functions-656k / README.md
crumb's picture
Create README.md
c14a7ae verified
metadata
license: other

Functions-656K

six hundred fifty six thousand samples of code, annotated with input and output types as well as any dependencies or libraries required for their use. completely anonymized variables. i anonymized all the variables to sort of low-pass a bunch of high frequency features, a model of the space should be smoother to navigate.

1105k files from the stack -> 1376k cleanly typed functions w/ dependencies -> variable names anonymized -> 656k de-duplicated w minhash+lsh
licenses: check https://hf.co/datasets/bigcode/the-stack

filtering libraries: use the following: dataset.filter(lambda x: False not in [label in allowed_set for label in x['lib_used']])
where allowed_set is a subset of this list, which is the master list of allowed libraries in the dataset:

    "math", "cmath", "random", "statistics", "fractions", "decimal", "numbers",
    "itertools", "functools", "operator",
    "collections", "array", "bisect", "heapq", "weakref", "enum", "typing",
    "string", "re", "textwrap", "pprint", "reprlib", "difflib",
    "hashlib", "hmac", "secrets", "base64", "binascii", "quopri", "uu",
    "copy", "contextlib", "abc", "dataclasses", "warnings", "gc",
    "datetime", "calendar",
    "inspect", "traceback", "ast", "tokenize", "keyword", "dis", "sys",
    "json", "csv", "marshal",
    "uuid",
    "numpy", "scipy", "sympy", "sklearn", "statsmodels", "xgboost", "tqdm",
    "shapely", "PyWavelets", "z3", "cvxpy",
"os", "io", "pathlib", "glob", "fnmatch", "linecache", "tempfile", "shutil", "fileinput", "stat", "pickle", "shelve", "dbm", "sqlite3", "zipfile", "tarfile", "gzip", "bz2", "lzma", "zlib", "configparser", "argparse", "getopt", "optparse", "logging", "getpass", "profile", "cProfile", "pstats", "timeit", "tracemalloc", "faulthandler", "pdb", "curses", "termios", "tty", "pty", "platform", "plistlib", "netrc", "xdrlib", "mailbox", "mimetypes", "wave", "aifc", "sndhdr", "imghdr", "colorsys", "chunk", "sysconfig", "site", "resource", "pandas", "matplotlib", "seaborn", "plotly", "torch", "tensorflow", "keras", "gensim", "skimage", "PIL", "cv2", "nltk", "spacy", "dask", "polars", "vaex", "h5py", "netCDF4", "xarray", "Bio", "scikit-bio", "PyCogent", "geopandas", "rasterio", "rdkit", "openbabel", "pymatgen", "astropy", "sunpy", "healpy", "librosa", "MDAnalysis", "OpenMM", "cartopy",
"subprocess", "shlex", "socket", "ssl", "select", "selectors", "asyncio", "socketserver", "http", "urllib", "ftplib", "poplib", "imaplib", "smtplib", "telnetlib", "xmlrpc", "webbrowser", "cgi", "cgitb", "wsgiref", "http.server", "threading", "multiprocessing", "concurrent", "queue", "sched", "contextvars", "signal", "mmap", "importlib", "runpy", "code", "codeop", "ctypes", "cffi", "msvcrt", "winreg", "winsound", "posix", "pwd", "spwd", "grp", "crypt", "fcntl", "syslog", "pipes", "email", "ossaudiodev", "audioop", "bdb", "trace", "transformers", "datasets", "requests", "beautifulsoup4", "selenium", "folium", "qiskit"