mirror of
https://github.com/hwchase17/langchain.git
synced 2025-04-30 21:05:36 +00:00
a0c2281540
```python
"""python scripts/update_mypy_ruff.py"""
import glob
import tomllib
from pathlib import Path
import toml
import subprocess
import re
ROOT_DIR = Path(__file__).parents[1]
def main():
for path in glob.glob(str(ROOT_DIR / "libs/**/pyproject.toml"), recursive=True):
print(path)
with open(path, "rb") as f:
pyproject = tomllib.load(f)
try:
pyproject["tool"]["poetry"]["group"]["typing"]["dependencies"]["mypy"] = (
"^1.10"
)
pyproject["tool"]["poetry"]["group"]["lint"]["dependencies"]["ruff"] = (
"^0.5"
)
except KeyError:
continue
with open(path, "w") as f:
toml.dump(pyproject, f)
cwd = "/".join(path.split("/")[:-1])
completed = subprocess.run(
"poetry lock --no-update; poetry install --with typing; poetry run mypy . --no-color",
cwd=cwd,
shell=True,
capture_output=True,
text=True,
)
logs = completed.stdout.split("\n")
to_ignore = {}
for l in logs:
if re.match("^(.*)\:(\d+)\: error:.*\[(.*)\]", l):
path, line_no, error_type = re.match(
"^(.*)\:(\d+)\: error:.*\[(.*)\]", l
).groups()
if (path, line_no) in to_ignore:
to_ignore[(path, line_no)].append(error_type)
else:
to_ignore[(path, line_no)] = [error_type]
print(len(to_ignore))
for (error_path, line_no), error_types in to_ignore.items():
all_errors = ", ".join(error_types)
full_path = f"{cwd}/{error_path}"
try:
with open(full_path, "r") as f:
file_lines = f.readlines()
except FileNotFoundError:
continue
file_lines[int(line_no) - 1] = (
file_lines[int(line_no) - 1][:-1] + f" # type: ignore[{all_errors}]\n"
)
with open(full_path, "w") as f:
f.write("".join(file_lines))
subprocess.run(
"poetry run ruff format .; poetry run ruff --select I --fix .",
cwd=cwd,
shell=True,
capture_output=True,
text=True,
)
if __name__ == "__main__":
main()
```
116 lines
3.8 KiB
Python
116 lines
3.8 KiB
Python
"""Select and order examples based on ngram overlap score (sentence_bleu score).
|
|
|
|
https://www.nltk.org/_modules/nltk/translate/bleu_score.html
|
|
https://aclanthology.org/P02-1040.pdf
|
|
"""
|
|
|
|
from typing import Dict, List
|
|
|
|
import numpy as np
|
|
from langchain_core.example_selectors import BaseExampleSelector
|
|
from langchain_core.prompts import PromptTemplate
|
|
from langchain_core.pydantic_v1 import BaseModel, root_validator
|
|
|
|
|
|
def ngram_overlap_score(source: List[str], example: List[str]) -> float:
|
|
"""Compute ngram overlap score of source and example as sentence_bleu score
|
|
from NLTK package.
|
|
|
|
Use sentence_bleu with method1 smoothing function and auto reweighting.
|
|
Return float value between 0.0 and 1.0 inclusive.
|
|
https://www.nltk.org/_modules/nltk/translate/bleu_score.html
|
|
https://aclanthology.org/P02-1040.pdf
|
|
"""
|
|
from nltk.translate.bleu_score import (
|
|
SmoothingFunction, # type: ignore
|
|
sentence_bleu,
|
|
)
|
|
|
|
hypotheses = source[0].split()
|
|
references = [s.split() for s in example]
|
|
|
|
return float(
|
|
sentence_bleu(
|
|
references,
|
|
hypotheses,
|
|
smoothing_function=SmoothingFunction().method1,
|
|
auto_reweigh=True,
|
|
)
|
|
)
|
|
|
|
|
|
class NGramOverlapExampleSelector(BaseExampleSelector, BaseModel):
|
|
"""Select and order examples based on ngram overlap score (sentence_bleu score
|
|
from NLTK package).
|
|
|
|
https://www.nltk.org/_modules/nltk/translate/bleu_score.html
|
|
https://aclanthology.org/P02-1040.pdf
|
|
"""
|
|
|
|
examples: List[dict]
|
|
"""A list of the examples that the prompt template expects."""
|
|
|
|
example_prompt: PromptTemplate
|
|
"""Prompt template used to format the examples."""
|
|
|
|
threshold: float = -1.0
|
|
"""Threshold at which algorithm stops. Set to -1.0 by default.
|
|
|
|
For negative threshold:
|
|
select_examples sorts examples by ngram_overlap_score, but excludes none.
|
|
For threshold greater than 1.0:
|
|
select_examples excludes all examples, and returns an empty list.
|
|
For threshold equal to 0.0:
|
|
select_examples sorts examples by ngram_overlap_score,
|
|
and excludes examples with no ngram overlap with input.
|
|
"""
|
|
|
|
@root_validator(pre=True)
|
|
def check_dependencies(cls, values: Dict) -> Dict:
|
|
"""Check that valid dependencies exist."""
|
|
try:
|
|
from nltk.translate.bleu_score import ( # noqa: F401
|
|
SmoothingFunction,
|
|
sentence_bleu,
|
|
)
|
|
except ImportError as e:
|
|
raise ImportError(
|
|
"Not all the correct dependencies for this ExampleSelect exist."
|
|
"Please install nltk with `pip install nltk`."
|
|
) from e
|
|
|
|
return values
|
|
|
|
def add_example(self, example: Dict[str, str]) -> None:
|
|
"""Add new example to list."""
|
|
self.examples.append(example)
|
|
|
|
def select_examples(self, input_variables: Dict[str, str]) -> List[dict]:
|
|
"""Return list of examples sorted by ngram_overlap_score with input.
|
|
|
|
Descending order.
|
|
Excludes any examples with ngram_overlap_score less than or equal to threshold.
|
|
"""
|
|
inputs = list(input_variables.values())
|
|
examples = []
|
|
k = len(self.examples)
|
|
score = [0.0] * k
|
|
first_prompt_template_key = self.example_prompt.input_variables[0]
|
|
|
|
for i in range(k):
|
|
score[i] = ngram_overlap_score(
|
|
inputs, [self.examples[i][first_prompt_template_key]]
|
|
)
|
|
|
|
while True:
|
|
arg_max = np.argmax(score)
|
|
if (score[arg_max] < self.threshold) or abs(
|
|
score[arg_max] - self.threshold
|
|
) < 1e-9:
|
|
break
|
|
|
|
examples.append(self.examples[arg_max])
|
|
score[arg_max] = self.threshold - 1.0
|
|
|
|
return examples
|