Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -1347,3 +1347,5 @@
 {"step": 26940, "tokens": 14124318720, "loss": 2.700437545776367, "tokens_per_second": 983736.9920812615, "lr": 0.0015, "gradient_norm": 0.2033543586730957, "benchmark": false}
 {"step": 26960, "tokens": 14134804480, "loss": 2.7299921035766603, "tokens_per_second": 970803.0368257932, "lr": 0.0015, "gradient_norm": 0.16589409112930298, "benchmark": false}
 {"step": 26980, "tokens": 14145290240, "loss": 2.757349967956543, "tokens_per_second": 964245.0973920216, "lr": 0.0015, "gradient_norm": 0.18727724254131317, "benchmark": false}
+{"step": 27000, "tokens": 14155776000, "loss": 2.7379180908203127, "tokens_per_second": 974215.3208308646, "lr": 0.0015, "gradient_norm": 0.2257988452911377, "benchmark": false}
+{"step": 27020, "tokens": 14166261760, "loss": 2.7172765731811523, "tokens_per_second": 811788.9719791115, "lr": 0.0015, "gradient_norm": 0.19278764724731445, "benchmark": false}

reports/run/validation.jsonl

@@ -24,3 +24,4 @@
 {"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}}
 {"step": 26000, "source_validation_loss": {"fineweb_edu": 2.893357038497925, "dclm_edu": 2.934877634048462, "cosmopedia": 1.9365743398666382, "finemath": 2.1286120414733887}}
+{"step": 27000, "source_validation_loss": {"fineweb_edu": 2.890822649002075, "dclm_edu": 2.9303057193756104, "cosmopedia": 1.936927318572998, "finemath": 2.1255106925964355}}