Add files using upload-large-folder tool

.gitattributes

@@ -150,3 +150,6 @@ env-llmeval/lib/python3.10/site-packages/nvidia/cusolver/lib/libcusolver.so.11 f
 env-llmeval/lib/python3.10/site-packages/nvidia/cublas/lib/libcublasLt.so.12 filter=lfs diff=lfs merge=lfs -text
 env-llmeval/lib/python3.10/site-packages/nvidia/cusparse/lib/libcusparse.so.12 filter=lfs diff=lfs merge=lfs -text
 env-llmeval/lib/python3.10/site-packages/nvidia/cuda_nvrtc/lib/libnvrtc-builtins.so.12.1 filter=lfs diff=lfs merge=lfs -text
+env-llmeval/lib/python3.10/site-packages/nvidia/cuda_nvrtc/lib/libnvrtc.so.12 filter=lfs diff=lfs merge=lfs -text
+env-llmeval/lib/python3.10/site-packages/nvidia/cuda_cupti/lib/libnvperf_target.so filter=lfs diff=lfs merge=lfs -text
+env-llmeval/lib/python3.10/site-packages/nvidia/cuda_cupti/lib/libcheckpoint.so filter=lfs diff=lfs merge=lfs -text

env-llmeval/lib/python3.10/site-packages/functorch/dim/__init__.py

@@ -0,0 +1,179 @@
+import dis
+import inspect
+from typing import Sequence, Union
+
+import torch
+
+import functorch._C
+from functorch._C import dim as _C
+from .tree_map import tree_flatten, tree_map
+from .wrap_type import wrap_type
+
+_C._patch_tensor_class()
+dims, DimList, dimlists = _C.dims, _C.DimList, _C.dimlists
+
+
+class DimensionMismatchError(Exception):
+    pass
+
+
+class DimensionBindError(Exception):
+    pass
+
+
+from . import op_properties
+
+# use dict to avoid writing C++ bindings for set
+pointwise = {t: True for t in op_properties.pointwise}
+
+use_c = True
+if not use_c:
+    from . import reference
+
+
+class _Tensor:
+    # fast path around slow wrapping/unwrapping logic for simply queries used
+    # by the implementation...
+
+    @property
+    def dims(self):
+        return tuple(d for d in self._levels if isinstance(d, Dim))
+
+    def dim(self):
+        return self.ndim
+
+    if use_c:
+        __torch_function__ = classmethod(_C.__torch_function__)
+        expand = _C._instancemethod(_C.expand)
+    else:
+        __torch_function__ = reference.__torch_function__
+        expand = reference.expand
+
+    index = _C._instancemethod(_C.index)
+
+    def __repr__(self):
+        tensor, levels, ndim = self._tensor, self._levels, self.ndim
+        return f"{tensor}\nwith dims={tuple(l + ndim if isinstance(l, int) else l for l in levels)} sizes={tuple(tensor.size())}"
+
+
+TensorLike = (_Tensor, torch.Tensor)
+
+
+class Dim(_C.Dim, _Tensor):
+    # note that _C.Dim comes before tensor because we want the Dim API for things like size to take precendence.
+    # Tensor defines format, but we want to print Dims with special formatting
+    __format__ = object.__format__
+
+
+class Tensor(_Tensor, _C.Tensor):
+    if not use_c:
+        from_batched = staticmethod(_C.Tensor_from_batched)
+    from_positional = staticmethod(_C.Tensor_from_positional)
+    sum = _C._instancemethod(_C.Tensor_sum)
+
+
+def cat(tensors, dim, new_dim):
+    n = dims()
+    return stack(tensors, n, dim).index([n, dim], new_dim)
+
+
+if use_c:
+    _wrap = _C._wrap
+
+    def _def(name, *args, **kwargs):
+        orig = getattr(torch.Tensor, name)
+        setattr(_Tensor, name, _C._instancemethod(_wrap(orig, *args, **kwargs)))
+
+    t__getitem__ = _C._instancemethod(_C.__getitem__)
+    stack = _C.stack
+    split = _C._instancemethod(_C.split)
+else:
+    _wrap, _def = reference._wrap, reference._def
+    t__getitem__ = reference.t__getitem__
+    stack = reference.stack
+    split = reference.split
+
+# note: there is no python reference
+t__setitem__ = _C._instancemethod(_C.__setitem__)
+# this is patched in the C API because otherwise torch.Tensor will
+# no longer be considered a sequence and things will break
+# torch.Tensor.__getitem__ = t__getitem__
+
+_Tensor.__getitem__ = t__getitem__
+# torch.Tensor.__setitem__ = t__setitem__
+_Tensor.__setitem__ = t__setitem__
+
+torch.Tensor.split = split
+_Tensor.split = split
+torch.Tensor.expand = _C._instancemethod(_C.expand)
+torch.Tensor.index = _C._instancemethod(_C.index)
+wrap_type(use_c, _Tensor, torch.Tensor, _Tensor.__torch_function__)
+del _Tensor.ndim
+
+if use_c:
+    _Tensor.order = _C._instancemethod(_C.order)
+else:
+    _Tensor.order = reference.positional
+
+_def("mean")
+_def("sum")
+_def("all")
+_def("amax")
+_def("amin")
+_def("aminmax")
+_def("any")
+_def("count_nonzero")
+_def("logsumexp")
+_def("nanmean")
+_def("nansum")
+_def("prod")
+_def("std", keepdim_offset=2)
+_def("var", keepdim_offset=2)
+_def("max", single_dim=True)
+_def("min", single_dim=True)
+_def("argmax", single_dim=True)
+_def("argmin", single_dim=True)
+_def("kthvalue", single_dim=True)
+_def("median", single_dim=True)
+_def("nanmedian", single_dim=True)
+_def("mode", single_dim=True)
+_def("sort", reduce=False)
+_def("argsort", reduce=False)
+_def("unbind", single_dim=True)
+_def("chunk", dim_offset=1, reduce=False)
+_def("cummax", single_dim=True, reduce=False)
+_def("cummin", single_dim=True, reduce=False)
+_def("cumprod", single_dim=True, reduce=False)
+_def("cumprod_", single_dim=True, reduce=False)
+_def("cumsum", single_dim=True, reduce=False)
+_def("cumsum_", single_dim=True, reduce=False)
+_def("logcumsumexp", single_dim=True, reduce=False)
+_def("renorm", dim_offset=1, single_dim=True, reduce=False)
+_def("softmax", single_dim=True, reduce=False)
+softmax = _wrap(torch.nn.functional.softmax, single_dim=True, reduce=False)
+
+# stuff to handle in the future, because they require special
+# binding logic for dims
+# cross
+# diag_embed
+# diagonal
+# diagonal_scatter
+# diff
+# nanquantile
+# quantile
+# roll
+# rot90
+# topk (new dimes on output)
+# should these all be subsumed by inplace indexing?
+# index_add_
+# index_add
+# index_copy
+# index_copy_
+# index_fill
+# index_fill_
+# index_select
+# scatter
+# scatter_
+# scatter_add
+# scatter_add_
+# scatter_reduce

env-llmeval/lib/python3.10/site-packages/functorch/dim/batch_tensor.py

@@ -0,0 +1,25 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+from contextlib import contextmanager
+
+from torch._C._functorch import _vmap_add_layers, _vmap_remove_layers
+
+_enabled = False
+
+
+@contextmanager
+def _enable_layers(dims):
+    global _enabled
+    assert not _enabled
+    input = sorted((d._level, d.size) for d in dims if not isinstance(d, int))
+    n = len(input)
+    try:
+        _vmap_add_layers(input)
+        _enabled = True
+        yield
+    finally:
+        _enabled = False
+        _vmap_remove_layers(n)

env-llmeval/lib/python3.10/site-packages/functorch/dim/dim.py

@@ -0,0 +1,110 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+_vmap_levels = []
+
+
+@dataclass
+class LevelInfo:
+    level: int
+    alive: bool = True
+
+
+class Dim:
+    def __init__(self, name: str, size: Union[None, int] = None):
+        self.name = name
+        self._size = None
+        self._vmap_level = None
+        if size is not None:
+            self.size = size
+
+    def __del__(self):
+        if self._vmap_level is not None:
+            _vmap_active_levels[self._vmap_stack].alive = False
+            while (
+                not _vmap_levels[-1].alive and current_level() == _vmap_levels[-1].level
+            ):
+                _vmap_decrement_nesting()
+                _vmap_levels.pop()
+
+    @property
+    def size(self):
+        assert self.is_bound
+        return self._size
+
+    @size.setter
+    def size(self, size: int):
+        if self._size is None:
+            self._size = size
+            self._vmap_level = _vmap_increment_nesting(size, "same")
+            self._vmap_stack = len(_vmap_levels)
+            _vmap_levels.append(LevelInfo(self._vmap_level))
+
+        elif self._size != size:
+            raise DimensionBindError(
+                f"Dim '{self}' previously bound to a dimension of size {self._size} cannot bind to a dimension of size {size}"
+            )
+
+    @property
+    def is_bound(self):
+        return self._size is not None
+
+    def __repr__(self):
+        return self.name
+
+
+def extract_name(inst):
+    assert inst.opname == "STORE_FAST" or inst.opname == "STORE_NAME"
+    return inst.argval
+
+
+_cache = {}
+
+
+def dims(lists=0):
+    frame = inspect.currentframe()
+    assert frame is not None
+    calling_frame = frame.f_back
+    assert calling_frame is not None
+    code, lasti = calling_frame.f_code, calling_frame.f_lasti
+    key = (code, lasti)
+    if key not in _cache:
+        first = lasti // 2 + 1
+        instructions = list(dis.get_instructions(calling_frame.f_code))
+        unpack = instructions[first]
+
+        if unpack.opname == "STORE_FAST" or unpack.opname == "STORE_NAME":
+            # just a single dim, not a list
+            name = unpack.argval
+            ctor = Dim if lists == 0 else DimList
+            _cache[key] = lambda: ctor(name=name)
+        else:
+            assert unpack.opname == "UNPACK_SEQUENCE"
+            ndims = unpack.argval
+            names = tuple(
+                extract_name(instructions[first + 1 + i]) for i in range(ndims)
+            )
+            first_list = len(names) - lists
+            _cache[key] = lambda: tuple(
+                Dim(n) if i < first_list else DimList(name=n)
+                for i, n in enumerate(names)
+            )
+    return _cache[key]()
+
+
+def _dim_set(positional, arg):
+    def convert(a):
+        if isinstance(a, Dim):
+            return a
+        else:
+            assert isinstance(a, int)
+            return positional[a]
+
+    if arg is None:
+        return positional
+    elif not isinstance(arg, (Dim, int)):
+        return tuple(convert(a) for a in arg)
+    else:
+        return (convert(arg),)

env-llmeval/lib/python3.10/site-packages/functorch/dim/magic_trace.py

@@ -0,0 +1,42 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+import os
+import signal
+import subprocess
+from contextlib import contextmanager
+
+
+@contextmanager
+def magic_trace(output="trace.fxt", magic_trace_cache="/tmp/magic-trace"):
+    pid = os.getpid()
+    if not os.path.exists(magic_trace_cache):
+        print(f"Downloading magic_trace to: {magic_trace_cache}")
+        subprocess.run(
+            [
+                "wget",
+                "-O",
+                magic_trace_cache,
+                "-q",
+                "https://github.com/janestreet/magic-trace/releases/download/v1.0.2/magic-trace",
+            ]
+        )
+        subprocess.run(["chmod", "+x", magic_trace_cache])
+    args = [magic_trace_cache, "attach", "-pid", str(pid), "-o", output]
+    p = subprocess.Popen(args, stderr=subprocess.PIPE, encoding="utf-8")
+    while True:
+        x = p.stderr.readline()
+        print(x)
+        if "Attached" in x:
+            break
+    try:
+        yield
+    finally:
+        p.send_signal(signal.SIGINT)
+        r = p.wait()
+        print(p.stderr.read())
+        p.stderr.close()
+        if r != 0:
+            raise ValueError(f"magic_trace exited abnormally: {r}")

env-llmeval/lib/python3.10/site-packages/functorch/dim/op_properties.py

@@ -0,0 +1,311 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+import torch
+
+# pointwise operators can go through a faster pathway
+
+tensor_magic_methods = ["add", ""]
+pointwise_magic_methods_with_reverse = (
+    "add",
+    "sub",
+    "mul",
+    "floordiv",
+    "div",
+    "truediv",
+    "mod",
+    "pow",
+    "lshift",
+    "rshift",
+    "and",
+    "or",
+    "xor",
+)
+pointwise_magic_methods = (
+    *(x for m in pointwise_magic_methods_with_reverse for x in (m, "r" + m)),
+    "eq",
+    "gt",
+    "le",
+    "lt",
+    "ge",
+    "gt",
+    "ne",
+    "neg",
+    "pos",
+    "abs",
+    "invert",
+    "iadd",
+    "isub",
+    "imul",
+    "ifloordiv",
+    "idiv",
+    "itruediv",
+    "imod",
+    "ipow",
+    "ilshift",
+    "irshift",
+    "iand",
+    "ior",
+    "ixor",
+    "int",
+    "long",
+    "float",
+    "complex",
+)
+
+pointwise_methods = (*(f"__{m}__" for m in pointwise_magic_methods),)
+
+pointwise = (
+    *(getattr(torch.Tensor, m) for m in pointwise_methods),
+    torch.nn.functional.dropout,
+    torch.where,
+    torch.Tensor.abs,
+    torch.abs,
+    torch.Tensor.acos,
+    torch.acos,
+    torch.Tensor.acosh,
+    torch.acosh,
+    torch.Tensor.add,
+    torch.add,
+    torch.Tensor.addcdiv,
+    torch.addcdiv,
+    torch.Tensor.addcmul,
+    torch.addcmul,
+    torch.Tensor.addr,
+    torch.addr,
+    torch.Tensor.angle,
+    torch.angle,
+    torch.Tensor.asin,
+    torch.asin,
+    torch.Tensor.asinh,
+    torch.asinh,
+    torch.Tensor.atan,
+    torch.atan,
+    torch.Tensor.atan2,
+    torch.atan2,
+    torch.Tensor.atanh,
+    torch.atanh,
+    torch.Tensor.bitwise_and,
+    torch.bitwise_and,
+    torch.Tensor.bitwise_left_shift,
+    torch.bitwise_left_shift,
+    torch.Tensor.bitwise_not,
+    torch.bitwise_not,
+    torch.Tensor.bitwise_or,
+    torch.bitwise_or,
+    torch.Tensor.bitwise_right_shift,
+    torch.bitwise_right_shift,
+    torch.Tensor.bitwise_xor,
+    torch.bitwise_xor,
+    torch.Tensor.ceil,
+    torch.ceil,
+    torch.celu,
+    torch.nn.functional.celu,
+    torch.Tensor.clamp,
+    torch.clamp,
+    torch.Tensor.clamp_max,
+    torch.clamp_max,
+    torch.Tensor.clamp_min,
+    torch.clamp_min,
+    torch.Tensor.copysign,
+    torch.copysign,
+    torch.Tensor.cos,
+    torch.cos,
+    torch.Tensor.cosh,
+    torch.cosh,
+    torch.Tensor.deg2rad,
+    torch.deg2rad,
+    torch.Tensor.digamma,
+    torch.digamma,
+    torch.Tensor.div,
+    torch.div,
+    torch.dropout,
+    torch.nn.functional.dropout,
+    torch.nn.functional.elu,
+    torch.Tensor.eq,
+    torch.eq,
+    torch.Tensor.erf,
+    torch.erf,
+    torch.Tensor.erfc,
+    torch.erfc,
+    torch.Tensor.erfinv,
+    torch.erfinv,
+    torch.Tensor.exp,
+    torch.exp,
+    torch.Tensor.exp2,
+    torch.exp2,
+    torch.Tensor.expm1,
+    torch.expm1,
+    torch.feature_dropout,
+    torch.Tensor.float_power,
+    torch.float_power,
+    torch.Tensor.floor,
+    torch.floor,
+    torch.Tensor.floor_divide,
+    torch.floor_divide,
+    torch.Tensor.fmod,
+    torch.fmod,
+    torch.Tensor.frac,
+    torch.frac,
+    torch.Tensor.frexp,
+    torch.frexp,
+    torch.Tensor.gcd,
+    torch.gcd,
+    torch.Tensor.ge,
+    torch.ge,
+    torch.nn.functional.gelu,
+    torch.nn.functional.glu,
+    torch.Tensor.gt,
+    torch.gt,
+    torch.Tensor.hardshrink,
+    torch.hardshrink,
+    torch.nn.functional.hardshrink,
+    torch.nn.functional.hardsigmoid,
+    torch.nn.functional.hardswish,
+    torch.nn.functional.hardtanh,
+    torch.Tensor.heaviside,
+    torch.heaviside,
+    torch.Tensor.hypot,
+    torch.hypot,
+    torch.Tensor.i0,
+    torch.i0,
+    torch.Tensor.igamma,
+    torch.igamma,
+    torch.Tensor.igammac,
+    torch.igammac,
+    torch.Tensor.isclose,
+    torch.isclose,
+    torch.Tensor.isfinite,
+    torch.isfinite,
+    torch.Tensor.isinf,
+    torch.isinf,
+    torch.Tensor.isnan,
+    torch.isnan,
+    torch.Tensor.isneginf,
+    torch.isneginf,
+    torch.Tensor.isposinf,
+    torch.isposinf,
+    torch.Tensor.isreal,
+    torch.isreal,
+    torch.Tensor.kron,
+    torch.kron,
+    torch.Tensor.lcm,
+    torch.lcm,
+    torch.Tensor.ldexp,
+    torch.ldexp,
+    torch.Tensor.le,
+    torch.le,
+    torch.nn.functional.leaky_relu,
+    torch.Tensor.lerp,
+    torch.lerp,
+    torch.Tensor.lgamma,
+    torch.lgamma,
+    torch.Tensor.log,
+    torch.log,
+    torch.Tensor.log10,
+    torch.log10,
+    torch.Tensor.log1p,
+    torch.log1p,
+    torch.Tensor.log2,
+    torch.log2,
+    torch.nn.functional.logsigmoid,
+    torch.Tensor.logical_and,
+    torch.logical_and,
+    torch.Tensor.logical_not,
+    torch.logical_not,
+    torch.Tensor.logical_or,
+    torch.logical_or,
+    torch.Tensor.logical_xor,
+    torch.logical_xor,
+    torch.Tensor.logit,
+    torch.logit,
+    torch.Tensor.lt,
+    torch.lt,
+    torch.Tensor.maximum,
+    torch.maximum,
+    torch.Tensor.minimum,
+    torch.minimum,
+    torch.nn.functional.mish,
+    torch.Tensor.mvlgamma,
+    torch.mvlgamma,
+    torch.Tensor.nan_to_num,
+    torch.nan_to_num,
+    torch.Tensor.ne,
+    torch.ne,
+    torch.Tensor.neg,
+    torch.neg,
+    torch.Tensor.nextafter,
+    torch.nextafter,
+    torch.Tensor.outer,
+    torch.outer,
+    torch.polar,
+    torch.Tensor.polygamma,
+    torch.polygamma,
+    torch.Tensor.positive,
+    torch.positive,
+    torch.Tensor.pow,
+    torch.pow,
+    torch.Tensor.prelu,
+    torch.prelu,
+    torch.nn.functional.prelu,
+    torch.Tensor.rad2deg,
+    torch.rad2deg,
+    torch.Tensor.reciprocal,
+    torch.reciprocal,
+    torch.Tensor.relu,
+    torch.relu,
+    torch.nn.functional.relu,
+    torch.nn.functional.relu6,
+    torch.Tensor.remainder,
+    torch.remainder,
+    torch.Tensor.round,
+    torch.round,
+    torch.rrelu,
+    torch.nn.functional.rrelu,
+    torch.Tensor.rsqrt,
+    torch.rsqrt,
+    torch.rsub,
+    torch.selu,
+    torch.nn.functional.selu,
+    torch.Tensor.sgn,
+    torch.sgn,
+    torch.Tensor.sigmoid,
+    torch.sigmoid,
+    torch.nn.functional.sigmoid,
+    torch.Tensor.sign,
+    torch.sign,
+    torch.Tensor.signbit,
+    torch.signbit,
+    torch.nn.functional.silu,
+    torch.Tensor.sin,
+    torch.sin,
+    torch.Tensor.sinc,
+    torch.sinc,
+    torch.Tensor.sinh,
+    torch.sinh,
+    torch.nn.functional.softplus,
+    torch.nn.functional.softshrink,
+    torch.Tensor.sqrt,
+    torch.sqrt,
+    torch.Tensor.square,
+    torch.square,
+    torch.Tensor.sub,
+    torch.sub,
+    torch.Tensor.tan,
+    torch.tan,
+    torch.Tensor.tanh,
+    torch.tanh,
+    torch.nn.functional.tanh,
+    torch.threshold,
+    torch.nn.functional.threshold,
+    torch.trapz,
+    torch.Tensor.true_divide,
+    torch.true_divide,
+    torch.Tensor.trunc,
+    torch.trunc,
+    torch.Tensor.xlogy,
+    torch.xlogy,
+    torch.rand_like,
+)

env-llmeval/lib/python3.10/site-packages/functorch/dim/reference.py

@@ -0,0 +1,645 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+# reference python implementations for C ops
+import torch
+
+from functorch._C import dim as _C
+from . import op_properties
+from .batch_tensor import _enable_layers
+from .tree_map import tree_flatten, tree_map
+
+DimList = _C.DimList
+import operator
+from functools import reduce
+
+
+# use dict to avoid writing C++ bindings for set
+pointwise = set(op_properties.pointwise)
+
+
+def prod(x):
+    return reduce(operator.mul, x, 1)
+
+
+def _wrap_dim(d, N, keepdim):
+    from . import Dim
+
+    if isinstance(d, Dim):
+        assert not keepdim, "cannot preserve first-class dimensions with keepdim=True"
+        return d
+    elif d >= 0:
+        return d - N
+    else:
+        return d
+
+
+def _dims(d, N, keepdim, single_dim):
+    from . import Dim
+
+    if isinstance(d, (Dim, int)):
+        return ltuple((_wrap_dim(d, N, keepdim),))
+    assert not single_dim, f"expected a single dimension or int but found: {d}"
+    return ltuple(_wrap_dim(x, N, keepdim) for x in d)
+
+
+def _bind_dims_to_size(lhs_size, rhs, lhs_debug):
+    from . import DimensionMismatchError
+
+    not_bound = tuple((i, r) for i, r in enumerate(rhs) if not r.is_bound)
+    if len(not_bound) == 1:
+        idx, d = not_bound[0]
+        rhs_so_far = prod(r.size for r in rhs if r.is_bound)
+        if lhs_size % rhs_so_far != 0:
+            rhs_s = tuple("?" if not r.is_bound else str(r.size) for r in rhs)
+            raise DimensionMismatchError(
+                f"inferred dimension does not evenly fit into larger dimension: {lhs_size} vs {rhs_s}"
+            )
+        new_size = lhs_size // rhs_so_far
+        d.size = new_size
+    elif len(not_bound) > 1:
+        rhs_s = tuple("?" if not r.is_bound else str(r.size) for r in rhs)
+        raise DimensionMismatchError(
+            f"cannot infer the size of two dimensions at once: {rhs} with sizes {rhs_s}"
+        )
+    else:
+        rhs_size = prod(r.size for r in rhs)
+        if lhs_size != rhs_size:
+            raise DimensionMismatchError(
+                f"Dimension sizes to do not match ({lhs_size} != {rhs_size}) when matching {lhs_debug} to {rhs}"
+            )
+
+
+def _tensor_levels(inp):
+    from . import _Tensor
+
+    if isinstance(inp, _Tensor):
+        return inp._tensor, llist(inp._levels), inp._has_device
+    else:
+        return inp, llist(range(-inp.ndim, 0)), True
+
+
+def _match_levels(v, from_levels, to_levels):
+    view = []
+    permute = []
+    requires_view = False
+    size = v.size()
+    for t in to_levels:
+        try:
+            idx = from_levels.index(t)
+            permute.append(idx)
+            view.append(size[idx])
+        except ValueError:
+            view.append(1)
+            requires_view = True
+    if permute != list(range(len(permute))):
+        v = v.permute(*permute)
+    if requires_view:
+        v = v.view(*view)
+    return v
+
+
+# make a single dimension positional but do not permute it,
+# used to do multi-tensor operators where the dim being acted on
+# should not physically move if possible
+def _positional_no_permute(self, dim, expand_dim=False):
+    from . import Tensor
+
+    ptensor, levels = self._tensor, llist(self._levels)
+    try:
+        idx = levels.index(dim)
+    except ValueError:
+        if not expand_dim:
+            raise
+        idx = 0
+        ptensor = ptensor.expand(dim.size, *ptensor.size())
+        levels.insert(0, 0)
+    idx_batched = 0
+    for i in range(idx):
+        if isinstance(levels[i], int):
+            levels[i] -= 1
+            idx_batched += 1
+    levels[idx] = -idx_batched - 1
+    return Tensor.from_positional(ptensor, levels, self._has_device), idx_batched
+
+
+def seq(a, b):
+    from . import Dim
+
+    if isinstance(a, Dim) != isinstance(b, Dim):
+        return False
+    if isinstance(a, Dim):
+        return a is b
+    else:
+        return a == b
+
+
+class isin:
+    def __contains__(self, item):
+        for x in self:
+            if seq(item, x):
+                return True
+        return False
+
+    def index(self, item):
+        for i, x in enumerate(self):
+            if seq(item, x):
+                return i
+        raise ValueError
+
+
+class llist(isin, list):
+    pass
+
+
+class ltuple(isin, tuple):
+    pass
+
+
+empty_dict = {}
+
+
+@classmethod
+def __torch_function__(self, orig, cls, args, kwargs=empty_dict):
+    from . import _Tensor, Tensor, TensorLike
+    from .delayed_mul_tensor import DelayedMulTensor
+
+    if orig is torch.Tensor.__mul__:
+        lhs, rhs = args
+        if (
+            isinstance(lhs, _Tensor)
+            and isinstance(rhs, _Tensor)
+            and lhs.ndim == 0
+            and rhs.ndim == 0
+        ):
+            return DelayedMulTensor(lhs, rhs)
+    all_dims = llist()
+    flat_args, unflatten = tree_flatten((args, kwargs))
+    device_holding_tensor = None
+    for f in flat_args:
+        if isinstance(f, _Tensor):
+            if f._has_device:
+                device_holding_tensor = f._batchtensor
+            for d in f.dims:
+                if d not in all_dims:
+                    all_dims.append(d)
+
+    def unwrap(t):
+        if isinstance(t, _Tensor):
+            r = t._batchtensor
+            if device_holding_tensor is not None and not t._has_device:
+                r = r.to(device=device_holding_tensor.device)
+            return r
+        return t
+
+    if orig in pointwise:
+        result_levels = llist()
+        arg_levels = llist()
+        to_expand = []
+        for i, f in enumerate(flat_args):
+            if isinstance(f, TensorLike):
+                ptensor, levels, _ = _tensor_levels(f)
+                if (
+                    isinstance(f, _Tensor)
+                    and not f._has_device
+                    and device_holding_tensor is not None
+                ):
+                    ptensor = ptensor.to(device=device_holding_tensor.device)
+                flat_args[i] = ptensor
+                for l in levels:
+                    if l not in result_levels:
+                        result_levels.append(l)
+                to_expand.append((i, levels))
+
+        for i, levels in to_expand:
+            flat_args[i] = _match_levels(flat_args[i], levels, result_levels)
+        args, kwargs = unflatten(flat_args)
+        result = orig(*args, **kwargs)
+
+        def wrap(t):
+            if isinstance(t, TensorLike):
+                return Tensor.from_positional(
+                    t, result_levels, device_holding_tensor is not None
+                )
+            return t
+
+        return tree_map(wrap, result)
+    else:
+
+        def wrap(t):
+            if isinstance(t, TensorLike):
+                return Tensor.from_batched(t, device_holding_tensor is not None)
+            return t
+
+        with _enable_layers(all_dims):
+            print(f"batch_tensor for {orig}")
+            args, kwargs = unflatten(unwrap(f) for f in flat_args)
+            result = orig(*args, **kwargs)
+            # print("END", orig)
+            return tree_map(wrap, result)
+
+
+def positional(self, *dims):
+    from . import Dim, Tensor
+
+    ptensor, levels = self._tensor, llist(self._levels)
+    flat_dims = llist()
+    view = []
+    needs_view = False
+    ndim = self.ndim
+    for d in dims:
+        if isinstance(d, DimList):
+            flat_dims.extend(d)
+            view.extend(e.size for e in d)
+        elif isinstance(d, Dim):
+            flat_dims.append(d)
+            view.append(d.size)
+        elif isinstance(d, int):
+            d = _wrap_dim(d, ndim, False)
+            flat_dims.append(d)
+            view.append(ptensor.size(d))
+        else:
+            flat_dims.extend(d)
+            view.append(prod(e.size for e in d))
+            needs_view = True
+
+    permute = list(range(len(levels)))
+    nflat = len(flat_dims)
+    for i, d in enumerate(flat_dims):
+        try:
+            idx = levels.index(d)
+        except ValueError as e:
+            raise DimensionBindError(
+                f"tensor of dimensions {self.dims} does not contain dim {d}"
+            ) from e
+        p = permute[idx]
+        del levels[idx]
+        del permute[idx]
+        levels.insert(i, 0)
+        permute.insert(i, p)
+    ptensor = ptensor.permute(*permute)
+    seen = 0
+    for i in range(len(levels) - 1, -1, -1):
+        if isinstance(levels[i], int):
+            seen += 1
+            levels[i] = -seen
+    result = Tensor.from_positional(ptensor, levels, self._has_device)
+    if needs_view:
+        result = result.reshape(*view, *result.size()[len(flat_dims) :])
+    return result
+
+
+def _contains_dim(input):
+    from . import Dim
+
+    for i in input:
+        if isinstance(i, Dim):
+            return True
+
+
+def expand(self, *sizes):
+    if not _contains_dim(sizes):
+        return self.__torch_function__(torch.Tensor.expand, None, (self, *sizes))
+    dims = sizes
+    sizes = [d.size for d in dims] + [-1] * self.ndim
+    self = self.expand(*sizes)
+    return self[dims]
+
+
+_not_present = object()
+
+
+def _getarg(name, offset, args, kwargs, default):
+    if len(args) > offset:
+        return args[offset]
+    return kwargs.get(name, default)
+
+
+def _patcharg(name, offset, args, kwargs, value):
+    if len(args) > offset:
+        args[offset] = value
+    else:
+        kwargs[name] = value
+
+
+def _wrap(
+    orig, dim_offset=0, keepdim_offset=1, dim_name="dim", single_dim=False, reduce=True
+):
+    from . import Dim, Tensor, TensorLike
+
+    def fn(self, *args, **kwargs):
+        dim = _getarg(dim_name, dim_offset, args, kwargs, _not_present)
+        if dim is _not_present or (single_dim and not isinstance(dim, Dim)):
+            with _enable_layers(self.dims):
+                print(f"dim fallback batch_tensor for {orig}")
+                return Tensor.from_batched(
+                    orig(self._batchtensor, *args, **kwargs), self._has_device
+                )
+        keepdim = (
+            _getarg("keepdim", keepdim_offset, args, kwargs, False) if reduce else False
+        )
+        t, levels = self._tensor, llist(self._levels)
+        dims = _dims(dim, self._batchtensor.ndim, keepdim, single_dim)
+        dim_indices = tuple(levels.index(d) for d in dims)
+        if reduce and not keepdim:
+            new_levels = [l for i, l in enumerate(levels) if i not in dim_indices]
+        else:
+            new_levels = levels
+
+        if len(dim_indices) == 1:
+            dim_indices = dim_indices[
+                0
+            ]  # so that dims that really only take a single argument work...
+        args = list(args)
+        _patcharg(dim_name, dim_offset, args, kwargs, dim_indices)
+
+        def wrap(t):
+            if isinstance(t, TensorLike):
+                return Tensor.from_positional(t, new_levels, self._has_device)
+            return t
+
+        with _enable_layers(new_levels):
+            print(f"dim used batch_tensor for {orig}")
+            r = orig(t, *args, **kwargs)
+            return tree_map(wrap, r)
+
+    return fn
+
+
+def _def(name, *args, **kwargs):
+    from . import _Tensor
+
+    orig = getattr(torch.Tensor, name)
+    setattr(_Tensor, name, _wrap(orig, *args, **kwargs))
+
+
+no_slice = slice(None)
+
+_orig_getitem = torch.Tensor.__getitem__
+
+
+class dim_tracker:
+    def __init__(self):
+        self.dims = llist()
+        self.count = []
+
+    def record(self, d):
+        if d not in self.dims:
+            self.dims.append(d)
+            self.count.append(1)
+
+    def __getitem__(self, d):
+        return self.count[self.dims.index(d)]
+
+
+def t__getitem__(self, input):
+    from . import _Tensor, Dim, DimensionBindError, DimList, Tensor, TensorLike
+
+    # * bail to original example if we have a single non-Dim tensor, or a non-tensor
+    # * locate ... or an unbound tensor list, and determine its size, bind dim list
+    #   (remember that None does not count to the total dim count)
+    # * bind simple dims and dim-packs to their sizes, count the number of uses of each dim,
+    #   produce the re-view if needed
+    # * for each single-use dim index, replace with no_slice and mark that it will be added
+    #   (keep track of whether we have to call super)
+    # * call super if needed
+    # * if we have dims to bind, bind them (it will help if we eliminated ... and None before)
+
+    # this handles bool indexing handling, as well as some other simple cases.
+
+    is_simple = (
+        not isinstance(input, Dim)
+        and not isinstance(input, (tuple, list))
+        and
+        # WAR for functorch bug where zero time tensors in getitem are not handled correctly.
+        not (isinstance(input, TensorLike) and input.ndim == 0)
+    )
+
+    if is_simple:
+        if isinstance(self, _Tensor):
+            return _Tensor.__torch_function__(_orig_getitem, None, (self, input))
+        else:
+            return _orig_getitem(self, input)
+
+    # can further optimize this case
+    if not isinstance(input, tuple):
+        input = [input]
+    else:
+        input = list(input)
+
+    dims_indexed = 0
+    expanding_object = None
+    dimlists = []
+    for i, s in enumerate(input):
+        if s is ... or isinstance(s, DimList) and not s.is_bound:
+            if expanding_object is not None:
+                msg = (
+                    "at most one ... or unbound dimension list can exist in indexing list but"
+                    f" found 2 at offsets {i} and {expanding_object}"
+                )
+                raise DimensionBindError(msg)
+            expanding_object = i
+
+        if isinstance(s, DimList):
+            dims_indexed += len(s) if s.is_bound else 0
+            dimlists.append(i)
+        elif s is not None and s is not ...:
+            dims_indexed += 1
+
+    ndim = self.ndim
+    if dims_indexed > ndim:
+        raise IndexError(
+            f"at least {dims_indexed} indices were supplied but the tensor only has {ndim} dimensions."
+        )
+    if expanding_object is not None:
+        expanding_ndims = ndim - dims_indexed
+        obj = input[expanding_object]
+        if obj is ...:
+            input[expanding_object : expanding_object + 1] = [
+                no_slice
+            ] * expanding_ndims
+        else:
+            obj.bind_len(expanding_ndims)
+    # flatten the dimslists into the indexing
+    for i in reversed(dimlists):
+        input[i : i + 1] = input[i]
+    dims_indexed = 0
+    requires_view = False
+    size = self.size()
+    view_sizes = []
+    dims_seen = dim_tracker()
+
+    def add_dims(t):
+        if not isinstance(t, _Tensor):
+            return
+        for d in t.dims:
+            dims_seen.record(d)
+
+    add_dims(self)
+    dim_packs = []
+    for i, idx in enumerate(input):
+        if idx is None:
+            input[i] = no_slice
+            view_sizes.append(1)
+            requires_view = True
+        else:
+            sz = size[dims_indexed]
+            if isinstance(idx, Dim):
+                idx.size = sz
+                dims_seen.record(idx)
+                view_sizes.append(sz)
+            elif isinstance(idx, (tuple, list)) and idx and isinstance(idx[0], Dim):
+                for d in idx:
+                    dims_seen.record(idx)
+                _bind_dims_to_size(sz, idx, f"offset {i}")
+                view_sizes.extend(d.size for d in idx)
+                requires_view = True
+                dim_packs.append(i)
+            else:
+                add_dims(idx)
+                view_sizes.append(sz)
+            dims_indexed += 1
+    if requires_view:
+        self = self.view(*view_sizes)
+    for i in reversed(dim_packs):
+        input[i : i + 1] = input[i]
+
+    # currenty:
+    # input is flat, containing either Dim, or Tensor, or something valid for standard indexing
+    # self may have first-class dims as well.
+
+    # to index:
+    # drop the first class dims from self, they just become direct indices of their positions
+
+    # figure out the dimensions of the indexing tensors: union of all the dims in the tensors in the index.
+    # these dimensions will appear and need to be bound at the first place tensor occures
+
+    if isinstance(self, _Tensor):
+        ptensor_self, levels = self._tensor, list(self._levels)
+        # indices to ptensor rather than self which has first-class dimensions
+        input_it = iter(input)
+        flat_inputs = [next(input_it) if isinstance(l, int) else l for l in levels]
+        has_device = self._has_device
+        to_pad = 0
+    else:
+        ptensor_self, flat_inputs = self, input
+        to_pad = ptensor_self.ndim - len(flat_inputs)
+        has_device = True
+
+    result_levels = []
+    index_levels = []
+    tensor_insert_point = None
+    to_expand = {}
+    requires_getindex = False
+    for i, inp in enumerate(flat_inputs):
+        if isinstance(inp, Dim) and dims_seen[inp] == 1:
+            flat_inputs[i] = no_slice
+            result_levels.append(inp)
+        elif isinstance(inp, TensorLike):
+            requires_getindex = True
+            if tensor_insert_point is None:
+                tensor_insert_point = len(result_levels)
+            ptensor, levels, _ = _tensor_levels(inp)
+            to_expand[i] = levels
+            flat_inputs[i] = ptensor
+            for l in levels:
+                if l not in index_levels:
+                    index_levels.append(l)
+        else:
+            requires_getindex = True
+            result_levels.append(0)
+
+    if tensor_insert_point is not None:
+        result_levels[tensor_insert_point:tensor_insert_point] = index_levels
+
+    for i, levels in to_expand.items():
+        flat_inputs[i] = _match_levels(flat_inputs[i], levels, index_levels)
+
+    if requires_getindex:
+        result = _orig_getitem(ptensor_self, flat_inputs)
+    else:
+        result = ptensor_self
+
+    next_positional = -1
+    if to_pad > 0:
+        result_levels.extend([0] * to_pad)
+    for i, r in enumerate(reversed(result_levels)):
+        if isinstance(r, int):
+            result_levels[-1 - i] = next_positional
+            next_positional -= 1
+
+    return Tensor.from_positional(result, result_levels, has_device)
+
+
+# XXX - dim is optional and can be the outer-most dimension...
+def stack(tensors, new_dim, dim=0, out=None):
+    if isinstance(dim, int):
+        return torch.stack(tensors, dim, out).index(dim, new_dim)
+    index = None
+    if out is not None:
+        out, index = _positional_no_permute(out, dim, expand_dim=True)
+    ptensors = []
+    for t in tensors:
+        pt, pi = _positional_no_permute(t, dim, expand_dim=True)
+        if index is not None and pi != index:
+            pt = pt.move_dim(pi, index)
+        else:
+            index = pi
+        ptensors.append(pt)
+    pr = torch.stack(ptensors, index, out=out)
+    return pr.index((index, index + 1), (new_dim, dim))
+
+
+_orig_split = torch.Tensor.split
+
+
+def split(self, split_size_or_sections, dim=0):
+    from . import _Tensor, Dim
+
+    if isinstance(split_size_or_sections, int) or any(
+        isinstance(t, int) for t in split_size_or_sections
+    ):
+        if isinstance(dim, Dim):
+            raise ValueError(
+                "when dim is specified as a Dim object, split sizes must also be dimensions."
+            )
+        return _orig_split(self, split_size_or_sections, dim=dim)
+
+    if isinstance(dim, Dim):
+        assert isinstance(self, _Tensor), f"Tensor does not have dimension {dim}"
+        self, dim = _positional_no_permute(self, dim)
+
+    size = self.size(dim)
+    total_bound_size = 0
+    unbound = []
+    sizes = []
+    for i, d in enumerate(split_size_or_sections):
+        if d.is_bound:
+            sizes.append(d.size)
+            total_bound_size += d.size
+        else:
+            sizes.append(0)
+            unbound.append(i)
+
+    if unbound:
+        assert (
+            total_bound_size <= size
+        ), f"result dimensions are larger than original: {total_bound_size} vs {size} ({split_size_or_sections})"
+        remaining_size = size - total_bound_size
+        chunk_size = -(-remaining_size // len(unbound))
+        for u in unbound:
+            sz = min(chunk_size, remaining_size)
+            split_size_or_sections[u].size = sz
+            sizes[u] = sz
+            remaining_size -= sz
+    else:
+        assert (
+            total_bound_size == size
+        ), f"result dimensions do not match original: {total_bound_size} vs {size} ({split_size_or_sections})"
+    return tuple(
+        t.index(dim, d)
+        for d, t in zip(split_size_or_sections, _orig_split(self, sizes, dim=dim))
+    )

env-llmeval/lib/python3.10/site-packages/functorch/dim/tree_map.py

@@ -0,0 +1,14 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from functorch._C import dim
+
+tree_flatten = dim.tree_flatten
+
+
+def tree_map(fn, tree):
+    vs, unflatten = tree_flatten(tree)
+    return unflatten(fn(v) for v in vs)

env-llmeval/lib/python3.10/site-packages/functorch/dim/wrap_type.py

@@ -0,0 +1,71 @@
+# Copyright (c) Facebook, Inc. and its affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from types import (
+    BuiltinMethodType,
+    FunctionType,
+    GetSetDescriptorType,
+    MethodDescriptorType,
+    WrapperDescriptorType,
+)
+
+from functorch._C import dim as _C
+
+_wrap_method = _C._wrap_method
+
+FUNC_TYPES = (
+    FunctionType,
+    MethodDescriptorType,
+    BuiltinMethodType,
+    WrapperDescriptorType,
+)
+PROPERTY_TYPES = (GetSetDescriptorType, property)
+
+
+def _py_wrap_method(orig, __torch_function__):
+    def impl(*args, **kwargs):
+        return __torch_function__(orig, None, args, kwargs)
+
+    return impl
+
+
+def wrap_type(use_c, to_patch, pattern, __torch_function__):
+    if use_c:
+        wrap_method = _wrap_method
+    else:
+        wrap_method = _py_wrap_method
+
+    all = {}
+    for t in reversed(pattern.mro()[:-1]):  # skip object
+        all.update(t.__dict__)
+
+    def wrap_attr(orig):
+        return property(wrap_method(orig.__get__, __torch_function__))
+
+    for name, obj in all.items():
+        if name in (
+            "__dict__",
+            "__new__",
+            "__init__",
+            "__repr__",
+            "__weakref__",
+            "__doc__",
+            "__module__",
+            "__dir__",
+        ):
+            continue
+
+        # skip things that have been overloaded
+        # things that come from object like `__eq__` still need to be patched, however.
+        if hasattr(to_patch, name) and getattr(to_patch, name) is not getattr(
+            object, name, None
+        ):
+            continue
+
+        if isinstance(obj, FUNC_TYPES):
+            setattr(to_patch, name, wrap_method(obj, __torch_function__))
+        elif isinstance(obj, PROPERTY_TYPES):
+            setattr(to_patch, name, wrap_attr(obj))

env-llmeval/lib/python3.10/site-packages/nvidia/cuda_cupti/lib/libcheckpoint.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1626ff119582bca46605bc6d49769ab75314b9993dd647bd64a90dec747bc843
+size 1534104

env-llmeval/lib/python3.10/site-packages/nvidia/cuda_cupti/lib/libnvperf_target.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa2587c8d211fbc85e8b88cca0bcebe78c8cc40c81b0c3763ce57ac9e63f0669
+size 5895416

env-llmeval/lib/python3.10/site-packages/nvidia/cuda_nvrtc/lib/libnvrtc.so.12

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83ec9ad7775e89f6280286ba11eb9d28cafe49c2f777a3e051bcc881de7449fc
+size 56875328

env-llmeval/lib/python3.10/site-packages/pytz/lazy.py

@@ -0,0 +1,172 @@
+from threading import RLock
+try:
+    from collections.abc import Mapping as DictMixin
+except ImportError:  # Python < 3.3
+    try:
+        from UserDict import DictMixin  # Python 2
+    except ImportError:  # Python 3.0-3.3
+        from collections import Mapping as DictMixin
+
+
+# With lazy loading, we might end up with multiple threads triggering
+# it at the same time. We need a lock.
+_fill_lock = RLock()
+
+
+class LazyDict(DictMixin):
+    """Dictionary populated on first use."""
+    data = None
+
+    def __getitem__(self, key):
+        if self.data is None:
+            _fill_lock.acquire()
+            try:
+                if self.data is None:
+                    self._fill()
+            finally:
+                _fill_lock.release()
+        return self.data[key.upper()]
+
+    def __contains__(self, key):
+        if self.data is None:
+            _fill_lock.acquire()
+            try:
+                if self.data is None:
+                    self._fill()
+            finally:
+                _fill_lock.release()
+        return key in self.data
+
+    def __iter__(self):
+        if self.data is None:
+            _fill_lock.acquire()
+            try:
+                if self.data is None:
+                    self._fill()
+            finally:
+                _fill_lock.release()
+        return iter(self.data)
+
+    def __len__(self):
+        if self.data is None:
+            _fill_lock.acquire()
+            try:
+                if self.data is None:
+                    self._fill()
+            finally:
+                _fill_lock.release()
+        return len(self.data)
+
+    def keys(self):
+        if self.data is None:
+            _fill_lock.acquire()
+            try:
+                if self.data is None:
+                    self._fill()
+            finally:
+                _fill_lock.release()
+        return self.data.keys()
+
+
+class LazyList(list):
+    """List populated on first use."""
+
+    _props = [
+        '__str__', '__repr__', '__unicode__',
+        '__hash__', '__sizeof__', '__cmp__',
+        '__lt__', '__le__', '__eq__', '__ne__', '__gt__', '__ge__',
+        'append', 'count', 'index', 'extend', 'insert', 'pop', 'remove',
+        'reverse', 'sort', '__add__', '__radd__', '__iadd__', '__mul__',
+        '__rmul__', '__imul__', '__contains__', '__len__', '__nonzero__',
+        '__getitem__', '__setitem__', '__delitem__', '__iter__',
+        '__reversed__', '__getslice__', '__setslice__', '__delslice__']
+
+    def __new__(cls, fill_iter=None):
+
+        if fill_iter is None:
+            return list()
+
+        # We need a new class as we will be dynamically messing with its
+        # methods.
+        class LazyList(list):
+            pass
+
+        fill_iter = [fill_iter]
+
+        def lazy(name):
+            def _lazy(self, *args, **kw):
+                _fill_lock.acquire()
+                try:
+                    if len(fill_iter) > 0:
+                        list.extend(self, fill_iter.pop())
+                        for method_name in cls._props:
+                            delattr(LazyList, method_name)
+                finally:
+                    _fill_lock.release()
+                return getattr(list, name)(self, *args, **kw)
+            return _lazy
+
+        for name in cls._props:
+            setattr(LazyList, name, lazy(name))
+
+        new_list = LazyList()
+        return new_list
+
+# Not all versions of Python declare the same magic methods.
+# Filter out properties that don't exist in this version of Python
+# from the list.
+LazyList._props = [prop for prop in LazyList._props if hasattr(list, prop)]
+
+
+class LazySet(set):
+    """Set populated on first use."""
+
+    _props = (
+        '__str__', '__repr__', '__unicode__',
+        '__hash__', '__sizeof__', '__cmp__',
+        '__lt__', '__le__', '__eq__', '__ne__', '__gt__', '__ge__',
+        '__contains__', '__len__', '__nonzero__',
+        '__getitem__', '__setitem__', '__delitem__', '__iter__',
+        '__sub__', '__and__', '__xor__', '__or__',
+        '__rsub__', '__rand__', '__rxor__', '__ror__',
+        '__isub__', '__iand__', '__ixor__', '__ior__',
+        'add', 'clear', 'copy', 'difference', 'difference_update',
+        'discard', 'intersection', 'intersection_update', 'isdisjoint',
+        'issubset', 'issuperset', 'pop', 'remove',
+        'symmetric_difference', 'symmetric_difference_update',
+        'union', 'update')
+
+    def __new__(cls, fill_iter=None):
+
+        if fill_iter is None:
+            return set()
+
+        class LazySet(set):
+            pass
+
+        fill_iter = [fill_iter]
+
+        def lazy(name):
+            def _lazy(self, *args, **kw):
+                _fill_lock.acquire()
+                try:
+                    if len(fill_iter) > 0:
+                        for i in fill_iter.pop():
+                            set.add(self, i)
+                        for method_name in cls._props:
+                            delattr(LazySet, method_name)
+                finally:
+                    _fill_lock.release()
+                return getattr(set, name)(self, *args, **kw)
+            return _lazy
+
+        for name in cls._props:
+            setattr(LazySet, name, lazy(name))
+
+        new_set = LazySet()
+        return new_set
+
+# Not all versions of Python declare the same magic methods.
+# Filter out properties that don't exist in this version of Python
+# from the list.
+LazySet._props = [prop for prop in LazySet._props if hasattr(set, prop)]

env-llmeval/lib/python3.10/site-packages/pytz/reference.py

@@ -0,0 +1,140 @@
+'''
+Reference tzinfo implementations from the Python docs.
+Used for testing against as they are only correct for the years
+1987 to 2006. Do not use these for real code.
+'''
+
+from datetime import tzinfo, timedelta, datetime
+from pytz import HOUR, ZERO, UTC
+
+__all__ = [
+    'FixedOffset',
+    'LocalTimezone',
+    'USTimeZone',
+    'Eastern',
+    'Central',
+    'Mountain',
+    'Pacific',
+    'UTC'
+]
+
+
+# A class building tzinfo objects for fixed-offset time zones.
+# Note that FixedOffset(0, "UTC") is a different way to build a
+# UTC tzinfo object.
+class FixedOffset(tzinfo):
+    """Fixed offset in minutes east from UTC."""
+
+    def __init__(self, offset, name):
+        self.__offset = timedelta(minutes=offset)
+        self.__name = name
+
+    def utcoffset(self, dt):
+        return self.__offset
+
+    def tzname(self, dt):
+        return self.__name
+
+    def dst(self, dt):
+        return ZERO
+
+
+import time as _time
+
+STDOFFSET = timedelta(seconds=-_time.timezone)
+if _time.daylight:
+    DSTOFFSET = timedelta(seconds=-_time.altzone)
+else:
+    DSTOFFSET = STDOFFSET
+
+DSTDIFF = DSTOFFSET - STDOFFSET
+
+
+# A class capturing the platform's idea of local time.
+class LocalTimezone(tzinfo):
+
+    def utcoffset(self, dt):
+        if self._isdst(dt):
+            return DSTOFFSET
+        else:
+            return STDOFFSET
+
+    def dst(self, dt):
+        if self._isdst(dt):
+            return DSTDIFF
+        else:
+            return ZERO
+
+    def tzname(self, dt):
+        return _time.tzname[self._isdst(dt)]
+
+    def _isdst(self, dt):
+        tt = (dt.year, dt.month, dt.day,
+              dt.hour, dt.minute, dt.second,
+              dt.weekday(), 0, -1)
+        stamp = _time.mktime(tt)
+        tt = _time.localtime(stamp)
+        return tt.tm_isdst > 0
+
+Local = LocalTimezone()
+
+
+def first_sunday_on_or_after(dt):
+    days_to_go = 6 - dt.weekday()
+    if days_to_go:
+        dt += timedelta(days_to_go)
+    return dt
+
+
+# In the US, DST starts at 2am (standard time) on the first Sunday in April.
+DSTSTART = datetime(1, 4, 1, 2)
+# and ends at 2am (DST time; 1am standard time) on the last Sunday of Oct.
+# which is the first Sunday on or after Oct 25.
+DSTEND = datetime(1, 10, 25, 1)
+
+
+# A complete implementation of current DST rules for major US time zones.
+class USTimeZone(tzinfo):
+
+    def __init__(self, hours, reprname, stdname, dstname):
+        self.stdoffset = timedelta(hours=hours)
+        self.reprname = reprname
+        self.stdname = stdname
+        self.dstname = dstname
+
+    def __repr__(self):
+        return self.reprname
+
+    def tzname(self, dt):
+        if self.dst(dt):
+            return self.dstname
+        else:
+            return self.stdname
+
+    def utcoffset(self, dt):
+        return self.stdoffset + self.dst(dt)
+
+    def dst(self, dt):
+        if dt is None or dt.tzinfo is None:
+            # An exception may be sensible here, in one or both cases.
+            # It depends on how you want to treat them.  The default
+            # fromutc() implementation (called by the default astimezone()
+            # implementation) passes a datetime with dt.tzinfo is self.
+            return ZERO
+        assert dt.tzinfo is self
+
+        # Find first Sunday in April & the last in October.
+        start = first_sunday_on_or_after(DSTSTART.replace(year=dt.year))
+        end = first_sunday_on_or_after(DSTEND.replace(year=dt.year))
+
+        # Can't compare naive to aware objects, so strip the timezone from
+        # dt first.
+        if start <= dt.replace(tzinfo=None) < end:
+            return HOUR
+        else:
+            return ZERO
+
+Eastern = USTimeZone(-5, "Eastern", "EST", "EDT")
+Central = USTimeZone(-6, "Central", "CST", "CDT")
+Mountain = USTimeZone(-7, "Mountain", "MST", "MDT")
+Pacific = USTimeZone(-8, "Pacific", "PST", "PDT")

env-llmeval/lib/python3.10/site-packages/pytz/tzinfo.py

@@ -0,0 +1,580 @@
+'''Base classes and helpers for building zone specific tzinfo classes'''
+
+from datetime import datetime, timedelta, tzinfo
+from bisect import bisect_right
+try:
+    set
+except NameError:
+    from sets import Set as set
+
+import pytz
+from pytz.exceptions import AmbiguousTimeError, NonExistentTimeError
+
+__all__ = []
+
+_timedelta_cache = {}
+
+
+def memorized_timedelta(seconds):
+    '''Create only one instance of each distinct timedelta'''
+    try:
+        return _timedelta_cache[seconds]
+    except KeyError:
+        delta = timedelta(seconds=seconds)
+        _timedelta_cache[seconds] = delta
+        return delta
+
+
+_epoch = datetime(1970, 1, 1, 0, 0) # datetime.utcfromtimestamp(0)
+_datetime_cache = {0: _epoch}
+
+
+def memorized_datetime(seconds):
+    '''Create only one instance of each distinct datetime'''
+    try:
+        return _datetime_cache[seconds]
+    except KeyError:
+        # NB. We can't just do datetime.fromtimestamp(seconds, tz=timezone.utc).replace(tzinfo=None)
+        # as this fails with negative values under Windows (Bug #90096)
+        dt = _epoch + timedelta(seconds=seconds)
+        _datetime_cache[seconds] = dt
+        return dt
+
+
+_ttinfo_cache = {}
+
+
+def memorized_ttinfo(*args):
+    '''Create only one instance of each distinct tuple'''
+    try:
+        return _ttinfo_cache[args]
+    except KeyError:
+        ttinfo = (
+            memorized_timedelta(args[0]),
+            memorized_timedelta(args[1]),
+            args[2]
+        )
+        _ttinfo_cache[args] = ttinfo
+        return ttinfo
+
+
+_notime = memorized_timedelta(0)
+
+
+def _to_seconds(td):
+    '''Convert a timedelta to seconds'''
+    return td.seconds + td.days * 24 * 60 * 60
+
+
+class BaseTzInfo(tzinfo):
+    # Overridden in subclass
+    _utcoffset = None
+    _tzname = None
+    zone = None
+
+    def __str__(self):
+        return self.zone
+
+
+class StaticTzInfo(BaseTzInfo):
+    '''A timezone that has a constant offset from UTC
+
+    These timezones are rare, as most locations have changed their
+    offset at some point in their history
+    '''
+    def fromutc(self, dt):
+        '''See datetime.tzinfo.fromutc'''
+        if dt.tzinfo is not None and dt.tzinfo is not self:
+            raise ValueError('fromutc: dt.tzinfo is not self')
+        return (dt + self._utcoffset).replace(tzinfo=self)
+
+    def utcoffset(self, dt, is_dst=None):
+        '''See datetime.tzinfo.utcoffset
+
+        is_dst is ignored for StaticTzInfo, and exists only to
+        retain compatibility with DstTzInfo.
+        '''
+        return self._utcoffset
+
+    def dst(self, dt, is_dst=None):
+        '''See datetime.tzinfo.dst
+
+        is_dst is ignored for StaticTzInfo, and exists only to
+        retain compatibility with DstTzInfo.
+        '''
+        return _notime
+
+    def tzname(self, dt, is_dst=None):
+        '''See datetime.tzinfo.tzname
+
+        is_dst is ignored for StaticTzInfo, and exists only to
+        retain compatibility with DstTzInfo.
+        '''
+        return self._tzname
+
+    def localize(self, dt, is_dst=False):
+        '''Convert naive time to local time'''
+        if dt.tzinfo is not None:
+            raise ValueError('Not naive datetime (tzinfo is already set)')
+        return dt.replace(tzinfo=self)
+
+    def normalize(self, dt, is_dst=False):
+        '''Correct the timezone information on the given datetime.
+
+        This is normally a no-op, as StaticTzInfo timezones never have
+        ambiguous cases to correct:
+
+        >>> from pytz import timezone
+        >>> gmt = timezone('GMT')
+        >>> isinstance(gmt, StaticTzInfo)
+        True
+        >>> dt = datetime(2011, 5, 8, 1, 2, 3, tzinfo=gmt)
+        >>> gmt.normalize(dt) is dt
+        True
+
+        The supported method of converting between timezones is to use
+        datetime.astimezone(). Currently normalize() also works:
+
+        >>> la = timezone('America/Los_Angeles')
+        >>> dt = la.localize(datetime(2011, 5, 7, 1, 2, 3))
+        >>> fmt = '%Y-%m-%d %H:%M:%S %Z (%z)'
+        >>> gmt.normalize(dt).strftime(fmt)
+        '2011-05-07 08:02:03 GMT (+0000)'
+        '''
+        if dt.tzinfo is self:
+            return dt
+        if dt.tzinfo is None:
+            raise ValueError('Naive time - no tzinfo set')
+        return dt.astimezone(self)
+
+    def __repr__(self):
+        return '<StaticTzInfo %r>' % (self.zone,)
+
+    def __reduce__(self):
+        # Special pickle to zone remains a singleton and to cope with
+        # database changes.
+        return pytz._p, (self.zone,)
+
+
+class DstTzInfo(BaseTzInfo):
+    '''A timezone that has a variable offset from UTC
+
+    The offset might change if daylight saving time comes into effect,
+    or at a point in history when the region decides to change their
+    timezone definition.
+    '''
+    # Overridden in subclass
+
+    # Sorted list of DST transition times, UTC
+    _utc_transition_times = None
+
+    # [(utcoffset, dstoffset, tzname)] corresponding to
+    # _utc_transition_times entries
+    _transition_info = None
+
+    zone = None
+
+    # Set in __init__
+
+    _tzinfos = None
+    _dst = None  # DST offset
+
+    def __init__(self, _inf=None, _tzinfos=None):
+        if _inf:
+            self._tzinfos = _tzinfos
+            self._utcoffset, self._dst, self._tzname = _inf
+        else:
+            _tzinfos = {}
+            self._tzinfos = _tzinfos
+            self._utcoffset, self._dst, self._tzname = (
+                self._transition_info[0])
+            _tzinfos[self._transition_info[0]] = self
+            for inf in self._transition_info[1:]:
+                if inf not in _tzinfos:
+                    _tzinfos[inf] = self.__class__(inf, _tzinfos)
+
+    def fromutc(self, dt):
+        '''See datetime.tzinfo.fromutc'''
+        if (dt.tzinfo is not None and
+                getattr(dt.tzinfo, '_tzinfos', None) is not self._tzinfos):
+            raise ValueError('fromutc: dt.tzinfo is not self')
+        dt = dt.replace(tzinfo=None)
+        idx = max(0, bisect_right(self._utc_transition_times, dt) - 1)
+        inf = self._transition_info[idx]
+        return (dt + inf[0]).replace(tzinfo=self._tzinfos[inf])
+
+    def normalize(self, dt):
+        '''Correct the timezone information on the given datetime
+
+        If date arithmetic crosses DST boundaries, the tzinfo
+        is not magically adjusted. This method normalizes the
+        tzinfo to the correct one.
+
+        To test, first we need to do some setup
+
+        >>> from pytz import timezone
+        >>> utc = timezone('UTC')
+        >>> eastern = timezone('US/Eastern')
+        >>> fmt = '%Y-%m-%d %H:%M:%S %Z (%z)'
+
+        We next create a datetime right on an end-of-DST transition point,
+        the instant when the wallclocks are wound back one hour.
+
+        >>> utc_dt = datetime(2002, 10, 27, 6, 0, 0, tzinfo=utc)
+        >>> loc_dt = utc_dt.astimezone(eastern)
+        >>> loc_dt.strftime(fmt)
+        '2002-10-27 01:00:00 EST (-0500)'
+
+        Now, if we subtract a few minutes from it, note that the timezone
+        information has not changed.
+
+        >>> before = loc_dt - timedelta(minutes=10)
+        >>> before.strftime(fmt)
+        '2002-10-27 00:50:00 EST (-0500)'
+
+        But we can fix that by calling the normalize method
+
+        >>> before = eastern.normalize(before)
+        >>> before.strftime(fmt)
+        '2002-10-27 01:50:00 EDT (-0400)'
+
+        The supported method of converting between timezones is to use
+        datetime.astimezone(). Currently, normalize() also works:
+
+        >>> th = timezone('Asia/Bangkok')
+        >>> am = timezone('Europe/Amsterdam')
+        >>> dt = th.localize(datetime(2011, 5, 7, 1, 2, 3))
+        >>> fmt = '%Y-%m-%d %H:%M:%S %Z (%z)'
+        >>> am.normalize(dt).strftime(fmt)
+        '2011-05-06 20:02:03 CEST (+0200)'
+        '''
+        if dt.tzinfo is None:
+            raise ValueError('Naive time - no tzinfo set')
+
+        # Convert dt in localtime to UTC
+        offset = dt.tzinfo._utcoffset
+        dt = dt.replace(tzinfo=None)
+        dt = dt - offset
+        # convert it back, and return it
+        return self.fromutc(dt)
+
+    def localize(self, dt, is_dst=False):
+        '''Convert naive time to local time.
+
+        This method should be used to construct localtimes, rather
+        than passing a tzinfo argument to a datetime constructor.
+
+        is_dst is used to determine the correct timezone in the ambigous
+        period at the end of daylight saving time.
+
+        >>> from pytz import timezone
+        >>> fmt = '%Y-%m-%d %H:%M:%S %Z (%z)'
+        >>> amdam = timezone('Europe/Amsterdam')
+        >>> dt  = datetime(2004, 10, 31, 2, 0, 0)
+        >>> loc_dt1 = amdam.localize(dt, is_dst=True)
+        >>> loc_dt2 = amdam.localize(dt, is_dst=False)
+        >>> loc_dt1.strftime(fmt)
+        '2004-10-31 02:00:00 CEST (+0200)'
+        >>> loc_dt2.strftime(fmt)
+        '2004-10-31 02:00:00 CET (+0100)'
+        >>> str(loc_dt2 - loc_dt1)
+        '1:00:00'
+
+        Use is_dst=None to raise an AmbiguousTimeError for ambiguous
+        times at the end of daylight saving time
+
+        >>> try:
+        ...     loc_dt1 = amdam.localize(dt, is_dst=None)
+        ... except AmbiguousTimeError:
+        ...     print('Ambiguous')
+        Ambiguous
+
+        is_dst defaults to False
+
+        >>> amdam.localize(dt) == amdam.localize(dt, False)
+        True
+
+        is_dst is also used to determine the correct timezone in the
+        wallclock times jumped over at the start of daylight saving time.
+
+        >>> pacific = timezone('US/Pacific')
+        >>> dt = datetime(2008, 3, 9, 2, 0, 0)
+        >>> ploc_dt1 = pacific.localize(dt, is_dst=True)
+        >>> ploc_dt2 = pacific.localize(dt, is_dst=False)
+        >>> ploc_dt1.strftime(fmt)
+        '2008-03-09 02:00:00 PDT (-0700)'
+        >>> ploc_dt2.strftime(fmt)
+        '2008-03-09 02:00:00 PST (-0800)'
+        >>> str(ploc_dt2 - ploc_dt1)
+        '1:00:00'
+
+        Use is_dst=None to raise a NonExistentTimeError for these skipped
+        times.
+
+        >>> try:
+        ...     loc_dt1 = pacific.localize(dt, is_dst=None)
+        ... except NonExistentTimeError:
+        ...     print('Non-existent')
+        Non-existent
+        '''
+        if dt.tzinfo is not None:
+            raise ValueError('Not naive datetime (tzinfo is already set)')
+
+        # Find the two best possibilities.
+        possible_loc_dt = set()
+        for delta in [timedelta(days=-1), timedelta(days=1)]:
+            loc_dt = dt + delta
+            idx = max(0, bisect_right(
+                self._utc_transition_times, loc_dt) - 1)
+            inf = self._transition_info[idx]
+            tzinfo = self._tzinfos[inf]
+            loc_dt = tzinfo.normalize(dt.replace(tzinfo=tzinfo))
+            if loc_dt.replace(tzinfo=None) == dt:
+                possible_loc_dt.add(loc_dt)
+
+        if len(possible_loc_dt) == 1:
+            return possible_loc_dt.pop()
+
+        # If there are no possibly correct timezones, we are attempting
+        # to convert a time that never happened - the time period jumped
+        # during the start-of-DST transition period.
+        if len(possible_loc_dt) == 0:
+            # If we refuse to guess, raise an exception.
+            if is_dst is None:
+                raise NonExistentTimeError(dt)
+
+            # If we are forcing the pre-DST side of the DST transition, we
+            # obtain the correct timezone by winding the clock forward a few
+            # hours.
+            elif is_dst:
+                return self.localize(
+                    dt + timedelta(hours=6), is_dst=True) - timedelta(hours=6)
+
+            # If we are forcing the post-DST side of the DST transition, we
+            # obtain the correct timezone by winding the clock back.
+            else:
+                return self.localize(
+                    dt - timedelta(hours=6),
+                    is_dst=False) + timedelta(hours=6)
+
+        # If we get this far, we have multiple possible timezones - this
+        # is an ambiguous case occurring during the end-of-DST transition.
+
+        # If told to be strict, raise an exception since we have an
+        # ambiguous case
+        if is_dst is None:
+            raise AmbiguousTimeError(dt)
+
+        # Filter out the possiblilities that don't match the requested
+        # is_dst
+        filtered_possible_loc_dt = [
+            p for p in possible_loc_dt if bool(p.tzinfo._dst) == is_dst
+        ]
+
+        # Hopefully we only have one possibility left. Return it.
+        if len(filtered_possible_loc_dt) == 1:
+            return filtered_possible_loc_dt[0]
+
+        if len(filtered_possible_loc_dt) == 0:
+            filtered_possible_loc_dt = list(possible_loc_dt)
+
+        # If we get this far, we have in a wierd timezone transition
+        # where the clocks have been wound back but is_dst is the same
+        # in both (eg. Europe/Warsaw 1915 when they switched to CET).
+        # At this point, we just have to guess unless we allow more
+        # hints to be passed in (such as the UTC offset or abbreviation),
+        # but that is just getting silly.
+        #
+        # Choose the earliest (by UTC) applicable timezone if is_dst=True
+        # Choose the latest (by UTC) applicable timezone if is_dst=False
+        # i.e., behave like end-of-DST transition
+        dates = {}  # utc -> local
+        for local_dt in filtered_possible_loc_dt:
+            utc_time = (
+                local_dt.replace(tzinfo=None) - local_dt.tzinfo._utcoffset)
+            assert utc_time not in dates
+            dates[utc_time] = local_dt
+        return dates[[min, max][not is_dst](dates)]
+
+    def utcoffset(self, dt, is_dst=None):
+        '''See datetime.tzinfo.utcoffset
+
+        The is_dst parameter may be used to remove ambiguity during DST
+        transitions.
+
+        >>> from pytz import timezone
+        >>> tz = timezone('America/St_Johns')
+        >>> ambiguous = datetime(2009, 10, 31, 23, 30)
+
+        >>> str(tz.utcoffset(ambiguous, is_dst=False))
+        '-1 day, 20:30:00'
+
+        >>> str(tz.utcoffset(ambiguous, is_dst=True))
+        '-1 day, 21:30:00'
+
+        >>> try:
+        ...     tz.utcoffset(ambiguous)
+        ... except AmbiguousTimeError:
+        ...     print('Ambiguous')
+        Ambiguous
+
+        '''
+        if dt is None:
+            return None
+        elif dt.tzinfo is not self:
+            dt = self.localize(dt, is_dst)
+            return dt.tzinfo._utcoffset
+        else:
+            return self._utcoffset
+
+    def dst(self, dt, is_dst=None):
+        '''See datetime.tzinfo.dst
+
+        The is_dst parameter may be used to remove ambiguity during DST
+        transitions.
+
+        >>> from pytz import timezone
+        >>> tz = timezone('America/St_Johns')
+
+        >>> normal = datetime(2009, 9, 1)
+
+        >>> str(tz.dst(normal))
+        '1:00:00'
+        >>> str(tz.dst(normal, is_dst=False))
+        '1:00:00'
+        >>> str(tz.dst(normal, is_dst=True))
+        '1:00:00'
+
+        >>> ambiguous = datetime(2009, 10, 31, 23, 30)
+
+        >>> str(tz.dst(ambiguous, is_dst=False))
+        '0:00:00'
+        >>> str(tz.dst(ambiguous, is_dst=True))
+        '1:00:00'
+        >>> try:
+        ...     tz.dst(ambiguous)
+        ... except AmbiguousTimeError:
+        ...     print('Ambiguous')
+        Ambiguous
+
+        '''
+        if dt is None:
+            return None
+        elif dt.tzinfo is not self:
+            dt = self.localize(dt, is_dst)
+            return dt.tzinfo._dst
+        else:
+            return self._dst
+
+    def tzname(self, dt, is_dst=None):
+        '''See datetime.tzinfo.tzname
+
+        The is_dst parameter may be used to remove ambiguity during DST
+        transitions.
+
+        >>> from pytz import timezone
+        >>> tz = timezone('America/St_Johns')
+
+        >>> normal = datetime(2009, 9, 1)
+
+        >>> tz.tzname(normal)
+        'NDT'
+        >>> tz.tzname(normal, is_dst=False)
+        'NDT'
+        >>> tz.tzname(normal, is_dst=True)
+        'NDT'
+
+        >>> ambiguous = datetime(2009, 10, 31, 23, 30)
+
+        >>> tz.tzname(ambiguous, is_dst=False)
+        'NST'
+        >>> tz.tzname(ambiguous, is_dst=True)
+        'NDT'
+        >>> try:
+        ...     tz.tzname(ambiguous)
+        ... except AmbiguousTimeError:
+        ...     print('Ambiguous')
+        Ambiguous
+        '''
+        if dt is None:
+            return self.zone
+        elif dt.tzinfo is not self:
+            dt = self.localize(dt, is_dst)
+            return dt.tzinfo._tzname
+        else:
+            return self._tzname
+
+    def __repr__(self):
+        if self._dst:
+            dst = 'DST'
+        else:
+            dst = 'STD'
+        if self._utcoffset > _notime:
+            return '<DstTzInfo %r %s+%s %s>' % (
+                self.zone, self._tzname, self._utcoffset, dst
+            )
+        else:
+            return '<DstTzInfo %r %s%s %s>' % (
+                self.zone, self._tzname, self._utcoffset, dst
+            )
+
+    def __reduce__(self):
+        # Special pickle to zone remains a singleton and to cope with
+        # database changes.
+        return pytz._p, (
+            self.zone,
+            _to_seconds(self._utcoffset),
+            _to_seconds(self._dst),
+            self._tzname
+        )
+
+
+def unpickler(zone, utcoffset=None, dstoffset=None, tzname=None):
+    """Factory function for unpickling pytz tzinfo instances.
+
+    This is shared for both StaticTzInfo and DstTzInfo instances, because
+    database changes could cause a zones implementation to switch between
+    these two base classes and we can't break pickles on a pytz version
+    upgrade.
+    """
+    # Raises a KeyError if zone no longer exists, which should never happen
+    # and would be a bug.
+    tz = pytz.timezone(zone)
+
+    # A StaticTzInfo - just return it
+    if utcoffset is None:
+        return tz
+
+    # This pickle was created from a DstTzInfo. We need to
+    # determine which of the list of tzinfo instances for this zone
+    # to use in order to restore the state of any datetime instances using
+    # it correctly.
+    utcoffset = memorized_timedelta(utcoffset)
+    dstoffset = memorized_timedelta(dstoffset)
+    try:
+        return tz._tzinfos[(utcoffset, dstoffset, tzname)]
+    except KeyError:
+        # The particular state requested in this timezone no longer exists.
+        # This indicates a corrupt pickle, or the timezone database has been
+        # corrected violently enough to make this particular
+        # (utcoffset,dstoffset) no longer exist in the zone, or the
+        # abbreviation has been changed.
+        pass
+
+    # See if we can find an entry differing only by tzname. Abbreviations
+    # get changed from the initial guess by the database maintainers to
+    # match reality when this information is discovered.
+    for localized_tz in tz._tzinfos.values():
+        if (localized_tz._utcoffset == utcoffset and
+                localized_tz._dst == dstoffset):
+            return localized_tz
+
+    # This (utcoffset, dstoffset) information has been removed from the
+    # zone. Add it back. This might occur when the database maintainers have
+    # corrected incorrect information. datetime instances using this
+    # incorrect information will continue to do so, exactly as they were
+    # before being pickled. This is purely an overly paranoid safety net - I
+    # doubt this will ever been needed in real life.
+    inf = (utcoffset, dstoffset, tzname)
+    tz._tzinfos[inf] = tz.__class__(inf, tz._tzinfos)
+    return tz._tzinfos[inf]