Skip to content

P
Pytorch Widedeep

Greenplum / Pytorch Widedeep
12 个月 前同步成功

通知 9
Star 1
Fork 0
P
Pytorch Widedeep

体验新版 GitCode，发现更多精彩内容 >>

提交 04d40076 编写于 作者: P Pavol Mulinka
浏览文件
操作

code cleaned + added workaround for ¨ImportError: libGL.so.1: cannot open... 

code cleaned + added workaround for ¨ImportError: libGL.so.1: cannot open shared object file: No such file or directory ``` apt-get update apt-get install ffmpeg libsm6 libxext6  -y ```¨
上级 ab150000
隐藏空白更改
内联 并排
Showing with 83 addition and 118 deletion
.github/workflows/build.yml
浏览文件 @ 04d40076
......@@ -44,6 +44,8 @@ jobs:
python-version: ${{ matrix.python-version }}
- name: Install dependencies
run: |
apt-get update
apt-get install ffmpeg libsm6 libxext6 -y
python -m pip install --upgrade pip
python -m pip install pytest-cov codecov .
if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
......
examples/12_HyperParameter_tuning_w_RayTune.ipynb
浏览文件 @ 04d40076
......@@ -30,7 +30,7 @@
},
{
"cell_type": "code",
"execution_count": 17,
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
......@@ -72,7 +72,7 @@
},
{
"cell_type": "code",
"execution_count": 6,
"execution_count": 2,
"metadata": {},
"outputs": [
{
......@@ -631,7 +631,7 @@
"4 0.68 -0.59 2.0 -36.0 -6.9 2.02 0.14 -0.23 "
]
},
"execution_count": 6,
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
......@@ -644,7 +644,7 @@
},
{
"cell_type": "code",
"execution_count": 7,
"execution_count": 3,
"metadata": {},
"outputs": [
{
......@@ -655,7 +655,7 @@
"Name: target, dtype: int64"
]
},
"execution_count": 7,
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
......@@ -667,7 +667,7 @@
},
{
"cell_type": "code",
"execution_count": 8,
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
......@@ -677,7 +677,7 @@
},
{
"cell_type": "code",
"execution_count": 9,
"execution_count": 5,
"metadata": {},
"outputs": [],
"source": [
......@@ -694,7 +694,7 @@
},
{
"cell_type": "code",
"execution_count": 10,
"execution_count": 6,
"metadata": {},
"outputs": [],
"source": [
......@@ -703,7 +703,7 @@
},
{
"cell_type": "code",
"execution_count": 11,
"execution_count": 7,
"metadata": {},
"outputs": [],
"source": [
......@@ -729,7 +729,7 @@
},
{
"cell_type": "code",
"execution_count": 12,
"execution_count": 8,
"metadata": {},
"outputs": [],
"source": [
......@@ -740,7 +740,7 @@
},
{
"cell_type": "code",
"execution_count": 13,
"execution_count": 9,
"metadata": {},
"outputs": [
{
......@@ -787,7 +787,7 @@
")"
]
},
"execution_count": 13,
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
......@@ -802,7 +802,7 @@
},
{
"cell_type": "code",
"execution_count": 14,
"execution_count": 10,
"metadata": {},
"outputs": [],
"source": [
......@@ -815,7 +815,7 @@
},
{
"cell_type": "code",
"execution_count": null,
"execution_count": 11,
"metadata": {},
"outputs": [],
"source": [
......@@ -828,19 +828,19 @@
},
{
"cell_type": "code",
"execution_count": 18,
"execution_count": 12,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"== Status ==<br>Memory usage on this node: 2.1/12.2 GiB<br>Using FIFO scheduling algorithm.<br>Resources requested: 0/8 CPUs, 0/0 GPUs, 0.0/6.69 GiB heap, 0.0/2.29 GiB objects<br>Result logdir: /home/palo/ray_results/_inner_2021-10-15_01-12-33<br>Number of trials: 2/2 (2 TERMINATED)<br><table>\n",
"== Status ==<br>Memory usage on this node: 3.8/12.2 GiB<br>Using FIFO scheduling algorithm.<br>Resources requested: 0/8 CPUs, 0/0 GPUs, 0.0/5.86 GiB heap, 0.0/2.0 GiB objects<br>Result logdir: /home/palo/ray_results/_inner_2021-10-17_19-06-33<br>Number of trials: 2/2 (2 TERMINATED)<br><table>\n",
"<thead>\n",
"<tr><th>Trial name </th><th>status </th><th>loc </th><th style=\"text-align: right;\"> batch_size</th><th style=\"text-align: right;\"> iter</th><th style=\"text-align: right;\"> total time (s)</th></tr>\n",
"</thead>\n",
"<tbody>\n",
"<tr><td>_inner_367dc_00000</td><td>TERMINATED</td><td> </td><td style=\"text-align: right;\"> 1000</td><td style=\"text-align: right;\"> 5</td><td style=\"text-align: right;\"> 14.1115</td></tr>\n",
"<tr><td>_inner_367dc_00001</td><td>TERMINATED</td><td> </td><td style=\"text-align: right;\"> 5000</td><td style=\"text-align: right;\"> 5</td><td style=\"text-align: right;\"> 13.1224</td></tr>\n",
"<tr><td>_inner_94413_00000</td><td>TERMINATED</td><td> </td><td style=\"text-align: right;\"> 1000</td><td style=\"text-align: right;\"> 5</td><td style=\"text-align: right;\"> 16.337</td></tr>\n",
"<tr><td>_inner_94413_00001</td><td>TERMINATED</td><td> </td><td style=\"text-align: right;\"> 5000</td><td style=\"text-align: right;\"> 5</td><td style=\"text-align: right;\"> 15.021</td></tr>\n",
"</tbody>\n",
"</table><br><br>"
],
......@@ -855,7 +855,7 @@
"name": "stderr",
"output_type": "stream",
"text": [
"2021-10-15 01:12:50,327\tINFO tune.py:448 -- Total run time: 16.66 seconds (16.61 seconds for the tuning loop).\n"
"2021-10-17 19:06:52,530\tINFO tune.py:448 -- Total run time: 21.26 seconds (19.21 seconds for the tuning loop).\n"
]
}
],
......@@ -882,7 +882,7 @@
" initializers={'deeptabular': XavierNormal},\n",
" optimizers={'deeptabular': deep_opt},\n",
" metrics=[accuracy, precision, recall, f1],\n",
" verbose=0)\n",
" verbose=1)\n",
"\n",
" trainer.fit(X_train=X_train,\n",
" X_val=X_val,\n",
......@@ -900,7 +900,7 @@
},
{
"cell_type": "code",
"execution_count": 20,
"execution_count": 13,
"metadata": {},
"outputs": [],
"source": [
......@@ -909,60 +909,38 @@
},
{
"cell_type": "code",
"execution_count": 21,
"execution_count": 14,
"metadata": {},
"outputs": [],
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"\u001b[2m\u001b[36m(pid=14987)\u001b[0m sys:1: ResourceWarning: unclosed file <_io.TextIOWrapper name='/tmp/ray/session_2021-10-17_19-06-31_273508_14906/logs/worker-edb8ddf5edcb134690e06916577fafd5d2bb26af-01000000-14987.out' mode='a' encoding='utf-8'>\n",
"\u001b[2m\u001b[36m(pid=14987)\u001b[0m sys:1: ResourceWarning: unclosed file <_io.TextIOWrapper name='/tmp/ray/session_2021-10-17_19-06-31_273508_14906/logs/worker-edb8ddf5edcb134690e06916577fafd5d2bb26af-01000000-14987.err' mode='a' encoding='utf-8'>\n"
]
}
],
"source": [
"%load_ext tensorboard"
]
},
{
"cell_type": "code",
"execution_count": 22,
"execution_count": 16,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"Reusing TensorBoard on port 6006 (pid 2541), started 7:22:22 ago. (Use '!kill 2541' to kill it.)"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"\n",
" <iframe id=\"tensorboard-frame-de12d04f9fd792fb\" width=\"100%\" height=\"800\" frameborder=\"0\">\n",
" </iframe>\n",
" <script>\n",
" (function() {\n",
" const frame = document.getElementById(\"tensorboard-frame-de12d04f9fd792fb\");\n",
" const url = new URL(\"/\", window.location);\n",
" url.port = 6006;\n",
" frame.src = url;\n",
" })();\n",
" </script>\n",
" "
],
"text/plain": [
"<IPython.core.display.HTML object>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"outputs": [],
"source": [
"%tensorboard --logdir ~/ray_results"
]
}
],
"metadata": {
"interpreter": {
"hash": "3b99005fd577fa40f3cce433b2b92303885900e634b2b5344c07c59d06c8792d"
},
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"display_name": "Python 3.8.5 64-bit ('base': conda)",
"name": "python3"
},
"language_info": {
......
pytorch_widedeep/callbacks.py
浏览文件 @ 04d40076
......@@ -150,6 +150,7 @@ class History(Callback):
This callback runs by default within :obj:`Trainer`, therefore, should not
be passed to the :obj:`Trainer`. Is included here just for completion.
"""
def on_train_begin(self, logs: Optional[Dict] = None):
self.trainer.history = {}
......@@ -158,11 +159,11 @@ class History(Callback):
):
logs = logs or {}
for k, v in logs.items():
if isinstance(v, np.ndarray):# or isinstance(v, list):
if isinstance(v, np.ndarray):
v = v.tolist()
if isinstance(v, list) and len(v)>1:
if isinstance(v, list) and len(v) > 1:
for i in range(len(v)):
self.trainer.history.setdefault(k+'_'+str(i), []).append(v[i])
self.trainer.history.setdefault(k + "_" + str(i), []).append(v[i])
else:
self.trainer.history.setdefault(k, []).append(v)
......@@ -264,10 +265,9 @@ class LRHistory(Callback):
>>> trainer = Trainer(model, objective="regression", callbacks=[LRHistory(n_epochs=10)])
"""
def __init__(self, n_epochs: int, ray_tune: bool = False):
def __init__(self, n_epochs: int):
super(LRHistory, self).__init__()
self.n_epochs = n_epochs
self.ray_tune = ray_tune
def on_epoch_begin(self, epoch: int, logs: Optional[Dict] = None):
if epoch == 0 and self.trainer.lr_scheduler is not None:
......@@ -671,48 +671,15 @@ class EarlyStopping(Callback):
class RayTuneReporter(Callback):
r"""Callback that allows reporting history and lr_history values to RayTune for Hyperparameter tuning
Parameters
-----------
Attributes
----------
Examples
--------
>>> from pytorch_widedeep.callbacks import RayTuneReporter
>>> from pytorch_widedeep.models import TabMlp, Wide, WideDeep
>>> from pytorch_widedeep.training import Trainer
>>> from ray import tune
>>> import tracemalloc
>>> tracemalloc.start()
>>>
>>> config={"batch_size": tune.grid_search([1000, 5000]),}
>>> embed_input = [(u, i, j) for u, i, j in zip(["a", "b", "c"][:4], [4] * 3, [8] * 3)]
>>> column_idx = {k: v for v, k in enumerate(["a", "b", "c"])}
>>> wide = Wide(10, 1)
>>> deep = TabMlp(mlp_hidden_dims=[8, 4], column_idx=column_idx, embed_input=embed_input)
>>> model = WideDeep(wide, deep)
>>>
>>> def training_function(config, X_train, X_val):
>>> batch_size = config["batch_size"]
>>> trainer = Trainer(model, objective="regression", callbacks=[RayTuneReporter])
>>> trainer.fit(X_train=X_train,
>>> X_val=X_val,
>>> n_epochs=5,
>>> batch_size=batch_size)
>>> X_train = {"X_wide": X_wide_train, "X_tab": X_tab_train, "target": y_train}
>>> X_val = {"X_wide": X_wide_valid, "X_tab": X_tab_valid, "target": y_valid}
>>> analysis = tune.run(tune.with_parameters(training_function, X_train=X_train, X_val=X_val),
>>> config=config)
"""
def on_epoch_end(
self, epoch: int, logs: Optional[Dict] = None, metric: Optional[float] = None
):
report_dict = {}
for k, v in self.trainer.history.items():
report_dict.update({k: v[-1]})
if hasattr(self.trainer, 'lr_history'):
if hasattr(self.trainer, "lr_history"):
for k, v in self.trainer.lr_history.items():
report_dict.update({k: v[-1]})
tune.report(report_dict)
\ No newline at end of file
tune.report(report_dict)
pytorch_widedeep/tab2vec.py
浏览文件 @ 04d40076
......@@ -126,8 +126,12 @@ class Tab2Vec:
"""
return self
def transform(self, df: pd.DataFrame, new_embed_col_list: bool = False,
target_col: Optional[str] = None) -> Union[pd.DataFrame, Tuple[pd.DataFrame, pd.Series]]:
def transform(
self,
df: pd.DataFrame,
new_embed_col_list: bool = False,
target_col: Optional[str] = None,
) -> Union[pd.DataFrame, Tuple[pd.DataFrame, pd.Series]]:
r"""
Parameters
----------
......@@ -160,22 +164,36 @@ class Tab2Vec:
col_names = list(self.tab_preprocessor.column_idx.keys())
embed_col_names = []
for col, vec_size in tab_preprocessor.embed_cols:
embed_col_names_temp = [col+'_'+str(i) for i in range(vec_size)]
embed_col_names.extend(embed_col_names_temp)
col_names = list(chain.from_iterable(embed_col_names_temp if item == col
else [item] for item in col_names))
if target_col:
if new_embed_col_list:
return pd.DataFrame(data=X_vec, columns=col_names), df[target_col], embed_col_names
if self.tab_preprocessor.for_transformer:
if target_col:
return pd.DataFrame(data=X_vec), df[target_col]
else:
return pd.DataFrame(data=X_vec, columns=col_names), df[target_col]
return pd.DataFrame(data=X_vec)
else:
if new_embed_col_list:
return pd.DataFrame(data=X_vec, columns=col_names), embed_col_names
for col, vec_size in self.tab_preprocessor.embed_cols:
embed_col_names_temp = [col + "_" + str(i) for i in range(vec_size)]
embed_col_names.extend(embed_col_names_temp)
col_names = list(
chain.from_iterable(
embed_col_names_temp if item == col else [item]
for item in col_names
)
)
if target_col:
if new_embed_col_list:
return (
pd.DataFrame(data=X_vec, columns=col_names),
df[target_col],
embed_col_names,
)
else:
return pd.DataFrame(data=X_vec, columns=col_names), df[target_col]
else:
return pd.DataFrame(data=X_vec, columns=col_names)
if new_embed_col_list:
return pd.DataFrame(data=X_vec, columns=col_names), embed_col_names
else:
return pd.DataFrame(data=X_vec, columns=col_names)
def fit_transform(
self, df: pd.DataFrame, target_col: Optional[str] = None
......
pytorch_widedeep/training/trainer.py
浏览文件 @ 04d40076
......@@ -1242,7 +1242,7 @@ class Trainer:
else:
self.metric = None
sorted_callbacks_list = deepcopy(self.callbacks)
for obj,i in zip(self.callbacks[::-1], range(len(self.callbacks))[::-1]):
for obj, i in zip(self.callbacks[::-1], range(len(self.callbacks))[::-1]):
if isinstance(obj, RayTuneReporter):
sorted_callbacks_list.append(sorted_callbacks_list.pop(i))
self.callbacks = sorted_callbacks_list
......
setup.py
浏览文件 @ 04d40076
......@@ -65,7 +65,7 @@ setup_kwargs = {
"einops",
"wrapt",
"torchmetrics",
"ray",
"ray[tune]",
],
"extras_require": extras,
"python_requires": ">=3.7.0",
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或