Spaces:
Running
Running
more examples
Browse files
app.py
CHANGED
|
@@ -4,12 +4,14 @@ import gradio as gr
|
|
| 4 |
con = duckdb.connect(":memory:")
|
| 5 |
|
| 6 |
def greet(SQL_Query):
|
| 7 |
-
con.sql("CREATE TABLE IF NOT EXISTS fineweb_10k_samples AS SELECT * FROM 'hf://datasets/HuggingFaceFW/fineweb/sample/10BT/*.parquet' LIMIT 10000;")
|
| 8 |
return con.sql(SQL_Query).df()
|
| 9 |
|
| 10 |
examples = [
|
| 11 |
-
"SELECT dump, avg(token_count) FROM
|
|
|
|
|
|
|
| 12 |
]
|
|
|
|
| 13 |
description = "Run SQL queries on the FineWeb dataset"
|
| 14 |
-
demo = gr.Interface(fn=greet, inputs="text", outputs="dataframe", examples=examples, cache_examples=False, description=description)
|
| 15 |
demo.launch()
|
|
|
|
| 4 |
con = duckdb.connect(":memory:")
|
| 5 |
|
| 6 |
def greet(SQL_Query):
|
|
|
|
| 7 |
return con.sql(SQL_Query).df()
|
| 8 |
|
| 9 |
examples = [
|
| 10 |
+
"SELECT dump, avg(token_count) FROM\n(SELECT * FROM 'hf://datasets/HuggingFaceFW/fineweb/sample/10BT/*.parquet' LIMIT 10000)\nGROUP BY dump;",
|
| 11 |
+
"SELECT dump, max(token_count) FROM\n(SELECT * FROM 'hf://datasets/HuggingFaceFW/fineweb/sample/10BT/*.parquet' LIMIT 10000)\nGROUP BY dump;",
|
| 12 |
+
"SELECT dump, min(token_count) FROM\n(SELECT * FROM 'hf://datasets/HuggingFaceFW/fineweb/sample/10BT/*.parquet' LIMIT 10000)\nGROUP BY dump;",
|
| 13 |
]
|
| 14 |
+
css = "#component-4{display: block;}"
|
| 15 |
description = "Run SQL queries on the FineWeb dataset"
|
| 16 |
+
demo = gr.Interface(fn=greet, inputs="text", outputs="dataframe", examples=examples, cache_examples=False, description=description, css=css)
|
| 17 |
demo.launch()
|