Upload training_state.json with huggingface_hub

training_state.json

@@ -1,6 +1,6 @@
 {
-  "step": 9900,
-  "tokens": 973209600,
+  "step": 10000,
+  "tokens": 983040000,
   "parameters": {
     "total": 998694784,
     "active": 340154240,
@@ -9,29 +9,31 @@
   },
   "batch_size": 96,
   "sequence_length": 1024,
-  "loss": 1.0625,
-  "auxiliary_loss": 1.7109375,
-  "gradient_norm": 0.2578125,
-  "route": 2,
-  "lr_scale": 0.9821759560471955,
-  "tokens_per_second": 456298.6179129821,
-  "elapsed_seconds": 1576.9391847389998,
-  "average_patch_bytes": 8.144490472245236,
+  "loss": 0.99609375,
+  "auxiliary_loss": 1.3828125,
+  "gradient_norm": 0.1494140625,
+  "route": 0,
+  "lr_scale": 0.9817758610406299,
+  "tokens_per_second": 455814.1952219926,
+  "elapsed_seconds": 1598.9329262580013,
+  "average_patch_bytes": 8.525932350390287,
   "patch_coalesced_rate": 0.11458333333333333,
   "gate": 0.53515625,
   "peak_vram_gib": 24.91581106185913,
   "documents": {
-    "HuggingFaceFW/fineweb-edu:sample-10BT": 72014,
-    "HuggingFaceTB/finemath:finemath-4plus": 25642,
-    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 15533,
-    "open-r1/OpenR1-Math-220k:default": 31623,
-    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 457
+    "HuggingFaceFW/fineweb-edu:sample-10BT": 73231,
+    "HuggingFaceTB/finemath:finemath-4plus": 26090,
+    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 15790,
+    "open-r1/OpenR1-Math-220k:default": 32147,
+    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 464
   },
   "source_bytes": {
-    "HuggingFaceFW/fineweb-edu:sample-10BT": 336934389,
-    "HuggingFaceTB/finemath:finemath-4plus": 127805349,
-    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 58094772,
-    "open-r1/OpenR1-Math-220k:default": 40664196,
-    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 17456161
-  }
+    "HuggingFaceFW/fineweb-edu:sample-10BT": 342662456,
+    "HuggingFaceTB/finemath:finemath-4plus": 129972402,
+    "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 59077727,
+    "open-r1/OpenR1-Math-220k:default": 41355928,
+    "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 17732336
+  },
+  "validation_loss": 1.064453125,
+  "checkpoint_started": true
 }