Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -296,3 +296,5 @@
 {"step": 5920, "tokens": 3103784960, "loss": 2.9021526336669923, "tokens_per_second": 986599.4897803419, "lr": 0.0015, "gradient_norm": 0.0749964788556099, "benchmark": false}
 {"step": 5940, "tokens": 3114270720, "loss": 2.9089988708496093, "tokens_per_second": 978099.6904409037, "lr": 0.0015, "gradient_norm": 0.08997740596532822, "benchmark": false}
 {"step": 5960, "tokens": 3124756480, "loss": 2.9387432098388673, "tokens_per_second": 978047.3818877105, "lr": 0.0015, "gradient_norm": 0.0761580765247345, "benchmark": false}
+{"step": 5980, "tokens": 3135242240, "loss": 2.8938913345336914, "tokens_per_second": 971113.1983575885, "lr": 0.0015, "gradient_norm": 0.07130406051874161, "benchmark": false}
+{"step": 6000, "tokens": 3145728000, "loss": 2.910635757446289, "tokens_per_second": 968513.8750935235, "lr": 0.0015, "gradient_norm": 0.07274629175662994, "benchmark": false}

reports/run/validation.jsonl

@@ -3,3 +3,4 @@
 {"step": 3000, "source_validation_loss": {"fineweb_edu": 3.2158007621765137, "dclm_edu": 3.24115252494812, "cosmopedia": 2.362274169921875, "finemath": 2.494218349456787}}
 {"step": 4000, "source_validation_loss": {"fineweb_edu": 3.1462671756744385, "dclm_edu": 3.168818950653076, "cosmopedia": 2.2698538303375244, "finemath": 2.4097580909729004}}
 {"step": 5000, "source_validation_loss": {"fineweb_edu": 3.097052812576294, "dclm_edu": 3.1254994869232178, "cosmopedia": 2.2071025371551514, "finemath": 2.3563427925109863}}
+{"step": 6000, "source_validation_loss": {"fineweb_edu": 3.0642285346984863, "dclm_edu": 3.0997276306152344, "cosmopedia": 2.1647276878356934, "finemath": 2.31894588470459}}