Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -1296,3 +1296,5 @@
 {"step": 25920, "tokens": 13589544960, "loss": 2.7093645095825196, "tokens_per_second": 850106.0294220387, "lr": 0.0015, "gradient_norm": 0.199713796377182, "benchmark": false}
 {"step": 25940, "tokens": 13600030720, "loss": 2.7047935485839845, "tokens_per_second": 987127.2054923556, "lr": 0.0015, "gradient_norm": 0.17227420210838318, "benchmark": false}
 {"step": 25960, "tokens": 13610516480, "loss": 2.708211135864258, "tokens_per_second": 854841.7415926263, "lr": 0.0015, "gradient_norm": 0.17577874660491943, "benchmark": false}
+{"step": 25980, "tokens": 13621002240, "loss": 2.712994384765625, "tokens_per_second": 973681.908757307, "lr": 0.0015, "gradient_norm": 0.1609300822019577, "benchmark": false}
+{"step": 26000, "tokens": 13631488000, "loss": 2.724579620361328, "tokens_per_second": 967542.1501508587, "lr": 0.0015, "gradient_norm": 0.1633884310722351, "benchmark": false}

reports/run/validation.jsonl

@@ -23,3 +23,4 @@
 {"step": 23000, "source_validation_loss": {"fineweb_edu": 2.9029648303985596, "dclm_edu": 2.9421579837799072, "cosmopedia": 1.955125093460083, "finemath": 2.1363375186920166}}
 {"step": 24000, "source_validation_loss": {"fineweb_edu": 2.9004201889038086, "dclm_edu": 2.94045090675354, "cosmopedia": 1.9505376815795898, "finemath": 2.12982439994812}}
 {"step": 25000, "source_validation_loss": {"fineweb_edu": 2.8956382274627686, "dclm_edu": 2.936393976211548, "cosmopedia": 1.9424511194229126, "finemath": 2.1288459300994873}}
+{"step": 26000, "source_validation_loss": {"fineweb_edu": 2.893357038497925, "dclm_edu": 2.934877634048462, "cosmopedia": 1.9365743398666382, "finemath": 2.1286120414733887}}