| {"data": {"dedup_minhash_independent_output_CC-MAIN-2021-04": {"x": ["2021-04"], "y": [4.49282993950334e-06], "label": "2021-04", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-10": {"x": ["2021-10"], "y": [4.4320506429770015e-06], "label": "2021-10", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-17": {"x": ["2021-17"], "y": [4.525422769007163e-06], "label": "2021-17", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-21": {"x": ["2021-21"], "y": [4.337506002649248e-06], "label": "2021-21", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-25": {"x": ["2021-25"], "y": [4.31209874426111e-06], "label": "2021-25", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-31": {"x": ["2021-31"], "y": [4.549984757361091e-06], "label": "2021-31", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-39": {"x": ["2021-39"], "y": [4.476627474039729e-06], "label": "2021-39", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-43": {"x": ["2021-43"], "y": [4.515299780479149e-06], "label": "2021-43", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2021-49": {"x": ["2021-49"], "y": [3.711316662237467e-06], "label": "2021-49", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2022-05": {"x": ["2022-05"], "y": [3.833222545464518e-06], "label": "2022-05", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2022-21": {"x": ["2022-21"], "y": [4.033461274185702e-06], "label": "2022-21", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2022-27": {"x": ["2022-27"], "y": [3.977774743517952e-06], "label": "2022-27", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2022-33": {"x": ["2022-33"], "y": [3.735864798219777e-06], "label": "2022-33", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2022-40": {"x": ["2022-40"], "y": [4.000056932617072e-06], "label": "2022-40", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2022-49": {"x": ["2022-49"], "y": [4.073748029584546e-06], "label": "2022-49", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2023-06": {"x": ["2023-06"], "y": [4.084332699102002e-06], "label": "2023-06", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2023-14": {"x": ["2023-14"], "y": [4.092708480795626e-06], "label": "2023-14", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2023-23": {"x": ["2023-23"], "y": [4.986546453643025e-06], "label": "2023-23", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2023-40": {"x": ["2023-40"], "y": [8.10525888274716e-06], "label": "2023-40", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2023-50": {"x": ["2023-50"], "y": [1.1324709237817275e-05], "label": "2023-50", "color": "blue 1.0"}, "dedup_minhash_independent_output_CC-MAIN-2024-10": {"x": ["2024-10"], "y": [1.645258450134167e-05], "label": "2024-10", "color": "blue 1.0"}, "dedup_minhash_independent_output_new_CC-MAIN-2024-18": {"x": ["2024-18"], "y": [2.1801896395395062e-05], "label": "2024-18", "color": "blue 1.0"}}, "traces": [{"x": ["2021-04", "2021-10", "2021-17", "2021-21", "2021-25", "2021-31", "2021-39", "2021-43", "2021-49", "2022-05", "2022-21", "2022-27", "2022-33", "2022-40", "2022-49", "2023-06", "2023-14", "2023-23", "2023-40", "2023-50", "2024-10", "2024-18"], "y": [0.4303327426314354, 0.4285315126180649, 0.43032742974658805, 0.42827240626017254, 0.42655438867708045, 0.43065756373107433, 0.42699132797618705, 0.432743809496363, 0.426107710848252, 0.4314096675564845, 0.4288001439223687, 0.43330975994467735, 0.4283793264379104, 0.43007233180105686, 0.43284351378679276, 0.43421785968045395, 0.4327918446312348, 0.43310930766165257, 0.4327431519826253, 0.4331989834705989, 0.43582472391426563, 0.4374971886475881], "type": "scattergl", "mode": "lines+markers", "name": "Dump Score", "yaxis": "y2", "line": {"color": "black 1.0"}, "marker": {"size": 10, "color": "black 0.5", "line": {"color": "black 1.0", "width": 2}}}], "layout": {"showlegend": false, "title": {"text": "Synthetic Data Contamination"}, "margin": {"b": 80}, "xaxis": {"title": {"text": "Year", "standoff": 20}, "type": "category"}, "yaxis": {"title": {"text": "Synthetic proxy Words Ratio"}, "side": "right"}, "shapes": [{"type": "line", "x0": 13.92, "y0": 0.424, "x1": 13.92, "y1": 0.439, "xref": "x", "yref": "y2", "line": {"color": "red 1.0", "width": 2, "dash": "dashdot"}}], "annotations": [{"text": "Chat-GPT Release", "x": 12.3, "y": 0.4357, "xref": "x", "yref": "y2", "showarrow": false}], "yaxis2": {"title": {"text": "Aggregate Score"}, "overlaying": "y", "side": "left"}}} |