Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -1147,3 +1147,5 @@
 {"step": 22940, "tokens": 12027166720, "loss": 2.746495819091797, "tokens_per_second": 973142.2797433673, "lr": 0.0015, "gradient_norm": 0.1785355806350708, "benchmark": false}
 {"step": 22960, "tokens": 12037652480, "loss": 2.741334342956543, "tokens_per_second": 982407.4831120628, "lr": 0.0015, "gradient_norm": 0.16480481624603271, "benchmark": false}
 {"step": 22980, "tokens": 12048138240, "loss": 2.6979450225830077, "tokens_per_second": 983168.0924193689, "lr": 0.0015, "gradient_norm": 0.15850849449634552, "benchmark": false}
+{"step": 23000, "tokens": 12058624000, "loss": 2.7307880401611326, "tokens_per_second": 839554.3026868133, "lr": 0.0015, "gradient_norm": 0.24211981892585754, "benchmark": false}
+{"step": 23020, "tokens": 12069109760, "loss": 2.7809255599975584, "tokens_per_second": 823841.549096626, "lr": 0.0015, "gradient_norm": 0.1513972282409668, "benchmark": false}

reports/run/validation.jsonl

@@ -20,3 +20,4 @@
 {"step": 20000, "source_validation_loss": {"fineweb_edu": 2.914907693862915, "dclm_edu": 2.9553182125091553, "cosmopedia": 1.9704746007919312, "finemath": 2.1503968238830566}}
 {"step": 21000, "source_validation_loss": {"fineweb_edu": 2.9123802185058594, "dclm_edu": 2.9512526988983154, "cosmopedia": 1.9636504650115967, "finemath": 2.145289421081543}}
 {"step": 22000, "source_validation_loss": {"fineweb_edu": 2.906125783920288, "dclm_edu": 2.946148633956909, "cosmopedia": 1.962751865386963, "finemath": 2.1453704833984375}}
+{"step": 23000, "source_validation_loss": {"fineweb_edu": 2.9029648303985596, "dclm_edu": 2.9421579837799072, "cosmopedia": 1.955125093460083, "finemath": 2.1363375186920166}}