Spaces:
Sleeping
Sleeping
| import sys | |
| import re | |
| from importlib.metadata import version | |
| import polars as pl | |
| import gradio as gr | |
| # Config | |
| concurrency_limit = 5 | |
| title = "See ASR Outputs" | |
| # https://www.tablesgenerator.com/markdown_tables | |
| authors_table = """ | |
| ## Authors | |
| Follow them on social networks and **contact** if you need any help or have any questions: | |
| | <img src="https://avatars.githubusercontent.com/u/7875085?v=4" width="100"> **Yehor Smoliakov** | | |
| |-------------------------------------------------------------------------------------------------| | |
| | https://t.me/smlkw in Telegram | | |
| | https://x.com/yehor_smoliakov at X | | |
| | https://github.com/egorsmkv at GitHub | | |
| | https://huggingface.co/Yehor at Hugging Face | | |
| | or use [email protected] | | |
| """.strip() | |
| examples = [ | |
| ["evaluation_results.jsonl", False], | |
| ["evaluation_results_batch.jsonl", True], | |
| ] | |
| description_head = f""" | |
| # {title} | |
| ## Overview | |
| See generated JSONL files made by ASR models as a dataframe. | |
| """.strip() | |
| description_foot = f""" | |
| {authors_table} | |
| """.strip() | |
| metrics_value = """ | |
| Metrics will appear here. | |
| """.strip() | |
| tech_env = f""" | |
| #### Environment | |
| - Python: {sys.version} | |
| """.strip() | |
| tech_libraries = f""" | |
| #### Libraries | |
| - gradio: {version("gradio")} | |
| - polars: {version("polars")} | |
| """.strip() | |
| def inference(file_name, _batch_mode): | |
| if not file_name: | |
| raise gr.Error("Please paste your JSON file.") | |
| df = pl.read_ndjson(file_name) | |
| required_columns = [ | |
| "filename", | |
| "inference_start", | |
| "inference_end", | |
| "inference_total", | |
| "duration", | |
| "reference", | |
| "prediction", | |
| ] | |
| required_columns_batch = [ | |
| "inference_start", | |
| "inference_end", | |
| "inference_total", | |
| "filenames", | |
| "durations", | |
| "references", | |
| "predictions", | |
| ] | |
| if _batch_mode: | |
| if not all(col in df.columns for col in required_columns_batch): | |
| raise gr.Error( | |
| f"Please provide a JSONL file with the following columns: {required_columns_batch}" | |
| ) | |
| else: | |
| if not all(col in df.columns for col in required_columns): | |
| raise gr.Error( | |
| f"Please provide a JSONL file with the following columns: {required_columns}" | |
| ) | |
| # exclude inference_start, inference_end | |
| if _batch_mode: | |
| df = df.drop(["inference_start", "inference_end", "filenames"]) | |
| else: | |
| df = df.drop(["inference_start", "inference_end", "filename"]) | |
| # round "inference_total" field to 2 decimal places | |
| df = df.with_columns(pl.col("inference_total").round(2)) | |
| return df | |
| demo = gr.Blocks( | |
| title=title, | |
| analytics_enabled=False, | |
| theme=gr.themes.Base(), | |
| ) | |
| with demo: | |
| gr.Markdown(description_head) | |
| gr.Markdown("## Usage") | |
| with gr.Row(): | |
| df = gr.DataFrame( | |
| label="Dataframe", | |
| ) | |
| with gr.Row(): | |
| with gr.Column(): | |
| jsonl_file = gr.File(label="A JSONL file") | |
| batch_mode = gr.Checkbox( | |
| label="Use batch mode", | |
| ) | |
| gr.Button("Show").click( | |
| inference, | |
| concurrency_limit=concurrency_limit, | |
| inputs=[jsonl_file, batch_mode], | |
| outputs=df, | |
| ) | |
| with gr.Row(): | |
| gr.Examples( | |
| label="Choose an example", | |
| inputs=[jsonl_file, batch_mode], | |
| examples=examples, | |
| ) | |
| gr.Markdown(description_foot) | |
| gr.Markdown("### Gradio app uses:") | |
| gr.Markdown(tech_env) | |
| gr.Markdown(tech_libraries) | |
| if __name__ == "__main__": | |
| demo.queue() | |
| demo.launch() | |