Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -897,3 +897,5 @@
 {"step": 17940, "tokens": 9405726720, "loss": 2.742411422729492, "tokens_per_second": 993852.2885264372, "lr": 0.0015, "gradient_norm": 0.12873972952365875, "benchmark": false}
 {"step": 17960, "tokens": 9416212480, "loss": 2.7404382705688475, "tokens_per_second": 979842.152883283, "lr": 0.0015, "gradient_norm": 0.14923682808876038, "benchmark": false}
 {"step": 17980, "tokens": 9426698240, "loss": 2.7674840927124023, "tokens_per_second": 984205.3643317616, "lr": 0.0015, "gradient_norm": 0.1542074829339981, "benchmark": false}
+{"step": 18000, "tokens": 9437184000, "loss": 2.7930335998535156, "tokens_per_second": 990893.5270222354, "lr": 0.0015, "gradient_norm": 0.14015617966651917, "benchmark": false}
+{"step": 18020, "tokens": 9447669760, "loss": 2.758540153503418, "tokens_per_second": 818144.637390868, "lr": 0.0015, "gradient_norm": 0.13568595051765442, "benchmark": false}

reports/run/validation.jsonl

@@ -15,3 +15,4 @@
 {"step": 15000, "source_validation_loss": {"fineweb_edu": 2.9435317516326904, "dclm_edu": 2.9802112579345703, "cosmopedia": 2.0042612552642822, "finemath": 2.18125581741333}}
 {"step": 16000, "source_validation_loss": {"fineweb_edu": 2.9361798763275146, "dclm_edu": 2.9739556312561035, "cosmopedia": 1.9942559003829956, "finemath": 2.171142578125}}
 {"step": 17000, "source_validation_loss": {"fineweb_edu": 2.931060791015625, "dclm_edu": 2.966684341430664, "cosmopedia": 1.9891211986541748, "finemath": 2.1641457080841064}}
+{"step": 18000, "source_validation_loss": {"fineweb_edu": 2.923886775970459, "dclm_edu": 2.9643118381500244, "cosmopedia": 1.983628511428833, "finemath": 2.1608521938323975}}