Spaces:
Running
Running
File size: 3,886 Bytes
8af54b8 a6d7b1c 4c7982b 8af54b8 e01a5f6 a6d7b1c e01a5f6 a6d7b1c |
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 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 |
# %%
try:
from ipytorch import logging
except Exception as e:
import logging
from typing import Any, Optional, Protocol, Iterable, Callable
from tqdm.auto import tqdm
from evaluate.evaluation_suite import EvaluationSuite
import evaluate
import numpy as np
import datasets
from tasks import Task, Metrics, fake_pipeline
from utils import is_equiv
# %%
# %cd ../tlem
# %load_ext ipytorch
# %ls
# TODO: Add BibTeX citation
_CITATION = """\
@InProceedings{huggingface:module,
title = {A great new module},
authors={huggingface, Inc.},
year={2020}
}
"""
# TODO: Add description of the module here
_DESCRIPTION = """\
A simple measurement that returns the number of elements in dataset.
"""
# TODO: Add description of the arguments of the module here
_KWARGS_DESCRIPTION = """
Calculates number of elements in dataset
Args:
data: list of elements.
Returns:
element_count: number of elements in dataset,
Examples:
>>> measure = evaluate.load("lvwerra/element_count")
>>> measure.compute(["a", "b", "c")
{"element_count": 3}
"""
# TODO: Define external resources urls if needed
BAD_WORDS_URL = "http://url/to/external/resource/bad_words.txt"
@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
class ReasoningMetric(evaluate.Metric):
"""TODO: Short description of my evaluation module."""
def _info(self):
features = datasets.Features(
{
"responses": datasets.Value("string"),
"references": datasets.Value("string"),
}
)
if self.config_name == "svamp":
features = datasets.Features(
{
"responses": datasets.Value("string"),
"references": datasets.Value("float"),
}
)
# TODO: Specifies the evaluate.EvaluationModuleInfo object
return evaluate.EvaluationModuleInfo(
# This is the description that will appear on the modules page.
# module_type="measurement",
description=_DESCRIPTION,
citation=_CITATION,
inputs_description=_KWARGS_DESCRIPTION,
# This defines the format of each prediction and reference
features=features,
# Homepage of the module for documentation
homepage="http://module.homepage",
# Additional links to the codebase or references
codebase_urls=["http://github.com/path/to/codebase/of/new_module"],
reference_urls=["http://path.to.reference.url/new_module"],
)
def _compute(self, responses, references, verbose=False):
results = {}
scores = getattr(Metrics, self.config_name)(responses, references)
acc = np.asarray(scores).mean()
results = {
"accuracy": acc,
"scores": scores,
}
if verbose:
results["references"] = references
results["answers"] = responses
# results["scores"] = scores
return results
class Suite(EvaluationSuite):
def run(
self, model_or_pipeline: Any, prompt: str = "{instruction}"
) -> dict[str, float]:
self.assert_suite_nonempty()
results_all = {}
for task in tqdm(self.suite, desc="Running tasks"):
task_name = task.name
results = task.run(model_or_pipeline)
results_all[task_name] = results
return results_all
def __init__(self, name):
super().__init__(name)
self.suite = [
Task(
dataset_name=("gsm8k", "main"),
metric_name=("sustech/tlem", "gsm8k"),
input_column="question",
label_column="answer",
)
# TASK_REGISTRY["gsm8k"],
# TASK_REGISTRY["competition_math"],
]
# %%
|