.04s/step
+[31500/46793] loss 2.271 mtp 2.576 aux 0.20 gnorm 0.25 lr 7.2e-05 2.04s/step
+[31550/46793] loss 1.843 mtp 2.080 aux 0.17 gnorm 0.25 lr 7.2e-05 2.04s/step
+[31600/46793] loss 1.991 mtp 2.327 aux 0.28 gnorm 0.23 lr 7.1e-05 2.04s/step
+[31650/46793] loss 2.139 mtp 2.386 aux 0.18 gnorm 0.24 lr 7.1e-05 2.04s/step
+[31700/46793] loss 1.372 mtp 1.571 aux 0.23 gnorm 0.22 lr 7.1e-05 2.04s/step
+[31750/46793] loss 1.409 mtp 1.620 aux 0.16 gnorm 0.23 lr 7.0e-05 2.04s/step
+[31900/46793] loss 1.475 mtp 1.706 aux 0.19 gnorm 0.22 lr 6.9e-05 2.04s/step
+[31950/46793] loss 1.379 mtp 1.632 aux 0.22 gnorm 0.23 lr 6.9e-05 2.04s/step
+[val 31999] global=2.6714  code=1.196  fineweb_edu=3.456  glm_magpie=1.886  glm_regen=2.933  reap_calib=3.699  spdx=2.873  tinystories=1.030  wiki_en=3.153  wiki_zh=3.816
+[32000/46793] loss 1.912 mtp 2.187 aux 0.23 gnorm 0.24 lr 6.8e-05 2.04s/step
+[32050/46793] loss 1.850 mtp 2.121 aux 0.19 gnorm 0.24 lr 6.8e-05 2.04s/step
+[32100/46793] loss 2.477 mtp 2.794 aux 0.20 gnorm 0.21 lr 6.7e-05 2.04s/step
+[32150/46793] loss 1.928 mtp 2.271 aux 0.27 gnorm 0.23 lr 6.7e-05 2.04s/step
+[32200/46793] loss 1.422 mtp 1.629 aux 0.14 gnorm 0.23 lr 6.6e-05 2.04s/step
+[32250/46793] loss 2.375 mtp 2.676 aux 0.18 gnorm 0.24 lr 6.6e-05 2.04s/step
+[32300/46793] loss 1.442 mtp 1.629 aux 0.14 gnorm 0.24 lr 6.6e-05 2.04s/step
+[32350/46793] loss 2.619 mtp 2.947 aux 0.21 gnorm 0.25 lr 6.5e-05 2.04s/step
+[32500/46793] loss 1.246 mtp 1.543 aux 0.26 gnorm 0.24 lr 6.4e-05 2.04s/step
+[32550/46793] loss 1.658 mtp 1.890 aux 0.17 gnorm 0.25 lr 6.4e-05 2.04s/step
+[32600/46793] loss 1.132 mtp 1.315 aux 0.15 gnorm 0.24 lr 6.3e-05 2.04s/step
+[32650/46793] loss 1.876 mtp 2.225 aux 0.27 gnorm 0.25 lr 6.3e-05 2.04s/step
+[32700/46793] loss 1.535 mtp 1.882 aux 0.30 gnorm 0.23 lr 6.2e-05 2.04s/step
+[32750/46793] loss 1.804 mtp 2.065 aux 0.21 gnorm 0.26 lr 6.2e-05 2.04s/step
+[32800/46793] loss 1.746 mtp 1.972 aux 0.19 gnorm 0.26 lr 6.1e-05 2.04s/step
+[32850/46793] loss 1.787 mtp 2.088 aux 0.26 gnorm 0.23 lr 6.1e-05 2.04s/step
+[32900/46793] loss 1.552 mtp 1.792 aux 0.23 gnorm 0.24 lr 6.1e-05 2.04s/step
+[32950/46793] loss 2.017 mtp 2.289 aux 0.17 gnorm 0.24 lr 6.0e-05 2.04s/step
+[33100/46793] loss 1.811 mtp 2.071 aux 0.18 gnorm 0.20 lr 5.9e-05 2.04s/step
+[33150/46793] loss 2.114 mtp 2.427 aux 0.19 gnorm 0.24 lr 5.9e-05 2.04s/step
+[33200/46793] loss 1.556 mtp 1.796 aux 0.17 gnorm 0.23 lr 5.8e-05 2.04s/step
+[33250/46793] loss 2.151 mtp 2.421 aux 0.16 gnorm 0.23 lr 5.8e-05 2.04s/step
+[33300/46793] loss 2.164 mtp 2.485 aux 0.18 gnorm 0.24 lr 5.7e-05 2.04s/step
+[33350/46793] loss 2.596 mtp 2.925 aux 0.19 gnorm 0.25 lr 5.7e-05 2.04s/step
+[33400/46793] loss 2.202 mtp 2.490 aux 0.17 gnorm 0.22 lr 5.7e-05 2.04s/step
+[33450/46793] loss 1.580 mtp 1.809 aux 0.18 gnorm 0.22 lr 5.6e-05 2.04s/step
+[33500/46793] loss 1.645 mtp 1.920 aux 0.24 gnorm 0.24 lr 5.6e-05 2.04s/step
+[33550/46793] loss 1.891 mtp 2.154 aux 0.19 gnorm 0.24 lr 5.5e-05 2.04s/step
+[33700/46793] loss 2.116 mtp 2.348 aux 0.14 gnorm 0.25 lr 5.4e-05 2.04s/step
+[33750/46793] loss 2.197 mtp 2.500 aux 0.25 gnorm 0.24 lr 5.4e-05 2.04s/step
+[33800/46793] loss 2.058 mtp 2.345 aux 0.15 gnorm 0.24 lr 5.4e-05 2.04s/step
+[33850/46793] loss 2.122 mtp 2.516 aux 0.28 gnorm 0.23 lr 5.3e-05 2.04s/step
+[33900/46793] loss 2.164 mtp 2.380 aux 0.13 gnorm 0.21 lr 5.3e-05 2.04s/step
+[33950/46793] loss 1.611 mtp 1.844 aux 0.17 gnorm 0.24 lr 5.2e-05 2.04s/step
+[val 33999] global=2.6627  code=1.189  fineweb_edu=3.449  glm_magpie=1.883  glm_regen=2.922  reap_calib=3.709  spdx=2.827  tinystories=1.029  wiki_en=3.148  wiki_zh=3.809
+[34000/46793] loss 1.555 mtp 1.799 aux 0.16 gnorm 0.24 lr 5.2e-05 2.04s/step
+[34050/46793] loss 1.767 mtp 2.120 aux 0.28 gnorm 0.23 lr 5.2e-05 2.04s/step
+[34100/46793] loss 1.764 mtp 2.082 aux 0.25 gnorm 0.24 lr 5.1e-05 2.04s/step
+[34150/46793] loss 1.745 mtp 2.090 aux 0.28 gnorm 0.23 lr 5.1e-05 2.04s/step
+[34300/46793] loss 2.453 mtp 2.717 aux 0.16 gnorm 0.26 lr 5.0e-05 2.05s/step
+[34350/46793] loss 1.771 mtp 1.979 aux 0.12 gnorm 0.22 lr 4.9e-05 2.05s/step
+[34400/46793] loss 1.945 mtp 2.199 aux 0.15 gnorm 0.25 lr 4.9e-05 2.05s/step
+[34450/46793] loss 1.862 mtp 2.191 aux 0.29 gnorm 0.22 lr 4.9e-05 2.05s/step
+[34500/46793] loss 1.123 mtp 1.453 aux 0.38 gnorm 0.23 lr 4.8e-05 2.05s/step
+[34550/46793] loss 2.047 mtp 2.301 aux 0.18 gnorm 0.25 lr 4.8e-05 2.05s/step
+[34600/46793] loss 1.958 mtp 2.203 aux 0.16 gnorm 0.24 lr 4.8e-05 2.05s/step
+[34650/46793] loss 1.543 mtp 1.717 aux 0.13 gnorm 0.23 lr 4.7e-05 2.05s/step
+[34700/46793] loss 1.909 mtp 2.217 aux 0.19 gnorm 0.25 lr 4.7e-05 2.05s/step
+[34750/46793] loss 2.711 mtp 3.021 aux 0.17 gnorm 0.25 lr 4.6e-05 2.05s/step
+[34900/46793] loss 2.972 mtp 3.290 aux 0.19 gnorm 0.23 lr 4.5e-05 2.05s/step
+[34950/46793] loss 2.042 mtp 2.318 aux 0.14 gnorm 0.23 lr 4.5e-05 2.05s/step
+[val 34999] global=2.6604  code=1.187  fineweb_edu=3.452  glm_magpie=1.881  glm_regen=2.924  reap_calib=3.716  spdx=2.809  tinystories=1.026  wiki_en=3.145  wiki_zh=3.803
+[35000/46793] loss 1.991 mtp 2.236 aux 0.16 gnorm 0.24 lr 4.5e-05 2.05s/step
+[35050/46793] loss 2.882 mtp 3.254 aux 0.30 gnorm 0.25 lr 4.4e-05 2.05s/step
+[35100/46793] loss 2.036 mtp 2.297 aux 0.16 gnorm 0.25 lr 4.4e-05 2.05s/step
+[35150/46793] loss 2.162 mtp 2.492 aux 0.18 gnorm 0.25 lr 4.4e-05 2.05s/step
+[35200/46793] loss 1.831 mtp 2.243 aux 0.26 gnorm 0.22 lr 4.3e-05 2.05s/step
+[35250/46793] loss 1.148 mtp 1.334 aux 0.14 gnorm 0.22 lr 4.3e-05 2.05s/step
+[35300/46793] loss 1.884 mtp 2.218 aux 0.28 gnorm 0.24 lr 4.2e-05 2.05s/step
+[35350/46793] loss 2.128 mtp 2.368 aux 0.14 gnorm 0.23 lr 4.2e-05 2.05s/step
+[35500/46793] loss 1.220 mtp 1.470 aux 0.23 gnorm 0.26 lr 4.1e-05 2.05s/step
+[35550/46793] loss 1.585 mtp 1.943 aux 0.30 gnorm 0.22 lr 4.1e-05 2.05s/step
+[35600/46793] loss 1.310 mtp 1.569 aux 0.23 gnorm 0.24 lr 4.0e-05 2.05s/step
+[35650/46793] loss 2.507 mtp 2.768 aux 0.14 gnorm 0.25 lr 4.0e-05 2.05s/step
+[35700/46793] loss 1.871 mtp 2.167 aux 0.18 gnorm 0.23 lr 4.0e-05 2.05s/step
+[35750/46793] loss 1.114 mtp 1.374 aux 0.25 gnorm 0.25 lr 3.9e-05 2.05s/step
+[35800/46793] loss 1.513 mtp 1.754 aux 0.15 gnorm 0.23 lr 3.9e-05 2.05s/step
+[35850/46793] loss 1.867 mtp 2.165 aux 0.25 gnorm 0.23 lr 3.9e-05 2.05s/step
+[35900/46793] loss 2.365 mtp 2.695 aux 0.18 gnorm 0.23 lr 3.8e-05 2.05s/step
+[35950/46793] loss 2.119 mtp 2.373 aux 0.15 gnorm 0.24 lr 3.8e-05 2.05s/step
+[36100/46793] loss 1.784 mtp 2.035 aux 0.16 gnorm 0.23 lr 3.7e-05 2.05s/step
+[36150/46793] loss 1.545 mtp 1.759 aux 0.13 gnorm 0.23 lr 3.7e-05 2.05s/step
+[36200/46793] loss 1.196 mtp 1.451 aux 0.32 gnorm 0.21 lr 3.6e-05 2.05s/step
+[36250/46793] loss 2.148 mtp 2.466 aux 0.18 gnorm 0.24 lr 3.6e-05 2.05s/step
+[36300/46793] loss 2.149 mtp 2.451 aux 0.17 gnorm 0.26 lr 3.6e-05 2.05s/step
+[36350/46793] loss 1.246 mtp 1.479 aux 0.15 gnorm 0.24 lr 3.5e-05 2.05s/step
+[36400/46793] loss 2.299 mtp 2.573 aux 0.17 gnorm 0.23 lr 3.5e-05 2.05s/step
+[36450/46793] loss 2.322 mtp 2.590 aux 0.15 gnorm 0.25 lr 3.5e-05 2.05s/step
+[36500/46793] loss 1.005 mtp 1.411 aux 0.46 gnorm 0.24 lr 3.4e-05 2.05s/step
+[36550/46793] loss 1.353 mtp 1.697 aux 0.34 gnorm 0.22 lr 3.4e-05 2.05s/step
+[36700/46793] loss 2.222 mtp 2.471 aux 0.15 gnorm 0.25 lr 3.3e-05 2.05s/step
+[36750/46793] loss 1.739 mtp 1.929 aux 0.12 gnorm 0.22 lr 3.3e-05 2.05s/step
+[36800/46793] loss 2.002 mtp 2.230 aux 0.13 gnorm 0.24 lr 3.3e-05 2.05s/step
+[36850/46793] loss 1.317 mtp 1.610 aux 0.35 gnorm 0.21 lr 3.2e-05 2.05s/step
+[36900/46793] loss 2.292 mtp 2.543 aux 0.15 gnorm 0.22 lr 3.2e-05 2.05s/step
+[36950/46793] loss 1.083 mtp 1.271 aux 0.21 gnorm 0.23 lr 3.2e-05 2.05s/step
+[val 36999] global=2.6583  code=1.188  fineweb_edu=3.447  glm_magpie=1.881  glm_regen=2.921  reap_calib=3.710  spdx=2.808  tinystories=1.024  wiki_en=3.143  wiki_zh=3.802
+[37000/46793] loss 2.146 mtp 2.429 aux 0.16 gnorm 0.24 lr 3.1e-05 2.05s/step
+[37050/46793] loss 1.880 mtp 2.156 aux 0.23 gnorm 0.24 lr 3.1e-05 2.05s/step
+[37100/46793] loss 2.161 mtp 2.484 aux 0.18 gnorm 0.26 lr 3.1e-05 2.05s/step
+[37150/46793] loss 1.208 mtp 1.483 aux 0.31 gnorm 0.25 lr 3.0e-05 2.05s/step
+[37300/46793] loss 1.714 mtp 1.912 aux 0.12 gnorm 0.23 lr 2.9e-05 2.05s/step
+[37350/46793] loss 2.083 mtp 2.369 aux 0.16 gnorm 0.24 lr 2.9e-05 2.05s/step
+[37400/46793] loss 2.044 mtp 2.329 aux 0.18 gnorm 0.24 lr 2.9e-05 2.05s/step
+[37450/46793] loss 1.650 mtp 1.945 aux 0.25 gnorm 0.24 lr 2.9e-05 2.05s/step
+[37500/46793] loss 2.168 mtp 2.505 aux 0.23 gnorm 0.24 lr 2.8e-05 2.06s/step
+[37550/46793] loss 2.585 mtp 2.893 aux 0.18 gnorm 0.24 lr 2.8e-05 2.06s/step
+[37600/46793] loss 2.486 mtp 2.790 aux 0.17 gnorm 0.24 lr 2.8e-05 2.06s/step
+[37650/46793] loss 1.890 mtp 2.197 aux 0.27 gnorm 0.23 lr 2.7e-05 2.05s/step
+[37700/46793] loss 2.971 mtp 3.320 aux 0.19 gnorm 0.25 lr 2.7e-05 2.05s/step
+[37750/46793] loss 1.661 mtp 1.970 aux 0.26 gnorm 0.23 lr 2.7e-05 2.06s/step
+[37900/46793] loss 2.281 mtp 2.510 aux 0.12 gnorm 0.24 lr 2.6e-05 2.06s/step
+[37950/46793] loss 2.547 mtp 2.954 aux 0.30 gnorm 0.24 lr 2.6e-05 2.06s/step
+[val 37999] global=2.6565  code=1.188  fineweb_edu=3.449  glm_magpie=1.880  glm_regen=2.920  reap_calib=3.704  spdx=2.802  tinystories=1.023  wiki_en=3.142  wiki_zh=3.799
+[38000/46793] loss 1.879 mtp 2.189 aux 0.23 gnorm 0.24 lr 2.5e-05 2.06s/step
+[38050/46793] loss 1.935 mtp 2.220 aux 0.24 gnorm 0.22 lr 2.5e-05 2.06s/step
+[38100/46793] loss 1.553 mtp 1.798 aux 0.17 gnorm 0.24 lr 2.5e-05 2.06s/step
+[38150/46793] loss 1.738 mtp 1.999 aux 0.14 gnorm 0.24 lr 2.5e-05 2.06s/step
+[38200/46793] loss 2.289 mtp 2.565 aux 0.17 gnorm 0.25 lr 2.4e-05 2.06s/step
+[38250/46793] loss 2.040 mtp 2.268 aux 0.13 gnorm 0.22 lr 2.4e-05 2.06s/step
+[38300/46793] loss 2.134 mtp 2.465 aux 0.19 gnorm 0.25 lr 2.4e-05 2.06s/step
+[38350/46793] loss 1.847 mtp 2.063 aux 0.13 gnorm 0.24 lr 2.3e-05 2.06s/step
+[38700/46793] loss 2.028 mtp 2.372 aux 0.25 gnorm 0.21 lr 2.2e-05 2.06s/step
+[38750/46793] loss 1.997 mtp 2.281 aux 0.16 gnorm 0.24 lr 2.1e-05 2.06s/step
+[38800/46793] loss 1.454 mtp 1.654 aux 0.13 gnorm 0.24 lr 2.1e-05 2.06s/step
+[38850/46793] loss 1.344 mtp 1.595 aux 0.26 gnorm 0.22 lr 2.1e-05 2.06s/step
+[38900/46793] loss 1.195 mtp 1.369 aux 0.13 gnorm 0.21 lr 2.1e-05 2.06s/step
+[38950/46793] loss 1.412 mtp 1.658 aux 0.21 gnorm 0.22 lr 2.0e-05 2.06s/step
+[39100/46793] loss 2.085 mtp 2.553 aux 0.32 gnorm 0.24 lr 2.0e-05 2.06s/step
+[39150/46793] loss 2.005 mtp 2.291 aux 0.18 gnorm 0.22 lr 1.9e-05 2.06s/step
+[39200/46793] loss 2.631 mtp 3.022 aux 0.26 gnorm 0.23 lr 1.9e-05 2.06s/step
+[39250/46793] loss 2.563 mtp 2.863 aux 0.17 gnorm 0.24 lr 1.9e-05 2.06s/step
+[39300/46793] loss 1.618 mtp 1.917 aux 0.24 gnorm 0.26 lr 1.9e-05 2.06s/step
+[39350/46793] loss 2.147 mtp 2.409 aux 0.16 gnorm 0.22 lr 1.8e-05 2.06s/step
+[39400/46793] loss 1.209 mtp 1.368 aux 0.11 gnorm 0.23 lr 1.8e-05 2.06s/step
+[39450/46793] loss 1.686 mtp 1.993 aux 0.29 gnorm 0.23 lr 1.8e-05 2.06s/step
+[39500/46793] loss 2.086 mtp 2.394 aux 0.25 gnorm 0.21 lr 1.8e-05 2.06s/step
+[39550/46793] loss 2.247 mtp 2.556 aux 0.18 gnorm 0.24 lr 1.7e-05 2.06s/step
+[39700/46793] loss 2.267 mtp 2.583 aux 0.19 gnorm 0.24 lr 1.7e-05 2.06s/step
+[39750/46793] loss 2.652 mtp 2.971 aux 0.18 gnorm 0.24 lr 1.6e-05 2.06s/step
+[39800/46793] loss 1.745 mtp 1.997 aux 0.16 gnorm 0.24 lr 1.6e-05 2.06s/step
+[39850/46793] loss 1.985 mtp 2.242 aux 0.15 gnorm 0.22 lr 1.6e-05 2.06s/step
+[39900/46793] loss 2.159 mtp 2.408 aux 0.14 gnorm 0.24 lr 1.6e-05 2.06s/step
+[39950/46793] loss 2.914 mtp 3.222 aux 0.17 gnorm 0.24 lr 1.6e-05 2.06s/step
+[val 39999] global=2.6556  code=1.187  fineweb_edu=3.446  glm_magpie=1.878  glm_regen=2.921  reap_calib=3.707  spdx=2.796  tinystories=1.022  wiki_en=3.141  wiki_zh=3.800
+[40000/46793] loss 1.339 mtp 1.681 aux 0.28 gnorm 0.23 lr 1.5e-05 2.06s/step
+[40050/46793] loss 2.681 mtp 3.001 aux 0.18 gnorm 0.24 lr 1.5e-05 2.06s/step
+[40100/46793] loss 0.983 mtp 1.152 aux 0.11 gnorm 0.23 lr 1.5e-05 2.06s/step
+[40150/46793] loss 1.700 mtp 1.986 aux 0.24 gnorm 0.23 lr 1.5e-05 2.06s/step
+[40350/46793] loss 1.176 mtp 1.499 aux 0.32 gnorm 0.23 lr 1.4e-05 2.06s/step
+[40400/46793] loss 2.530 mtp 2.852 aux 0.19 gnorm 0.25 lr 1.4e-05 2.06s/step
+[40450/46793] loss 1.658 mtp 2.034 aux 0.34 gnorm 0.23 lr 1.3e-05 2.06s/step
+[40500/46793] loss 1.626 mtp 1.910 aux 0.16 gnorm 0.24 lr 1.3e-05 2.06s/step
+[40550/46793] loss 1.736 mtp 1.996 aux 0.17 gnorm 0.23 lr 1.3e-05 2.06s/step
+[40600/46793] loss 1.283 mtp 1.504 aux 0.16 gnorm 0.24 lr 1.3e-05 2.06s/step
+[40650/46793] loss 2.101 mtp 2.523 aux 0.32 gnorm 0.26 lr 1.3e-05 2.06s/step
+[40700/46793] loss 1.240 mtp 1.426 aux 0.11 gnorm 0.23 lr 1.2e-05 2.06s/step
+[40750/46793] loss 1.840 mtp 2.111 aux 0.24 gnorm 0.24 lr 1.2e-05 2.06s/step
+[40900/46793] loss 2.433 mtp 2.692 aux 0.15 gnorm 0.22 lr 1.2e-05 2.06s/step
+[40950/46793] loss 2.655 mtp 2.980 aux 0.17 gnorm 0.23 lr 1.1e-05 2.06s/step
+[val 40999] global=2.6581  code=1.187  fineweb_edu=3.447  glm_magpie=1.880  glm_regen=2.921  reap_calib=3.721  spdx=2.804  tinystories=1.022  wiki_en=3.140  wiki_zh=3.801
+[41000/46793] loss 1.533 mtp 1.717 aux 0.12 gnorm 0.23 lr 1.1e-05 2.06s/step
+[41050/46793] loss 1.596 mtp 1.848 aux 0.20 gnorm 0.23 lr 1.1e-05 2.06s/step
+[41100/46793] loss 1.419 mtp 1.664 aux 0.23 gnorm 0.25 lr 1.1e-05 2.06s/step
+[41150/46793] loss 1.555 mtp 1.822 aux 0.25 gnorm 0.25 lr 1.1e-05 2.06s/step
+[41200/46793] loss 2.034 mtp 2.291 aux 0.16 gnorm 0.23 lr 1.0e-05 2.06s/step
+[41250/46793] loss 1.842 mtp 2.031 aux 0.13 gnorm 0.22 lr 1.0e-05 2.06s/step
+[41300/46793] loss 2.337 mtp 2.584 aux 0.13 gnorm 0.23 lr 1.0e-05 2.06s/step
+[41350/46793] loss 2.528 mtp 2.810 aux 0.15 gnorm 0.24 lr 9.9e-06 2.06s/step
+[41500/46793] loss 1.737 mtp 2.053 aux 0.26 gnorm 0.25 lr 9.4e-06 2.06s/step
+[41550/46793] loss 1.928 mtp 2.250 aux 0.18 gnorm 0.23 lr 9.2e-06 2.06s/step
+[41600/46793] loss 3.082 mtp 3.465 aux 0.20 gnorm 0.25 lr 9.0e-06 2.06s/step
+[41650/46793] loss 1.627 mtp 1.864 aux 0.20 gnorm 0.22 lr 8.9e-06 2.06s/step
+[41700/46793] loss 1.993 mtp 2.251 aux 0.17 gnorm 0.23 lr 8.7e-06 2.06s/step
+[41750/46793] loss 1.589 mtp 1.789 aux 0.11 gnorm 0.23 lr 8.5e-06 2.07s/step
+[41800/46793] loss 1.818 mtp 2.063 aux 0.14 gnorm 0.23 lr 8.3e-06 2.06s/step
+[41850/46793] loss 1.981 mtp 2.274 aux 0.25 gnorm 0.22 lr 8.2e-06 2.06s/step
+[41900/46793] loss 1.556 mtp 1.772 aux 0.13 gnorm 0.24 lr 8.0e-06 2.06s/step
+[41950/46793] loss 2.314 mtp 2.623 aux 0.22 gnorm 0.24 lr 7.9e-06 2.06s/step
+[42100/46793] loss 2.478 mtp 2.818 aux 0.17 gnorm 0.24 lr 7.4e-06 2.07s/step
+[42150/46793] loss 1.213 mtp 1.438 aux 0.16 gnorm 0.25 lr 7.2e-06 2.07s/step
+[42200/46793] loss 2.403 mtp 2.683 aux 0.17 gnorm 0.22 lr 7.1e-06 2.07s/step
+[42250/46793] loss 1.532 mtp 1.892 aux 0.29 gnorm 0.23 lr 6.9e-06 2.07s/step
+[42300/46793] loss 2.501 mtp 2.862 aux 0.19 gnorm 0.23 lr 6.8e-06 2.07s/step
+[42350/46793] loss 0.940 mtp 1.230 aux 0.42 gnorm 0.20 lr 6.6e-06 2.07s/step
+[42400/46793] loss 1.800 mtp 2.060 aux 0.24 gnorm 0.23 lr 6.5e-06 2.07s/step
+[42450/46793] loss 2.044 mtp 2.321 aux 0.23 gnorm 0.24 lr 6.3e-06 2.07s/step
+[42500/46793] loss 1.658 mtp 1.912 aux 0.15 gnorm 0.23 lr 6.2e-06 2.07s/step
+[42550/46793] loss 1.698 mtp 2.061 aux 0.29 gnorm 0.23 lr 6.0e-06 2.07s/step
+[42700/46793] loss 1.519 mtp 1.768 aux 0.22 gnorm 0.24 lr 5.6e-06 2.07s/step
+[42750/46793] loss 1.503 mtp 1.686 aux 0.11 gnorm 0.23 lr 5.5e-06 2.07s/step
+[42800/46793] loss 2.685 mtp 2.979 aux 0.16 gnorm 0.24 lr 5.4e-06 2.07s/step
+[42850/46793] loss 2.350 mtp 2.623 aux 0.14 gnorm 0.24 lr 5.2e-06 2.07s/step
+[42900/46793] loss 2.638 mtp 2.974 aux 0.18 gnorm 0.23 lr 5.1e-06 2.07s/step
+[42950/46793] loss 2.786 mtp 3.111 aux 0.19 gnorm 0.22 lr 5.0e-06 2.07s/step
+[val 42999] global=2.6560  code=1.188  fineweb_edu=3.447  glm_magpie=1.880  glm_regen=2.920  reap_calib=3.707  spdx=2.799  tinystories=1.022  wiki_en=3.141  wiki_zh=3.800
+[43000/46793] loss 2.350 mtp 2.655 aux 0.17 gnorm 0.23 lr 4.8e-06 2.07s/step
+[43050/46793] loss 1.590 mtp 1.956 aux 0.26 gnorm 0.24 lr 4.7e-06 2.07s/step
+[43100/46793] loss 1.898 mtp 2.165 aux 0.15 gnorm 0.23 lr 4.6e-06 2.07s/step
+[43150/46793] loss 2.621 mtp 2.969 aux 0.19 gnorm 0.24 lr 4.5e-06 2.07s/step
+[43300/46793] loss 1.696 mtp 2.156 aux 0.34 gnorm 0.25 lr 4.1e-06 2.07s/step
+[43350/46793] loss 1.009 mtp 1.207 aux 0.18 gnorm 0.23 lr 4.0e-06 2.07s/step
+[43400/46793] loss 1.825 mtp 2.117 aux 0.25 gnorm 0.21 lr 3.9e-06 2.07s/step
+[43450/46793] loss 2.136 mtp 2.440 aux 0.17 gnorm 0.24 lr 3.8e-06 2.07s/step
+[43500/46793] loss 1.706 mtp 1.938 aux 0.13 gnorm 0.23 lr 3.7e-06 2.07s/step
+[43550/46793] loss 1.945 mtp 2.256 aux 0.25 gnorm 0.23 lr 3.5e-06 2.07s/step
+[43600/46793] loss 1.962 mtp 2.258 aux 0.17 gnorm 0.23 lr 3.4e-06 2.07s/step
+[43650/46793] loss 2.677 mtp 3.030 aux 0.18 gnorm 0.23 lr 3.3e-06 2.07s/step
+[43700/46793] loss 1.032 mtp 1.243 aux 0.20 gnorm 0.21 lr 3.2e-06 2.07s/step
+[43750/46793] loss 1.949 mtp 2.329 aux 0.28 gnorm 0.24 lr 3.1e-06 2.07s/step
+[43900/46793] loss 2.712 mtp 3.040 aux 0.18 gnorm 0.23 lr 2.8e-06 2.07s/step
+[43950/46793] loss 1.833 mtp 2.059 aux 0.14 gnorm 0.24 lr 2.7e-06 2.07s/step
+[val 43999] global=2.6575  code=1.187  fineweb_edu=3.449  glm_magpie=1.879  glm_regen=2.920  reap_calib=3.715  spdx=2.802  tinystories=1.023  wiki_en=3.141  wiki_zh=3.800
+[44000/46793] loss 2.301 mtp 2.691 aux 0.30 gnorm 0.25 lr 2.6e-06 2.07s/step
+[44050/46793] loss 1.720 mtp 2.045 aux 0.17 gnorm 0.23 lr 2.5e-06 2.07s/step
+[44100/46793] loss 2.322 mtp 2.609 aux 0.17 gnorm 0.25 lr 2.4e-06 2.07s/step
+[44150/46793] loss 2.200 mtp 2.508 aux 0.17 gnorm 0.24 lr 2.4e-06 2.07s/step
+[44200/46793] loss 1.746 mtp 2.013 aux 0.14 gnorm 0.22 lr 2.3e-06 2.07s/step
+[44250/46793] loss 1.120 mtp 1.448 aux 0.32 gnorm 0.22 lr 2.2e-06 2.07s/step
+[44300/46793] loss 1.884 mtp 2.196 aux 0.15 gnorm 0.25 lr 2.1e-06 2.07s/step
+[44350/46793] loss 1.831 mtp 2.146 aux 0.19 gnorm 0.26 lr 2.0e-06 2.07s/step
+[44500/46793] loss 1.903 mtp 2.187 aux 0.16 gnorm 0.23 lr 1.8e-06 2.07s/step
+[44550/46793] loss 1.486 mtp 1.786 aux 0.32 gnorm 0.22 lr 1.7e-06 2.07s/step
+[44600/46793] loss 1.752 mtp 2.054 aux 0.24 gnorm 0.25 lr 1.6e-06 2.07s/step
+[44650/46793] loss 2.279 mtp 2.560 aux 0.16 gnorm 0.24 lr 1.5e-06 2.07s/step
+[44700/46793] loss 1.369 mtp 1.648 aux 0.23 gnorm 0.24 lr 1.5e-06 2.07s/step
+[44750/46793] loss 2.184 mtp 2.540 aux 0.26 gnorm 0.24 lr 1.4e-06 2.07s/step
+[44800/46793] loss 2.030 mtp 2.274 aux 0.17 gnorm 0.25 lr 1.3e-06 2.07s/step
+[44850/46793] loss 2.811 mtp 3.125 aux 0.18 gnorm 0.24 lr 1.3e-06 2.07s/step
+[44900/46793] loss 1.363 mtp 1.587 aux 0.13 gnorm 0.23 lr 1.2e-06 2.07s/step
+[44950/46793] loss 2.126 mtp 2.410 aux 0.15 gnorm 0.25 lr 1.1e-06 2.07s/step
+[45100/46793] loss 1.755 mtp 2.010 aux 0.17 gnorm 0.25 lr 9.7e-07 2.07s/step
+[45150/46793] loss 2.239 mtp 2.566 aux 0.18 gnorm 0.24 lr 9.1e-07 2.07s/step
+[45200/46793] loss 1.876 mtp 2.201 aux 0.18 gnorm 0.24 lr 8.6e-07 2.07s/step
+[45250/46793] loss 2.334 mtp 2.671 aux 0.23 gnorm 0.25 lr 8.0e-07 2.07s/step
+[45300/46793] loss 1.453 mtp 1.656 aux 0.14 gnorm 0.23 lr 7.5e-07 2.07s/step
+[45350/46793] loss 1.948 mtp 2.202 aux 0.15 gnorm 0.25 lr 7.0e-07 2.07s/step
+[45400/46793] loss 2.685 mtp 3.009 aux 0.19 gnorm 0.24 lr 6.6e-07 2.07s/step
+[45450/46793] loss 2.692 mtp 2.985 aux 0.17 gnorm 0.23 lr 6.1e-07 2.07s/step
+[45500/46793] loss 2.620 mtp 3.019 aux 0.28 gnorm 0.22 lr 5.6e-07 2.07s/step
+[45550/46793] loss 2.245 mtp 2.552 aux 0.17 gnorm 0.25 lr 5.2e-07 2.07s/step
+[45700/46793] loss 1.625 mtp 1.885 aux 0.22 gnorm 0.24 lr 4.0e-07 2.07s/step
+[45750/46793] loss 1.204 mtp 1.524 aux 0.23 gnorm 0.23 lr 3.7e-07 2.07s/step
+[45800/46793] loss 2.572 mtp 2.855 aux 0.17 gnorm 0.25 lr 3.3e-07 2.07s/step
+[45850/46793] loss 1.635 mtp 1.990 aux 0.26 gnorm 0.23 lr 3.0e-07 2.07s/step
+[45900/46793] loss 2.889 mtp 3.226 aux 0.16 gnorm 0.25 lr 2.7e-07 2.07s/step
+[45950/46793] loss 1.282 mtp 1.482 aux 0.13 gnorm 0.24 lr 2.4e-07 2.07s/step
+[val 45999] global=2.6577  code=1.188  fineweb_edu=3.447  glm_magpie=1.879  glm_regen=2.920  reap_calib=3.719  spdx=2.803  tinystories=1.023  wiki_en=3.142  wiki_zh=3.799
+[46000/46793] loss 1.551 mtp 1.785 aux 0.18 gnorm 0.23 lr 2.1e-07 2.07s/step
+[46050/46793] loss 2.885 mtp 3.213 aux 0.17 gnorm 0.23 lr 1.9e-07 2.07s/step
+[46100/46793] loss 2.407 mtp 2.680 aux 0.15 gnorm 0.25 lr 1.6e-07 2.07s/step
+[46150/46793] loss 1.537 mtp 1.729 aux 0.14 gnorm 0.63 lr 1.4e-07 2.07s/step
+[46300/46793] loss 2.196 mtp 2.488 aux 0.16 gnorm 0.25 lr 8.2e-08 2.07s/step
+[46350/46793] loss 1.882 mtp 2.145 aux 0.14 gnorm 0.23 lr 6.6e-08 2.07s/step
+[46400/46793] loss 1.349 mtp 1.646 aux 0.35 gnorm 0.24 lr 5.2e-08 2.07s/step
+[46450/46793] loss 2.251 mtp 2.555 aux 0.24 gnorm 0.22 lr 4.0e-08 2.07s/step
+[46500/46793] loss 1.666 mtp 1.931 aux 0.22 gnorm 0.24 lr 2.9e-08 2.07s/step
+[46550/46793] loss 1.496 mtp 1.718 aux 0.15 gnorm 0.23 lr 2.0e-08 2.07s/step
+[46600/46793] loss 1.455 mtp 1.701 aux 0.23 gnorm 0.23 lr 1.3e-08 2.07s/step
+[46650/46793] loss 2.797 mtp 3.147 aux 0.19 gnorm 0.24 lr 6.9e-09 2.07s/step
+[46700/46793] loss 1.175 mtp 1.344 aux 0.09 gnorm 0.22 lr 2.9e-09 2.07s/step
+[46750/46793] loss 1.404 mtp 1.620 aux 0.13 gnorm 0.23 lr 6.3e-10 2.07s/step
+[46792/46793] loss 1.580 mtp 1.896 aux 0.24 gnorm 0.23 lr 3.4e-13 2.07s/step