nicer bars

data/plots/edu-100k/arc_acc_norm.json

@@ -1 +1 @@
-{"data": {"C4": {"x": ["C4"], "y": [0.4390000104904175], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.4329999983310699], "label": "Dolma"}, "FineWeb (ours)": {"x": ["FineWeb (ours)"], "y": [0.4404999911785126], "label": "FineWeb (ours)"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.4395000040531158], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.4375], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.4514999985694885], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.4090000092983246], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.5475000143051147], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "FineWeb-Edu at 100k steps"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "margin": {"b": 100}}}
+{"data": {"C4": {"x": ["C4"], "y": [0.4390000104904175], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.4329999983310699], "label": "Dolma"}, "FineWeb": {"x": ["FineWeb"], "y": [0.4404999911785126], "label": "FineWeb"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.4395000040531158], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.4375], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.4514999985694885], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.4090000092983246], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.5475000143051147], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "Evaluation results at 350B tokens"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "yaxis": {"range": [0.25, 0.6070000171661377]}, "margin": {"b": 100}}}

data/plots/edu-100k/hellaswag_acc_norm.json

@@ -1 +1 @@
-{"data": {"C4": {"x": ["C4"], "y": [0.6010000109672546], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.5879999995231628], "label": "Dolma"}, "FineWeb (ours)": {"x": ["FineWeb (ours)"], "y": [0.6060000061988831], "label": "FineWeb (ours)"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.5440000295639038], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.5709999799728394], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.5649999976158142], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.515999972820282], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.5830000042915344], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "FineWeb-Edu at 100k steps"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "margin": {"b": 100}}}
+{"data": {"C4": {"x": ["C4"], "y": [0.6010000109672546], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.5879999995231628], "label": "Dolma"}, "FineWeb": {"x": ["FineWeb"], "y": [0.6060000061988831], "label": "FineWeb"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.5440000295639038], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.5709999799728394], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.5649999976158142], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.515999972820282], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.5830000042915344], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "Evaluation results at 350B tokens"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "yaxis": {"range": [0.5, 0.6272000074386597]}, "margin": {"b": 100}}}

data/plots/edu-100k/mmlu_acc_norm.json

@@ -1 +1 @@
-{"data": {"C4": {"x": ["C4"], "y": [0.3100601136684418], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.3082475662231445], "label": "Dolma"}, "FineWeb (ours)": {"x": ["FineWeb (ours)"], "y": [0.3212694227695465], "label": "FineWeb (ours)"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.3182428181171417], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.3192791938781738], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.3266949653625488], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.3129254281520843], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.3591959178447723], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "FineWeb-Edu at 100k steps"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "margin": {"b": 100}}}
+{"data": {"C4": {"x": ["C4"], "y": [0.3100601136684418], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.3082475662231445], "label": "Dolma"}, "FineWeb": {"x": ["FineWeb"], "y": [0.3212694227695465], "label": "FineWeb"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.3182428181171417], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.3192791938781738], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.3266949653625488], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.3129254281520843], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.3591959178447723], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "Evaluation results at 350B tokens"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "yaxis": {"range": [0.25, 0.38103510141372676]}, "margin": {"b": 100}}}

data/plots/edu-100k/openbookqa_acc_norm.json

@@ -1 +1 @@
-{"data": {"C4": {"x": ["C4"], "y": [0.3740000128746032], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.3499999940395355], "label": "Dolma"}, "FineWeb (ours)": {"x": ["FineWeb (ours)"], "y": [0.3680000007152557], "label": "FineWeb (ours)"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.3339999914169311], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.3459999859333038], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.3499999940395355], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.3339999914169311], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.4000000059604645], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "FineWeb-Edu at 100k steps"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "margin": {"b": 100}}}
+{"data": {"C4": {"x": ["C4"], "y": [0.3740000128746032], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.3499999940395355], "label": "Dolma"}, "FineWeb": {"x": ["FineWeb"], "y": [0.3680000007152557], "label": "FineWeb"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.3339999914169311], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.3459999859333038], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.3499999940395355], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.3339999914169311], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.4000000059604645], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "Evaluation results at 350B tokens"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "yaxis": {"range": [0.25, 0.4300000071525574]}, "margin": {"b": 100}}}

data/plots/edu-100k/piqa_acc_norm.json

@@ -1 +1 @@
-{"data": {"C4": {"x": ["C4"], "y": [0.7739999890327454], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.7549999952316284], "label": "Dolma"}, "FineWeb (ours)": {"x": ["FineWeb (ours)"], "y": [0.7580000162124634], "label": "FineWeb (ours)"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.7239999771118164], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.7710000276565552], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.7379999756813049], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.7179999947547913], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.7590000033378601], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "FineWeb-Edu at 100k steps"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "margin": {"b": 100}}}
+{"data": {"C4": {"x": ["C4"], "y": [0.7739999890327454], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.7549999952316284], "label": "Dolma"}, "FineWeb": {"x": ["FineWeb"], "y": [0.7580000162124634], "label": "FineWeb"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.7239999771118164], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.7710000276565552], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.7379999756813049], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.7179999947547913], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.7590000033378601], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "Evaluation results at 350B tokens"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "yaxis": {"range": [0.5, 0.8287999868392945]}, "margin": {"b": 100}}}

data/plots/edu-100k/winogrande_acc_norm.json

@@ -1 +1 @@
-{"data": {"C4": {"x": ["C4"], "y": [0.5490000247955322], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.550000011920929], "label": "Dolma"}, "FineWeb (ours)": {"x": ["FineWeb (ours)"], "y": [0.5559999942779541], "label": "FineWeb (ours)"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.5509999990463257], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.5680000185966492], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.5519999861717224], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.5350000262260437], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.5720000267028809], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "FineWeb-Edu at 100k steps"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "margin": {"b": 100}}}
+{"data": {"C4": {"x": ["C4"], "y": [0.5490000247955322], "label": "C4"}, "Dolma": {"x": ["Dolma"], "y": [0.550000011920929], "label": "Dolma"}, "FineWeb": {"x": ["FineWeb"], "y": [0.5559999942779541], "label": "FineWeb"}, "RedPajama2": {"x": ["RedPajama2"], "y": [0.5509999990463257], "label": "RedPajama2"}, "RefinedWeb": {"x": ["RefinedWeb"], "y": [0.5680000185966492], "label": "RefinedWeb"}, "SlimPajama": {"x": ["SlimPajama"], "y": [0.5519999861717224], "label": "SlimPajama"}, "The Pile": {"x": ["The Pile"], "y": [0.5350000262260437], "label": "The Pile"}, "FineWeb-Edu": {"x": ["FineWeb-Edu"], "y": [0.5720000267028809], "label": "FineWeb-Edu"}}, "layout": {"showlegend": false, "title": {"text": "Evaluation results at 350B tokens"}, "xaxis": {"title": {"text": "Dataset", "standoff": 30}, "tickangle": 30}, "yaxis": {"range": [0.5, 0.586400032043457]}, "margin": {"b": 100}}}

src/clusters.js

@@ -230,7 +230,6 @@ async function plotClusters() {
                 parent,
                 { "marker.size": BASE_SIZE * zoomLevel },
                 { annotations: relevant_annotations },
-                relevantDataIdx
             );
         }
         // Zoom reset