File size: 2,543 Bytes
36821d3
 
dff7018
36821d3
dff7018
1f6b0d7
d2e7f91
 
dff7018
1f6b0d7
 
dff7018
36821d3
 
 
dff7018
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36821d3
dff7018
36821d3
 
 
 
dff7018
36821d3
 
dff7018
 
 
 
 
 
36821d3
dff7018
 
 
 
 
36821d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dff7018
36821d3
 
dff7018
36821d3
 
dff7018
36821d3
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
import json

import spaces
import gradio as gr
import torch
from distilabel.llms import vLLM
from distilabel.steps.tasks.argillalabeller import ArgillaLabeller


llm = vLLM(
    model="Qwen/Qwen2-0.5B-Instruct-GGUF"
)
task = ArgillaLabeller(llm=llm)
task.load()


def load_examples():
    with open("examples.json", "r") as f:
        return json.load(f)


# Create Gradio examples
examples = load_examples()


def process_fields(fields):
    if isinstance(fields, str):
        fields = json.loads(fields)
    if isinstance(fields, dict):
        fields = [fields]
    return [field if isinstance(field, dict) else json.loads(field) for field in fields]


@spaces.GPU
def process_records_gradio(records, example_records, fields, question):
    try:
        # Convert string inputs to dictionaries
        records = json.loads(records)
        example_records = json.loads(example_records) if example_records else None
        fields = process_fields(fields) if fields else None
        question = json.loads(question) if question else None

        print(fields)
        print(question)
        print(example_records)

        if not fields and not question:
            return "Error: Either fields or question must be provided"

        runtime_parameters = {"fields": fields, "question": question}
        if example_records:
            runtime_parameters["example_records"] = example_records
        print(runtime_parameters)
        task.set_runtime_parameters(runtime_parameters)

        results = []
        for record in records:
            output = next(task.process(inputs=[{"records": record}]))
            results.append(output[0]["suggestions"])

        return json.dumps({"results": results}, indent=2)
    except Exception as e:
        return f"Error: {str(e)}"


interface = gr.Interface(
    fn=process_records_gradio,
    inputs=[
        gr.Code(label="Records (JSON)", language="json", lines=5),
        gr.Code(label="Example Records (JSON, optional)", language="json", lines=5),
        gr.Code(label="Fields (JSON, optional)", language="json"),
        gr.Code(label="Question (JSON, optional)", language="json"),
    ],
    examples=examples,
    outputs=gr.Code(label="Suggestions", language="json", lines=10),
    title="Record Processing Interface",
    description="Enter JSON data for `rg.Record.to_dict()`, `List[rg.Record.to_dict()]`, `List[Field].serialize()`, or `List[rg.Question.serialize()]` At least one of fields or question must be provided.",
)

if __name__ == "__main__":
    interface.launch()