Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -746,3 +746,5 @@
 {"step": 14920, "tokens": 7822376960, "loss": 2.771735191345215, "tokens_per_second": 991036.9703833835, "lr": 0.0015, "gradient_norm": 0.12381613999605179, "benchmark": false}
 {"step": 14940, "tokens": 7832862720, "loss": 2.7973310470581056, "tokens_per_second": 982117.8814135678, "lr": 0.0015, "gradient_norm": 0.13259257376194, "benchmark": false}
 {"step": 14960, "tokens": 7843348480, "loss": 2.8001480102539062, "tokens_per_second": 989712.792759918, "lr": 0.0015, "gradient_norm": 0.12138649076223373, "benchmark": false}
+{"step": 14980, "tokens": 7853834240, "loss": 2.7922641754150392, "tokens_per_second": 990706.2731206587, "lr": 0.0015, "gradient_norm": 0.11834993213415146, "benchmark": false}
+{"step": 15000, "tokens": 7864320000, "loss": 2.810902976989746, "tokens_per_second": 985215.5444990908, "lr": 0.0015, "gradient_norm": 0.13044802844524384, "benchmark": false}

reports/run/validation.jsonl

@@ -12,3 +12,4 @@
 {"step": 12000, "source_validation_loss": {"fineweb_edu": 2.9684195518493652, "dclm_edu": 3.0029680728912354, "cosmopedia": 2.0343093872070312, "finemath": 2.201632261276245}}
 {"step": 13000, "source_validation_loss": {"fineweb_edu": 2.9597086906433105, "dclm_edu": 2.994127035140991, "cosmopedia": 2.02358341217041, "finemath": 2.192626714706421}}
 {"step": 14000, "source_validation_loss": {"fineweb_edu": 2.948702096939087, "dclm_edu": 2.9865922927856445, "cosmopedia": 2.0160582065582275, "finemath": 2.18440580368042}}
+{"step": 15000, "source_validation_loss": {"fineweb_edu": 2.9435317516326904, "dclm_edu": 2.9802112579345703, "cosmopedia": 2.0042612552642822, "finemath": 2.18125581741333}}