Upload 8k_v3/flexitok--bpe_ltr_swe_Latn_8000_v3_overlap.json with huggingface_hub

8k_v3/flexitok--bpe_ltr_swe_Latn_8000_v3_overlap.json

@@ -0,0 +1 @@
+{"1": {"ratio_to_total_tokens": 0.4234981745768337, "expected_training_ratio_in_superset": 0.020166579741753985, "num_tokens": 3828}, "2": {"ratio_to_total_tokens": 0.14747206549397057, "expected_training_ratio_in_superset": 0.014044958618473387, "num_tokens": 1333}, "3": {"ratio_to_total_tokens": 0.05100121694877752, "expected_training_ratio_in_superset": 0.0072858881355396445, "num_tokens": 461}, "4": {"ratio_to_total_tokens": 0.031530036508463324, "expected_training_ratio_in_superset": 0.0060057212397073, "num_tokens": 285}, "5": {"ratio_to_total_tokens": 0.02610908286314858, "expected_training_ratio_in_superset": 0.006216448300749662, "num_tokens": 236}, "6": {"ratio_to_total_tokens": 0.02068812921783383, "expected_training_ratio_in_superset": 0.005910894062238237, "num_tokens": 187}, "7": {"ratio_to_total_tokens": 0.019471180440314192, "expected_training_ratio_in_superset": 0.006490393480104731, "num_tokens": 176}, "8": {"ratio_to_total_tokens": 0.01648412435003872, "expected_training_ratio_in_superset": 0.00627966641906237, "num_tokens": 149}, "9": {"ratio_to_total_tokens": 0.015377807279566323, "expected_training_ratio_in_superset": 0.006590488834099853, "num_tokens": 139}, "10": {"ratio_to_total_tokens": 0.011837592654054652, "expected_training_ratio_in_superset": 0.005636948882883167, "num_tokens": 107}, "11": {"ratio_to_total_tokens": 0.01272264631043257, "expected_training_ratio_in_superset": 0.00666424330546468, "num_tokens": 115}, "12": {"ratio_to_total_tokens": 0.011284434118818453, "expected_training_ratio_in_superset": 0.006448248067896259, "num_tokens": 102}, "13": {"ratio_to_total_tokens": 0.011948224361101892, "expected_training_ratio_in_superset": 0.0073965198425868855, "num_tokens": 108}, "14": {"ratio_to_total_tokens": 0.01228011948224361, "expected_training_ratio_in_superset": 0.008186746321495741, "num_tokens": 111}, "15": {"ratio_to_total_tokens": 0.007301692665117823, "expected_training_ratio_in_superset": 0.005215494760798445, "num_tokens": 66}, "16": {"ratio_to_total_tokens": 0.005642217059409226, "expected_training_ratio_in_superset": 0.0042988320452641724, "num_tokens": 51}, "17": {"ratio_to_total_tokens": 0.004867795110078548, "expected_training_ratio_in_superset": 0.003940596041492159, "num_tokens": 44}, "18": {"ratio_to_total_tokens": 0.003761478039606151, "expected_training_ratio_in_superset": 0.0032241240339481296, "num_tokens": 34}, "19": {"ratio_to_total_tokens": 0.0028764243832282333, "expected_training_ratio_in_superset": 0.0026024792038731636, "num_tokens": 26}, "20": {"ratio_to_total_tokens": 0.0036508463325589115, "expected_training_ratio_in_superset": 0.003476996507198963, "num_tokens": 33}, "21": {"ratio_to_total_tokens": 0.16019471180440314, "expected_training_ratio_in_superset": 0.16019471180440314, "num_tokens": 1448}, "total_training_compared_to_full_model": 0.2962769796490341}