fix chart

pages/index.md

@@ -5,25 +5,29 @@ title: 🤗 Hub Stats
 _Note: The charts are updated daily via the <a class="font-bold text-blue" href="https://huggingface.co/datasets/cfahlgren1/hub-stats" target="_blank" rel="noopener noreferrer">hub-stats dataset</a>. It may take a little while for the data to load as the queries are running entirely in the browser._
 
 ```sql hub_growth
+-- Hub Growth by Month (using DATE_TRUNC for proper date ordering)
 WITH all_data AS (
   SELECT 
-    DATE_TRUNC('month', createdAt::TIMESTAMP) AS month, 
-    'model' AS repo 
-  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/models/train/0000.parquet?download=true')
-  
+    CAST(DATE_TRUNC('month', createdAt) AS DATE) AS month, 
+    'model'   AS repo 
+  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/models.parquet?download=true')
+  WHERE createdAt IS NOT NULL
+
   UNION ALL
-  
+
   SELECT 
-    DATE_TRUNC('month', createdAt::TIMESTAMP) AS month, 
+    CAST(DATE_TRUNC('month', createdAt) AS DATE) AS month, 
     'dataset' AS repo 
-  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/datasets/train/0000.parquet?download=true')
-  
+  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/datasets.parquet')
+  WHERE createdAt IS NOT NULL
+
   UNION ALL
-  
+
   SELECT 
-    DATE_TRUNC('month', createdAt::TIMESTAMP) AS month, 
-    'space' AS repo 
-  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/spaces/train/0000.parquet?download=true')
+    CAST(DATE_TRUNC('month', createdAt) AS DATE) AS month, 
+    'space'   AS repo 
+  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/spaces.parquet?download=true')
+  WHERE createdAt IS NOT NULL
 )
 SELECT
   month,
@@ -37,6 +41,7 @@ ORDER BY month, repo
 ## Cumulative Hub Growth
 
 ```sql cumulative_growth_by_repo
+-- Cumulative Hub Growth
 WITH ordered_data AS (
   SELECT
     month,
@@ -75,9 +80,14 @@ ORDER BY month, repo
 ## Models Created Each Month
 
 ```sql model_creations_by_month
-SELECT month, repo, creations
+-- Models Created Each Month
+SELECT
+  month,
+  repo,
+  creations
 FROM ${hub_growth}
 WHERE repo = 'model'
+ORDER BY month;
 ```
 
 <AreaChart 
@@ -91,9 +101,14 @@ WHERE repo = 'model'
 ## Datasets Created Each Month
 
 ```sql dataset_creations_by_month
-SELECT month, repo, creations
+-- Datasets Created Each Month
+SELECT
+  month,
+  repo,
+  creations
 FROM ${hub_growth}
 WHERE repo = 'dataset'
+ORDER BY month;
 ```
 <AreaChart 
     data={dataset_creations_by_month}
@@ -106,9 +121,14 @@ WHERE repo = 'dataset'
 ## Spaces Created Each Month
 
 ```sql space_creations_by_month
-SELECT month, repo, creations
+-- Spaces Created Each Month
+SELECT
+  month,
+  repo,
+  creations
 FROM ${hub_growth}
 WHERE repo = 'space'
+ORDER BY month;
 ```
 
 <AreaChart 
@@ -123,10 +143,11 @@ WHERE repo = 'space'
 
 ```sql model_pipeline_downloads
 SELECT
-pipeline_tag AS name, SUM(downloads) AS value
-FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/models/train/0000.parquet?download=true')
+    pipeline_tag AS name,
+    SUM(downloads) AS value
+FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/models.parquet?download=true')
 GROUP BY pipeline_tag
-ORDER BY value DESC;
+ORDER BY value DESC
 ```
 
 <DataTable data={model_pipeline_downloads} search=true>
@@ -159,10 +180,11 @@ ORDER BY value DESC;
 
 ```sql model_library_downloads
 SELECT
-library_name AS name, SUM(downloads) AS value
-FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/models/train/0000.parquet?download=true')
+    library_name AS name,
+    SUM(downloads) AS value
+FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/models.parquet?download=true')
 GROUP BY library_name
-ORDER BY value DESC;
+ORDER BY value DESC
 ```
 
 <DataTable data={model_library_downloads} search=true>
@@ -194,11 +216,13 @@ ORDER BY value DESC;
 # Model Licenses
 
 ```sql model_license_ratio
-  SELECT SUBSTRING(name, 9) AS name, COUNT(*) as value
-  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/models/train/0000.parquet?download=true'),
-   UNNEST(tags) AS t(name)
-  WHERE name LIKE 'license:%'
-  GROUP BY name;
+SELECT
+    SUBSTRING(name, 9) AS name,
+    COUNT(*) as value
+FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/models.parquet?download=true'),
+    UNNEST(tags) AS t(name)
+WHERE name LIKE 'license:%'
+GROUP BY name
 ```
 
 <ECharts config={
@@ -220,11 +244,13 @@ ORDER BY value DESC;
 # Dataset Licenses
 
 ```sql dataset_license_ratio
-  SELECT SUBSTRING(name, 9) AS name, COUNT(*) as value
-  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/datasets/train/0000.parquet?download=true'),
-  UNNEST(tags) AS t(name)
-  WHERE name LIKE 'license:%'
-  GROUP BY name;
+SELECT
+    SUBSTRING(name, 9) AS name,
+    COUNT(*) as value
+FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/datasets.parquet?download=true'),
+    UNNEST(tags) AS t(name)
+WHERE name LIKE 'license:%'
+GROUP BY name
 ```
 
 <ECharts config={
@@ -246,9 +272,11 @@ ORDER BY value DESC;
 # Space SDKs
 
 ```sql space_sdk_ratio
-  SELECT sdk as name, COUNT(*) as value
-  FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/refs%2Fconvert%2Fparquet/spaces/train/0000.parquet?download=true')
-  GROUP BY name;
+SELECT
+    sdk as name,
+    COUNT(*) as value
+FROM read_parquet('https://huggingface.co/datasets/cfahlgren1/hub-stats/resolve/main/spaces.parquet?download=true')
+GROUP BY name
 ```