Spaces:
Sleeping
Sleeping
no subprocess
Browse files- app.py +10 -146
- evaluation_logic.py +193 -0
- requirements.txt +2 -0
app.py
CHANGED
|
@@ -1,157 +1,21 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
-
import
|
| 3 |
-
import sys
|
| 4 |
-
from pathlib import Path
|
| 5 |
-
from datetime import datetime
|
| 6 |
-
import json
|
| 7 |
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
duckdb_nsql_dir = current_dir / 'duckdb-nsql'
|
| 11 |
-
eval_dir = duckdb_nsql_dir / 'eval'
|
| 12 |
-
sys.path.extend([str(current_dir), str(duckdb_nsql_dir), str(eval_dir)])
|
| 13 |
-
|
| 14 |
-
# Import necessary functions and classes from predict.py and evaluate.py
|
| 15 |
-
from eval.predict import predict, console, get_manifest, DefaultLoader
|
| 16 |
-
from eval.constants import PROMPT_FORMATTERS
|
| 17 |
-
from eval.evaluate import evaluate, compute_metrics, get_to_print
|
| 18 |
-
from eval.evaluate import test_suite_evaluation, read_tables_json
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
def run_evaluation(model_name):
|
| 22 |
-
results = []
|
| 23 |
-
|
| 24 |
-
if "OPENROUTER_API_KEY" not in os.environ:
|
| 25 |
-
return "Error: OPENROUTER_API_KEY not found in environment variables."
|
| 26 |
-
|
| 27 |
-
try:
|
| 28 |
-
# Set up the arguments similar to the CLI in predict.py
|
| 29 |
-
dataset_path = "duckdb-nsql/eval/data/dev.json"
|
| 30 |
-
table_meta_path = "duckdb-nsql/eval/data/tables.json"
|
| 31 |
-
output_dir = "duckdb-nsql/output/"
|
| 32 |
-
prompt_format = "duckdbinstgraniteshort"
|
| 33 |
-
stop_tokens = [';']
|
| 34 |
-
max_tokens = 30000
|
| 35 |
-
temperature = 0.1
|
| 36 |
-
num_beams = -1
|
| 37 |
-
manifest_client = "openrouter"
|
| 38 |
-
manifest_engine = model_name
|
| 39 |
-
manifest_connection = "http://localhost:5000"
|
| 40 |
-
overwrite_manifest = True
|
| 41 |
-
parallel = False
|
| 42 |
-
|
| 43 |
-
# Initialize necessary components
|
| 44 |
-
data_formatter = DefaultLoader()
|
| 45 |
-
prompt_formatter = PROMPT_FORMATTERS[prompt_format]()
|
| 46 |
-
|
| 47 |
-
# Load manifest
|
| 48 |
-
manifest = get_manifest(
|
| 49 |
-
manifest_client=manifest_client,
|
| 50 |
-
manifest_connection=manifest_connection,
|
| 51 |
-
manifest_engine=manifest_engine,
|
| 52 |
-
)
|
| 53 |
-
|
| 54 |
-
results.append(f"Using model: {manifest_engine}")
|
| 55 |
-
|
| 56 |
-
# Load data and metadata
|
| 57 |
-
results.append("Loading metadata and data...")
|
| 58 |
-
db_to_tables = data_formatter.load_table_metadata(table_meta_path)
|
| 59 |
-
data = data_formatter.load_data(dataset_path)
|
| 60 |
-
|
| 61 |
-
# Generate output filename
|
| 62 |
-
date_today = datetime.now().strftime("%y-%m-%d")
|
| 63 |
-
pred_filename = f"{prompt_format}_0docs_{manifest_engine.split('/')[-1]}_{Path(dataset_path).stem}_{date_today}.json"
|
| 64 |
-
pred_path = Path(output_dir) / pred_filename
|
| 65 |
-
results.append(f"Prediction will be saved to: {pred_path}")
|
| 66 |
-
|
| 67 |
-
# Debug: Print predict function signature
|
| 68 |
-
yield f"Predict function signature: {inspect.signature(predict)}"
|
| 69 |
-
|
| 70 |
-
# Run prediction
|
| 71 |
-
yield "Starting prediction..."
|
| 72 |
-
try:
|
| 73 |
-
predict(
|
| 74 |
-
dataset_path=dataset_path,
|
| 75 |
-
table_meta_path=table_meta_path,
|
| 76 |
-
output_dir=output_dir,
|
| 77 |
-
prompt_format=prompt_format,
|
| 78 |
-
stop_tokens=stop_tokens,
|
| 79 |
-
max_tokens=max_tokens,
|
| 80 |
-
temperature=temperature,
|
| 81 |
-
num_beams=num_beams,
|
| 82 |
-
manifest_client=manifest_client,
|
| 83 |
-
manifest_engine=manifest_engine,
|
| 84 |
-
manifest_connection=manifest_connection,
|
| 85 |
-
overwrite_manifest=overwrite_manifest,
|
| 86 |
-
parallel=parallel
|
| 87 |
-
)
|
| 88 |
-
except TypeError as e:
|
| 89 |
-
yield f"TypeError in predict function: {str(e)}"
|
| 90 |
-
yield "Attempting to call predict with only expected arguments..."
|
| 91 |
-
# Try calling predict with only the arguments it expects
|
| 92 |
-
predict_args = inspect.getfullargspec(predict).args
|
| 93 |
-
filtered_args = {k: v for k, v in locals().items() if k in predict_args}
|
| 94 |
-
predict(**filtered_args)
|
| 95 |
-
|
| 96 |
-
results.append("Prediction completed.")
|
| 97 |
-
|
| 98 |
-
# Run evaluation
|
| 99 |
-
results.append("Starting evaluation...")
|
| 100 |
-
|
| 101 |
-
# Set up evaluation arguments
|
| 102 |
-
gold_path = Path(dataset_path)
|
| 103 |
-
db_dir = "duckdb-nsql/eval/data/databases/"
|
| 104 |
-
tables_path = Path(table_meta_path)
|
| 105 |
-
|
| 106 |
-
kmaps = test_suite_evaluation.build_foreign_key_map_from_json(str(tables_path))
|
| 107 |
-
db_schemas = read_tables_json(str(tables_path))
|
| 108 |
-
|
| 109 |
-
gold_sqls_dict = json.load(gold_path.open("r", encoding="utf-8"))
|
| 110 |
-
pred_sqls_dict = [json.loads(l) for l in pred_path.open("r").readlines()]
|
| 111 |
-
|
| 112 |
-
gold_sqls = [p.get("query", p.get("sql", "")) for p in gold_sqls_dict]
|
| 113 |
-
setup_sqls = [p["setup_sql"] for p in gold_sqls_dict]
|
| 114 |
-
validate_sqls = [p["validation_sql"] for p in gold_sqls_dict]
|
| 115 |
-
gold_dbs = [p.get("db_id", p.get("db", "")) for p in gold_sqls_dict]
|
| 116 |
-
pred_sqls = [p["pred"] for p in pred_sqls_dict]
|
| 117 |
-
categories = [p.get("category", "") for p in gold_sqls_dict]
|
| 118 |
-
|
| 119 |
-
metrics = compute_metrics(
|
| 120 |
-
gold_sqls=gold_sqls,
|
| 121 |
-
pred_sqls=pred_sqls,
|
| 122 |
-
gold_dbs=gold_dbs,
|
| 123 |
-
setup_sqls=setup_sqls,
|
| 124 |
-
validate_sqls=validate_sqls,
|
| 125 |
-
kmaps=kmaps,
|
| 126 |
-
db_schemas=db_schemas,
|
| 127 |
-
database_dir=db_dir,
|
| 128 |
-
lowercase_schema_match=False,
|
| 129 |
-
model_name=model_name,
|
| 130 |
-
categories=categories,
|
| 131 |
-
)
|
| 132 |
-
|
| 133 |
-
results.append("Evaluation completed.")
|
| 134 |
-
|
| 135 |
-
# Format and add the evaluation metrics to the results
|
| 136 |
-
if metrics:
|
| 137 |
-
to_print = get_to_print({"all": metrics}, "all", model_name, len(gold_sqls))
|
| 138 |
-
formatted_metrics = "\n".join([f"{k}: {v}" for k, v in to_print.items() if k not in ["slice", "model"]])
|
| 139 |
-
results.append(f"Evaluation metrics:\n{formatted_metrics}")
|
| 140 |
-
else:
|
| 141 |
-
results.append("No evaluation metrics returned.")
|
| 142 |
-
|
| 143 |
-
except Exception as e:
|
| 144 |
-
results.append(f"An unexpected error occurred: {str(e)}")
|
| 145 |
-
|
| 146 |
-
return "\n\n".join(results)
|
| 147 |
|
| 148 |
with gr.Blocks() as demo:
|
| 149 |
gr.Markdown("# DuckDB SQL Evaluation App")
|
| 150 |
|
| 151 |
model_name = gr.Textbox(label="Model Name (e.g., qwen/qwen-2.5-72b-instruct)")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
start_btn = gr.Button("Start Evaluation")
|
| 153 |
output = gr.Textbox(label="Output", lines=20)
|
| 154 |
|
| 155 |
-
start_btn.click(fn=
|
| 156 |
|
| 157 |
-
demo.launch()
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
+
from evaluation_logic import run_evaluation, AVAILABLE_PROMPT_FORMATS
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
|
| 4 |
+
def gradio_run_evaluation(model_name, prompt_format):
|
| 5 |
+
return run_evaluation(model_name, prompt_format)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
|
| 7 |
with gr.Blocks() as demo:
|
| 8 |
gr.Markdown("# DuckDB SQL Evaluation App")
|
| 9 |
|
| 10 |
model_name = gr.Textbox(label="Model Name (e.g., qwen/qwen-2.5-72b-instruct)")
|
| 11 |
+
prompt_format = gr.Dropdown(
|
| 12 |
+
label="Prompt Format",
|
| 13 |
+
choices=AVAILABLE_PROMPT_FORMATS,
|
| 14 |
+
value="duckdbinstgraniteshort"
|
| 15 |
+
)
|
| 16 |
start_btn = gr.Button("Start Evaluation")
|
| 17 |
output = gr.Textbox(label="Output", lines=20)
|
| 18 |
|
| 19 |
+
start_btn.click(fn=gradio_run_evaluation, inputs=[model_name, prompt_format], outputs=output)
|
| 20 |
|
| 21 |
+
demo.queue().launch()
|
evaluation_logic.py
ADDED
|
@@ -0,0 +1,193 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import sys
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from datetime import datetime
|
| 5 |
+
import json
|
| 6 |
+
import traceback
|
| 7 |
+
|
| 8 |
+
# Add the necessary directories to the Python path
|
| 9 |
+
current_dir = Path(__file__).resolve().parent
|
| 10 |
+
duckdb_nsql_dir = current_dir / 'duckdb-nsql'
|
| 11 |
+
eval_dir = duckdb_nsql_dir / 'eval'
|
| 12 |
+
sys.path.extend([str(current_dir), str(duckdb_nsql_dir), str(eval_dir)])
|
| 13 |
+
|
| 14 |
+
# Import necessary functions and classes
|
| 15 |
+
from eval.predict import get_manifest, DefaultLoader, PROMPT_FORMATTERS, generate_sql
|
| 16 |
+
from eval.evaluate import evaluate, compute_metrics, get_to_print
|
| 17 |
+
from eval.evaluate import test_suite_evaluation, read_tables_json
|
| 18 |
+
from eval.schema import TextToSQLParams, Table
|
| 19 |
+
|
| 20 |
+
AVAILABLE_PROMPT_FORMATS = list(PROMPT_FORMATTERS.keys())
|
| 21 |
+
|
| 22 |
+
def run_prediction(model_name, prompt_format, output_file):
|
| 23 |
+
dataset_path = str(eval_dir / "data/dev.json")
|
| 24 |
+
table_meta_path = str(eval_dir / "data/tables.json")
|
| 25 |
+
stop_tokens = [';']
|
| 26 |
+
max_tokens = 30000
|
| 27 |
+
temperature = 0.1
|
| 28 |
+
num_beams = -1
|
| 29 |
+
manifest_client = "openrouter"
|
| 30 |
+
manifest_engine = model_name
|
| 31 |
+
manifest_connection = "http://localhost:5000"
|
| 32 |
+
overwrite_manifest = True
|
| 33 |
+
parallel = False
|
| 34 |
+
|
| 35 |
+
yield "Starting prediction..."
|
| 36 |
+
|
| 37 |
+
try:
|
| 38 |
+
# Initialize necessary components
|
| 39 |
+
data_formatter = DefaultLoader()
|
| 40 |
+
prompt_formatter = PROMPT_FORMATTERS[prompt_format]()
|
| 41 |
+
|
| 42 |
+
# Load manifest
|
| 43 |
+
manifest = get_manifest(
|
| 44 |
+
manifest_client=manifest_client,
|
| 45 |
+
manifest_connection=manifest_connection,
|
| 46 |
+
manifest_engine=manifest_engine,
|
| 47 |
+
)
|
| 48 |
+
|
| 49 |
+
# Load data
|
| 50 |
+
data = data_formatter.load_data(dataset_path)
|
| 51 |
+
db_to_tables = data_formatter.load_table_metadata(table_meta_path)
|
| 52 |
+
|
| 53 |
+
# Prepare input for generate_sql
|
| 54 |
+
text_to_sql_inputs = []
|
| 55 |
+
for input_question in data:
|
| 56 |
+
question = input_question["question"]
|
| 57 |
+
db_id = input_question.get("db_id", "none")
|
| 58 |
+
if db_id != "none":
|
| 59 |
+
table_params = list(db_to_tables.get(db_id, {}).values())
|
| 60 |
+
else:
|
| 61 |
+
table_params = []
|
| 62 |
+
|
| 63 |
+
if len(table_params) == 0:
|
| 64 |
+
yield f"[red] WARNING: No tables found for {db_id} [/red]"
|
| 65 |
+
|
| 66 |
+
text_to_sql_inputs.append(TextToSQLParams(
|
| 67 |
+
instruction=question,
|
| 68 |
+
database=db_id,
|
| 69 |
+
tables=table_params,
|
| 70 |
+
))
|
| 71 |
+
|
| 72 |
+
# Generate SQL
|
| 73 |
+
generated_sqls = generate_sql(
|
| 74 |
+
manifest=manifest,
|
| 75 |
+
text_to_sql_in=text_to_sql_inputs,
|
| 76 |
+
retrieved_docs=[[] for _ in text_to_sql_inputs], # Assuming no retrieved docs
|
| 77 |
+
prompt_formatter=prompt_formatter,
|
| 78 |
+
stop_tokens=stop_tokens,
|
| 79 |
+
overwrite_manifest=overwrite_manifest,
|
| 80 |
+
max_tokens=max_tokens,
|
| 81 |
+
temperature=temperature,
|
| 82 |
+
num_beams=num_beams,
|
| 83 |
+
parallel=parallel
|
| 84 |
+
)
|
| 85 |
+
|
| 86 |
+
# Save results
|
| 87 |
+
with output_file.open('w') as f:
|
| 88 |
+
for original_data, (sql, _) in zip(data, generated_sqls):
|
| 89 |
+
output = {**original_data, "pred": sql}
|
| 90 |
+
json.dump(output, f)
|
| 91 |
+
f.write('\n')
|
| 92 |
+
|
| 93 |
+
yield f"Prediction completed. Results saved to {output_file}"
|
| 94 |
+
except Exception as e:
|
| 95 |
+
yield f"Prediction failed with error: {str(e)}"
|
| 96 |
+
yield f"Error traceback: {traceback.format_exc()}"
|
| 97 |
+
|
| 98 |
+
def run_evaluation(model_name, prompt_format="duckdbinstgraniteshort"):
|
| 99 |
+
if "OPENROUTER_API_KEY" not in os.environ:
|
| 100 |
+
yield "Error: OPENROUTER_API_KEY not found in environment variables."
|
| 101 |
+
return
|
| 102 |
+
|
| 103 |
+
try:
|
| 104 |
+
# Set up the arguments
|
| 105 |
+
dataset_path = str(eval_dir / "data/dev.json")
|
| 106 |
+
table_meta_path = str(eval_dir / "data/tables.json")
|
| 107 |
+
output_dir = eval_dir / "output"
|
| 108 |
+
|
| 109 |
+
yield f"Using model: {model_name}"
|
| 110 |
+
yield f"Using prompt format: {prompt_format}"
|
| 111 |
+
|
| 112 |
+
output_file = output_dir / f"{prompt_format}_0docs_{model_name.trim().replace('/', '_')}_dev_{datetime.now().strftime('%y-%m-%d')}.json"
|
| 113 |
+
|
| 114 |
+
# Ensure the output directory exists
|
| 115 |
+
output_dir.mkdir(parents=True, exist_ok=True)
|
| 116 |
+
|
| 117 |
+
if output_file.exists():
|
| 118 |
+
yield f"Prediction file already exists: {output_file}"
|
| 119 |
+
yield "Skipping prediction step and proceeding to evaluation."
|
| 120 |
+
else:
|
| 121 |
+
# Run prediction
|
| 122 |
+
for output in run_prediction(model_name, prompt_format, output_file):
|
| 123 |
+
yield output
|
| 124 |
+
|
| 125 |
+
# Run evaluation
|
| 126 |
+
yield "Starting evaluation..."
|
| 127 |
+
|
| 128 |
+
# Set up evaluation arguments
|
| 129 |
+
gold_path = Path(dataset_path)
|
| 130 |
+
db_dir = str(eval_dir / "data/databases/")
|
| 131 |
+
tables_path = Path(table_meta_path)
|
| 132 |
+
|
| 133 |
+
kmaps = test_suite_evaluation.build_foreign_key_map_from_json(str(tables_path))
|
| 134 |
+
db_schemas = read_tables_json(str(tables_path))
|
| 135 |
+
|
| 136 |
+
gold_sqls_dict = json.load(gold_path.open("r", encoding="utf-8"))
|
| 137 |
+
pred_sqls_dict = [json.loads(l) for l in output_file.open("r").readlines()]
|
| 138 |
+
|
| 139 |
+
gold_sqls = [p.get("query", p.get("sql", "")) for p in gold_sqls_dict]
|
| 140 |
+
setup_sqls = [p["setup_sql"] for p in gold_sqls_dict]
|
| 141 |
+
validate_sqls = [p["validation_sql"] for p in gold_sqls_dict]
|
| 142 |
+
gold_dbs = [p.get("db_id", p.get("db", "")) for p in gold_sqls_dict]
|
| 143 |
+
pred_sqls = [p["pred"] for p in pred_sqls_dict]
|
| 144 |
+
categories = [p.get("category", "") for p in gold_sqls_dict]
|
| 145 |
+
|
| 146 |
+
yield "Computing metrics..."
|
| 147 |
+
metrics = compute_metrics(
|
| 148 |
+
gold_sqls=gold_sqls,
|
| 149 |
+
pred_sqls=pred_sqls,
|
| 150 |
+
gold_dbs=gold_dbs,
|
| 151 |
+
setup_sqls=setup_sqls,
|
| 152 |
+
validate_sqls=validate_sqls,
|
| 153 |
+
kmaps=kmaps,
|
| 154 |
+
db_schemas=db_schemas,
|
| 155 |
+
database_dir=db_dir,
|
| 156 |
+
lowercase_schema_match=False,
|
| 157 |
+
model_name=model_name,
|
| 158 |
+
categories=categories,
|
| 159 |
+
)
|
| 160 |
+
|
| 161 |
+
yield "Evaluation completed."
|
| 162 |
+
|
| 163 |
+
if metrics:
|
| 164 |
+
yield "Overall Results:"
|
| 165 |
+
overall_metrics = metrics['exec']['all']
|
| 166 |
+
yield f"Count: {overall_metrics['count']}"
|
| 167 |
+
yield f"Execution Accuracy: {overall_metrics['exec']:.3f}"
|
| 168 |
+
yield f"Exact Match Accuracy: {overall_metrics['exact']:.3f}"
|
| 169 |
+
yield f"Equality: {metrics['equality']['equality']:.3f}"
|
| 170 |
+
yield f"Edit Distance: {metrics['edit_distance']['edit_distance']:.3f}"
|
| 171 |
+
|
| 172 |
+
yield "\nResults by Category:"
|
| 173 |
+
categories = ['easy', 'medium', 'hard', 'duckdb', 'ddl', 'all']
|
| 174 |
+
|
| 175 |
+
for category in categories:
|
| 176 |
+
if category in metrics['exec']:
|
| 177 |
+
yield f"\n{category}:"
|
| 178 |
+
category_metrics = metrics['exec'][category]
|
| 179 |
+
yield f"Count: {category_metrics['count']}"
|
| 180 |
+
yield f"Execution Accuracy: {category_metrics['exec']:.3f}"
|
| 181 |
+
else:
|
| 182 |
+
yield f"\n{category}: No data available"
|
| 183 |
+
else:
|
| 184 |
+
yield "No evaluation metrics returned."
|
| 185 |
+
except Exception as e:
|
| 186 |
+
yield f"An unexpected error occurred: {str(e)}"
|
| 187 |
+
yield f"Error traceback: {traceback.format_exc()}"
|
| 188 |
+
|
| 189 |
+
if __name__ == "__main__":
|
| 190 |
+
model_name = input("Enter the model name: ")
|
| 191 |
+
prompt_format = input("Enter the prompt format (default is duckdbinstgraniteshort): ") or "duckdbinstgraniteshort"
|
| 192 |
+
for result in run_evaluation(model_name, prompt_format):
|
| 193 |
+
print(result, flush=True)
|
requirements.txt
CHANGED
|
@@ -20,6 +20,7 @@ peft==0.6.0
|
|
| 20 |
packaging==23.2
|
| 21 |
ninja==1.11.1.1
|
| 22 |
langchain
|
|
|
|
| 23 |
pydantic
|
| 24 |
packaging
|
| 25 |
#./duckdb-nsql/manifest
|
|
@@ -28,3 +29,4 @@ flask
|
|
| 28 |
diffusers
|
| 29 |
deepspeed
|
| 30 |
sentence_transformers
|
|
|
|
|
|
| 20 |
packaging==23.2
|
| 21 |
ninja==1.11.1.1
|
| 22 |
langchain
|
| 23 |
+
gradio
|
| 24 |
pydantic
|
| 25 |
packaging
|
| 26 |
#./duckdb-nsql/manifest
|
|
|
|
| 29 |
diffusers
|
| 30 |
deepspeed
|
| 31 |
sentence_transformers
|
| 32 |
+
tqdm
|