Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -1246,3 +1246,5 @@
 {"step": 24920, "tokens": 13065256960, "loss": 2.7452892303466796, "tokens_per_second": 960075.2523763671, "lr": 0.0015, "gradient_norm": 0.16753168404102325, "benchmark": false}
 {"step": 24940, "tokens": 13075742720, "loss": 2.724387359619141, "tokens_per_second": 991013.5827467868, "lr": 0.0015, "gradient_norm": 0.16112184524536133, "benchmark": false}
 {"step": 24960, "tokens": 13086228480, "loss": 2.6991626739501955, "tokens_per_second": 939572.5153784158, "lr": 0.0015, "gradient_norm": 0.1650470644235611, "benchmark": false}
+{"step": 24980, "tokens": 13096714240, "loss": 2.714500617980957, "tokens_per_second": 984047.7388723607, "lr": 0.0015, "gradient_norm": 0.1731661707162857, "benchmark": false}
+{"step": 25000, "tokens": 13107200000, "loss": 2.7156171798706055, "tokens_per_second": 999762.4758471423, "lr": 0.0015, "gradient_norm": 0.17807134985923767, "benchmark": false}

reports/run/validation.jsonl

@@ -22,3 +22,4 @@
 {"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}}
 {"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}}