Upload script_1/flexitok--bpe_script_Slav_16000_overlap.json with huggingface_hub

script_1/flexitok--bpe_script_Slav_16000_overlap.json

@@ -0,0 +1 @@
+{"1": {"ratio_to_total_tokens": 0.7000825082508251, "expected_training_ratio_in_superset": 0.06364386438643865, "num_tokens": 11879}, "2": {"ratio_to_total_tokens": 0.04667609618104668, "expected_training_ratio_in_superset": 0.008486562942008486, "num_tokens": 792}, "3": {"ratio_to_total_tokens": 0.026638378123526638, "expected_training_ratio_in_superset": 0.007265012215507264, "num_tokens": 452}, "4": {"ratio_to_total_tokens": 0.02775813295615276, "expected_training_ratio_in_superset": 0.010093866529510094, "num_tokens": 471}, "5": {"ratio_to_total_tokens": 0.03606789250353607, "expected_training_ratio_in_superset": 0.016394496592516395, "num_tokens": 612}, "6": {"ratio_to_total_tokens": 0.024811409712399812, "expected_training_ratio_in_superset": 0.013533496206763532, "num_tokens": 421}, "7": {"ratio_to_total_tokens": 0.02215935879302216, "expected_training_ratio_in_superset": 0.014101410141014101, "num_tokens": 376}, "8": {"ratio_to_total_tokens": 0.011786892975011787, "expected_training_ratio_in_superset": 0.008572285800008572, "num_tokens": 200}, "9": {"ratio_to_total_tokens": 0.0054219707685054215, "expected_training_ratio_in_superset": 0.004436157901504437, "num_tokens": 92}, "10": {"ratio_to_total_tokens": 0.00442008486562942, "expected_training_ratio_in_superset": 0.004018258968754017, "num_tokens": 75}, "11": {"ratio_to_total_tokens": 0.09417727487034418, "expected_training_ratio_in_superset": 0.09417727487034418, "num_tokens": 1598}, "total_training_compared_to_full_model": 0.24472268655436974}