Update tokenizer, configuration and training reports

reports/run/metrics.jsonl

@@ -347,3 +347,5 @@
 {"step": 6940, "tokens": 3638558720, "loss": 2.9081472396850585, "tokens_per_second": 973242.0789233029, "lr": 0.0015, "gradient_norm": 0.073855459690094, "benchmark": false}
 {"step": 6960, "tokens": 3649044480, "loss": 2.8929702758789064, "tokens_per_second": 964825.9607954713, "lr": 0.0015, "gradient_norm": 0.06530334800481796, "benchmark": false}
 {"step": 6980, "tokens": 3659530240, "loss": 2.8522924423217773, "tokens_per_second": 963148.4756019333, "lr": 0.0015, "gradient_norm": 0.07306957989931107, "benchmark": false}
+{"step": 7000, "tokens": 3670016000, "loss": 2.865734100341797, "tokens_per_second": 967669.6781736239, "lr": 0.0015, "gradient_norm": 0.0665232315659523, "benchmark": false}
+{"step": 7020, "tokens": 3680501760, "loss": 2.886398506164551, "tokens_per_second": 811736.6275375811, "lr": 0.0015, "gradient_norm": 0.0693461000919342, "benchmark": false}

reports/run/validation.jsonl

@@ -4,3 +4,4 @@
 {"step": 4000, "source_validation_loss": {"fineweb_edu": 3.1462671756744385, "dclm_edu": 3.168818950653076, "cosmopedia": 2.2698538303375244, "finemath": 2.4097580909729004}}
 {"step": 5000, "source_validation_loss": {"fineweb_edu": 3.097052812576294, "dclm_edu": 3.1254994869232178, "cosmopedia": 2.2071025371551514, "finemath": 2.3563427925109863}}
 {"step": 6000, "source_validation_loss": {"fineweb_edu": 3.0642285346984863, "dclm_edu": 3.0997276306152344, "cosmopedia": 2.1647276878356934, "finemath": 2.31894588470459}}
+{"step": 7000, "source_validation_loss": {"fineweb_edu": 3.0427091121673584, "dclm_edu": 3.072791576385498, "cosmopedia": 2.1341612339019775, "finemath": 2.284400701522827}}