Spaces:
Runtime error
Runtime error
Caleb Fahlgren
commited on
Commit
·
9fc2d21
1
Parent(s):
4ed656e
add examples and handle incorrect label / data key output better
Browse files
README.md
CHANGED
|
@@ -22,3 +22,5 @@ Powered by [Hermes-2-Pro-Llama-3-8B](https://huggingface.co/NousResearch/Hermes-
|
|
| 22 |
2. Get SQL DDL from Parquet column and data types
|
| 23 |
3. Prompt the LLM for SQL
|
| 24 |
4. Run the SQL against the Dataset Parquet
|
|
|
|
|
|
|
|
|
| 22 |
2. Get SQL DDL from Parquet column and data types
|
| 23 |
3. Prompt the LLM for SQL
|
| 24 |
4. Run the SQL against the Dataset Parquet
|
| 25 |
+
|
| 26 |
+
Inspired by [Datasets-Text2SQL](https://huggingface.co/spaces/asoria/datasets-text2sql)
|
app.py
CHANGED
|
@@ -46,10 +46,12 @@ class SQLResponse(BaseModel):
|
|
| 46 |
None, description="The type of visualization to display"
|
| 47 |
)
|
| 48 |
data_key: Optional[str] = Field(
|
| 49 |
-
None,
|
|
|
|
| 50 |
)
|
| 51 |
label_key: Optional[str] = Field(
|
| 52 |
-
None,
|
|
|
|
| 53 |
)
|
| 54 |
|
| 55 |
|
|
@@ -95,6 +97,9 @@ def generate_query(dataset_id: str, query: str) -> str:
|
|
| 95 |
```
|
| 96 |
|
| 97 |
Please assist the user by writing a SQL query that answers the user's question.
|
|
|
|
|
|
|
|
|
|
| 98 |
"""
|
| 99 |
|
| 100 |
print("Calling LLM with system prompt: ", system_prompt)
|
|
@@ -124,6 +129,12 @@ def query_dataset(dataset_id: str, query: str) -> Tuple[pd.DataFrame, str, plt.F
|
|
| 124 |
|
| 125 |
plot = None
|
| 126 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 127 |
if response.visualization_type == OutputTypes.LINECHART:
|
| 128 |
plot = df.plot(
|
| 129 |
kind="line", x=response.label_key, y=response.data_key
|
|
@@ -150,6 +161,13 @@ with gr.Blocks() as demo:
|
|
| 150 |
value="gretelai/synthetic_text_to_sql",
|
| 151 |
)
|
| 152 |
user_query = gr.Textbox("", label="Ask anything...")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 153 |
|
| 154 |
btn = gr.Button("Ask 🪄")
|
| 155 |
|
|
|
|
| 46 |
None, description="The type of visualization to display"
|
| 47 |
)
|
| 48 |
data_key: Optional[str] = Field(
|
| 49 |
+
None,
|
| 50 |
+
description="The column name from the sql query that contains the data for chart responses",
|
| 51 |
)
|
| 52 |
label_key: Optional[str] = Field(
|
| 53 |
+
None,
|
| 54 |
+
description="The column name from the sql query that contains the labels for chart responses",
|
| 55 |
)
|
| 56 |
|
| 57 |
|
|
|
|
| 97 |
```
|
| 98 |
|
| 99 |
Please assist the user by writing a SQL query that answers the user's question.
|
| 100 |
+
|
| 101 |
+
Use Label Key as the column name for the x-axis and Data Key as the column name for the y-axis for chart responses. The
|
| 102 |
+
label key and data key must be present in the SQL output.
|
| 103 |
"""
|
| 104 |
|
| 105 |
print("Calling LLM with system prompt: ", system_prompt)
|
|
|
|
| 129 |
|
| 130 |
plot = None
|
| 131 |
|
| 132 |
+
# handle incorrect data and label keys better
|
| 133 |
+
if response.label_key and response.label_key not in df.columns:
|
| 134 |
+
response.label_key = None
|
| 135 |
+
if response.data_key and response.data_key not in df.columns:
|
| 136 |
+
response.data_key = None
|
| 137 |
+
|
| 138 |
if response.visualization_type == OutputTypes.LINECHART:
|
| 139 |
plot = df.plot(
|
| 140 |
kind="line", x=response.label_key, y=response.data_key
|
|
|
|
| 161 |
value="gretelai/synthetic_text_to_sql",
|
| 162 |
)
|
| 163 |
user_query = gr.Textbox("", label="Ask anything...")
|
| 164 |
+
examples = [
|
| 165 |
+
["Show me a preview of the data"],
|
| 166 |
+
["Show me something interesting"],
|
| 167 |
+
["What is the largest length of sql query context?"],
|
| 168 |
+
["show me counts by sql_query_type in a bar chart"],
|
| 169 |
+
]
|
| 170 |
+
gr.Examples(examples=examples, inputs=[user_query], outputs=[])
|
| 171 |
|
| 172 |
btn = gr.Button("Ask 🪄")
|
| 173 |
|