Upload training_state.json with huggingface_hub

training_state.json

@@ -1,6 +1,6 @@
 {
-  "step": 11000,
-  "tokens": 1081344000,
+  "step": 11100,
+  "tokens": 1091174400,
   "parameters": {
     "total": 998694784,
     "active": 340154240,
@@ -9,31 +9,29 @@
   },
   "batch_size": 96,
   "sequence_length": 1024,
-  "loss": 1.0,
-  "auxiliary_loss": 1.2890625,
-  "gradient_norm": 0.1376953125,
-  "route": 1,
-  "lr_scale": 0.9775364585953472,
-  "tokens_per_second": 455876.52484148124,
-  "elapsed_seconds": 1821.3779653700021,
-  "average_patch_bytes": 9.266968325791856,
-  "patch_coalesced_rate": 0.052083333333333336,
+  "loss": 0.9765625,
+  "auxiliary_loss": 1.3515625,
+  "gradient_norm": 0.138671875,
+  "route": 2,
+  "lr_scale": 0.9770887670336095,
+  "tokens_per_second": 456579.6963227891,
+  "elapsed_seconds": 1844.6724838319988,
+  "average_patch_bytes": 8.908382419574082,
+  "patch_coalesced_rate": 0.07291666666666667,
   "gate": 0.53125,
   "peak_vram_gib": 24.91581106185913,
   "documents": {
-    "HuggingFaceFW/fineweb-edu:sample-10BT": 85621,
-    "HuggingFaceTB/finemath:finemath-4plus": 30470,
-    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 18410,
-    "open-r1/OpenR1-Math-220k:default": 37507,
-    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 533
+    "HuggingFaceFW/fineweb-edu:sample-10BT": 86801,
+    "HuggingFaceTB/finemath:finemath-4plus": 30911,
+    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 18665,
+    "open-r1/OpenR1-Math-220k:default": 38052,
+    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 538
   },
   "source_bytes": {
-    "HuggingFaceFW/fineweb-edu:sample-10BT": 399685809,
-    "HuggingFaceTB/finemath:finemath-4plus": 151625272,
-    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 68914502,
-    "open-r1/OpenR1-Math-220k:default": 48237830,
-    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 20731801
-  },
-  "validation_loss": 1.0546875,
-  "checkpoint_started": true
+    "HuggingFaceFW/fineweb-edu:sample-10BT": 405420184,
+    "HuggingFaceTB/finemath:finemath-4plus": 153781212,
+    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 69899623,
+    "open-r1/OpenR1-Math-220k:default": 48932159,
+    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 21002254
+  }
 }