Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -547,3 +547,5 @@
 {"step": 10940, "tokens": 5735710720, "loss": 2.8052703857421877, "tokens_per_second": 980657.4605480039, "lr": 0.0015, "gradient_norm": 0.0964113250374794, "benchmark": false}
 {"step": 10960, "tokens": 5746196480, "loss": 2.814346504211426, "tokens_per_second": 977156.1872256437, "lr": 0.0015, "gradient_norm": 0.08051484823226929, "benchmark": false}
 {"step": 10980, "tokens": 5756682240, "loss": 2.8020746231079103, "tokens_per_second": 977162.6497957591, "lr": 0.0015, "gradient_norm": 0.11768177151679993, "benchmark": false}
+{"step": 11000, "tokens": 5767168000, "loss": 2.8348876953125, "tokens_per_second": 987974.7411373939, "lr": 0.0015, "gradient_norm": 0.09453051537275314, "benchmark": false}
+{"step": 11020, "tokens": 5777653760, "loss": 2.8067771911621096, "tokens_per_second": 768149.5252021329, "lr": 0.0015, "gradient_norm": 0.09996025264263153, "benchmark": false}

reports/run/validation.jsonl

@@ -8,3 +8,4 @@
 {"step": 8000, "source_validation_loss": {"fineweb_edu": 3.022184133529663, "dclm_edu": 3.0521011352539062, "cosmopedia": 2.1067497730255127, "finemath": 2.2722864151000977}}
 {"step": 9000, "source_validation_loss": {"fineweb_edu": 3.0031962394714355, "dclm_edu": 3.0357918739318848, "cosmopedia": 2.0862884521484375, "finemath": 2.2452752590179443}}
 {"step": 10000, "source_validation_loss": {"fineweb_edu": 2.9892542362213135, "dclm_edu": 3.0245914459228516, "cosmopedia": 2.067286968231201, "finemath": 2.238250970840454}}
+{"step": 11000, "source_validation_loss": {"fineweb_edu": 2.9761719703674316, "dclm_edu": 3.0146288871765137, "cosmopedia": 2.05277943611145, "finemath": 2.2114815711975098}}