注意
点击此处下载完整示例代码
使用 Ray Tune 进行超参数调优¶
创建日期：2020 年 8 月 31 日 | 最后更新：2024 年 10 月 31 日 | 最后验证：2024 年 11 月 05 日
超参数调优能够使模型从表现平平跃升至高度准确。通常，一些简单的改动，如选择不同的学习率或更改网络层的大小，都可能对模型性能产生显著影响。
幸运的是，有一些工具可以帮助找到最佳参数组合。Ray Tune 是一个业界标准的分布式超参数调优工具。Ray Tune 包含了最新的超参数搜索算法，集成了各种分析库，并通过Ray 的分布式机器学习引擎原生支持分布式训练。
在本教程中，我们将展示如何将 Ray Tune 集成到您的 PyTorch 训练工作流程中。我们将扩展PyTorch 文档中的这个教程，用于训练 CIFAR10 图像分类器。
正如您将看到的，我们只需要进行一些微小的修改。具体来说，我们需要
将数据加载和训练封装到函数中，
使部分网络参数可配置，
添加检查点（可选），
并定义模型调优的搜索空间
要运行本教程，请确保已安装以下软件包
ray[tune]：分布式超参数调优库torchvision：用于数据变换器
设置 / 导入¶
我们从导入开始
from functools import partial
import os
import tempfile
from pathlib import Path
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torch.utils.data import random_split
import torchvision
import torchvision.transforms as transforms
from ray import tune
from ray import train
from ray.train import Checkpoint, get_checkpoint
from ray.tune.schedulers import ASHAScheduler
import ray.cloudpickle as pickle
大多数导入用于构建 PyTorch 模型。只有最后的导入是针对 Ray Tune 的。
数据加载器¶
我们将数据加载器封装到自己的函数中，并传递一个全局数据目录。这样我们可以在不同的试验之间共享数据目录。
def load_data(data_dir="./data"):
transform = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))]
)
trainset = torchvision.datasets.CIFAR10(
root=data_dir, train=True, download=True, transform=transform
)
testset = torchvision.datasets.CIFAR10(
root=data_dir, train=False, download=True, transform=transform
)
return trainset, testset
可配置的神经网络¶
我们只能调优那些可配置的参数。在本例中，我们可以指定全连接层的层大小
class Net(nn.Module):
def __init__(self, l1=120, l2=84):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(3, 6, 5)
self.pool = nn.MaxPool2d(2, 2)
self.conv2 = nn.Conv2d(6, 16, 5)
self.fc1 = nn.Linear(16 * 5 * 5, l1)
self.fc2 = nn.Linear(l1, l2)
self.fc3 = nn.Linear(l2, 10)
def forward(self, x):
x = self.pool(F.relu(self.conv1(x)))
x = self.pool(F.relu(self.conv2(x)))
x = torch.flatten(x, 1) # flatten all dimensions except batch
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
训练函数¶
现在有趣的地方来了，因为我们对PyTorch 文档中的示例进行了一些修改。
我们将训练脚本封装在一个函数 train_cifar(config, data_dir=None) 中。config 参数将接收我们希望用于训练的超参数。data_dir 指定我们加载和存储数据的目录，以便多个运行可以共享相同的数据源。如果在提供检查点的情况下，我们还在运行开始时加载模型和优化器状态。在本教程的后面部分，您将找到关于如何保存检查点及其用途的信息。
net = Net(config["l1"], config["l2"])
checkpoint = get_checkpoint()
if checkpoint:
with checkpoint.as_directory() as checkpoint_dir:
data_path = Path(checkpoint_dir) / "data.pkl"
with open(data_path, "rb") as fp:
checkpoint_state = pickle.load(fp)
start_epoch = checkpoint_state["epoch"]
net.load_state_dict(checkpoint_state["net_state_dict"])
optimizer.load_state_dict(checkpoint_state["optimizer_state_dict"])
else:
start_epoch = 0
优化器的学习率也设置为可配置的
optimizer = optim.SGD(net.parameters(), lr=config["lr"], momentum=0.9)
我们还将训练数据分割为训练集和验证集。因此，我们使用 80% 的数据进行训练，并在剩余的 20% 数据上计算验证损失。遍历训练集和测试集所使用的批量大小 (batch size) 也是可配置的。
使用 DataParallel 添加（多）GPU 支持¶
图像分类很大程度上受益于 GPU。幸运的是，我们可以在 Ray Tune 中继续使用 PyTorch 的抽象。因此，我们可以将模型封装在 nn.DataParallel 中，以支持在多个 GPU 上进行数据并行训练
device = "cpu"
if torch.cuda.is_available():
device = "cuda:0"
if torch.cuda.device_count() > 1:
net = nn.DataParallel(net)
net.to(device)
通过使用 device 变量，我们确保在没有 GPU 可用时训练也能正常进行。PyTorch 要求我们将数据显式地发送到 GPU 内存，如下所示
for i, data in enumerate(trainloader, 0):
inputs, labels = data
inputs, labels = inputs.to(device), labels.to(device)
现在代码支持在 CPU、单 GPU 和多 GPU 上进行训练。值得注意的是，Ray 还支持分数 GPU (fractional GPUs)，因此我们可以在试验之间共享 GPU，只要模型仍能容纳在 GPU 内存中。我们稍后会再讨论这一点。
与 Ray Tune 通信¶
最有趣的部分是与 Ray Tune 的通信
checkpoint_data = {
"epoch": epoch,
"net_state_dict": net.state_dict(),
"optimizer_state_dict": optimizer.state_dict(),
}
with tempfile.TemporaryDirectory() as checkpoint_dir:
data_path = Path(checkpoint_dir) / "data.pkl"
with open(data_path, "wb") as fp:
pickle.dump(checkpoint_data, fp)
checkpoint = Checkpoint.from_directory(checkpoint_dir)
train.report(
{"loss": val_loss / val_steps, "accuracy": correct / total},
checkpoint=checkpoint,
)
在这里，我们首先保存一个检查点，然后向 Ray Tune 报告一些指标。具体来说，我们将验证损失和准确率发送回 Ray Tune。Ray Tune 随后可以使用这些指标来决定哪种超参数配置带来了最佳结果。这些指标也可以用来提前停止表现不佳的试验，以避免浪费资源。
保存检查点是可选的，但是，如果我们想使用像基于总体的训练 (Population Based Training) 这样的高级调度器，则是必需的。此外，通过保存检查点，我们稍后可以加载训练好的模型并在测试集上进行验证。最后，保存检查点对于容错很有用，并且允许我们中断训练并在之后继续训练。
完整的训练函数¶
完整的代码示例如下所示
def train_cifar(config, data_dir=None):
net = Net(config["l1"], config["l2"])
device = "cpu"
if torch.cuda.is_available():
device = "cuda:0"
if torch.cuda.device_count() > 1:
net = nn.DataParallel(net)
net.to(device)
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=config["lr"], momentum=0.9)
checkpoint = get_checkpoint()
if checkpoint:
with checkpoint.as_directory() as checkpoint_dir:
data_path = Path(checkpoint_dir) / "data.pkl"
with open(data_path, "rb") as fp:
checkpoint_state = pickle.load(fp)
start_epoch = checkpoint_state["epoch"]
net.load_state_dict(checkpoint_state["net_state_dict"])
optimizer.load_state_dict(checkpoint_state["optimizer_state_dict"])
else:
start_epoch = 0
trainset, testset = load_data(data_dir)
test_abs = int(len(trainset) * 0.8)
train_subset, val_subset = random_split(
trainset, [test_abs, len(trainset) - test_abs]
)
trainloader = torch.utils.data.DataLoader(
train_subset, batch_size=int(config["batch_size"]), shuffle=True, num_workers=8
)
valloader = torch.utils.data.DataLoader(
val_subset, batch_size=int(config["batch_size"]), shuffle=True, num_workers=8
)
for epoch in range(start_epoch, 10): # loop over the dataset multiple times
running_loss = 0.0
epoch_steps = 0
for i, data in enumerate(trainloader, 0):
# get the inputs; data is a list of [inputs, labels]
inputs, labels = data
inputs, labels = inputs.to(device), labels.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward + backward + optimize
outputs = net(inputs)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
# print statistics
running_loss += loss.item()
epoch_steps += 1
if i % 2000 == 1999: # print every 2000 mini-batches
print(
"[%d, %5d] loss: %.3f"
% (epoch + 1, i + 1, running_loss / epoch_steps)
)
running_loss = 0.0
# Validation loss
val_loss = 0.0
val_steps = 0
total = 0
correct = 0
for i, data in enumerate(valloader, 0):
with torch.no_grad():
inputs, labels = data
inputs, labels = inputs.to(device), labels.to(device)
outputs = net(inputs)
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
loss = criterion(outputs, labels)
val_loss += loss.cpu().numpy()
val_steps += 1
checkpoint_data = {
"epoch": epoch,
"net_state_dict": net.state_dict(),
"optimizer_state_dict": optimizer.state_dict(),
}
with tempfile.TemporaryDirectory() as checkpoint_dir:
data_path = Path(checkpoint_dir) / "data.pkl"
with open(data_path, "wb") as fp:
pickle.dump(checkpoint_data, fp)
checkpoint = Checkpoint.from_directory(checkpoint_dir)
train.report(
{"loss": val_loss / val_steps, "accuracy": correct / total},
checkpoint=checkpoint,
)
print("Finished Training")
正如您所见，大部分代码都是直接改编自原始示例。
测试集准确率¶
通常，机器学习模型的性能会在一个保留的测试集上进行测试，测试集中的数据未用于模型训练。我们也将其封装在一个函数中
def test_accuracy(net, device="cpu"):
trainset, testset = load_data()
testloader = torch.utils.data.DataLoader(
testset, batch_size=4, shuffle=False, num_workers=2
)
correct = 0
total = 0
with torch.no_grad():
for data in testloader:
images, labels = data
images, labels = images.to(device), labels.to(device)
outputs = net(images)
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
return correct / total
该函数还期望一个 device 参数，这样我们就可以在 GPU 上进行测试集验证。
配置搜索空间¶
最后，我们需要定义 Ray Tune 的搜索空间。以下是一个示例
config = {
"l1": tune.choice([2 ** i for i in range(9)]),
"l2": tune.choice([2 ** i for i in range(9)]),
"lr": tune.loguniform(1e-4, 1e-1),
"batch_size": tune.choice([2, 4, 8, 16])
}
tune.choice() 接受一个值列表，这些值将从中进行均匀采样。在本例中，l1 和 l2 参数应为 4 到 256 之间的 2 的幂，即 4、8、16、32、64、128 或 256。 lr（学习率）应在 0.0001 到 0.1 之间均匀采样。最后，批量大小 (batch size) 是从 2、4、8 和 16 中选择一个。
在每次试验中，Ray Tune 将从这些搜索空间中随机采样参数组合。然后，它将并行训练多个模型，并从中找到性能最佳的模型。我们还使用 ASHAScheduler，它将提前终止表现不佳的试验。
我们使用 functools.partial 包装 train_cifar 函数，以设置常量 data_dir 参数。我们还可以告诉 Ray Tune 为每个试验分配哪些资源
gpus_per_trial = 2
# ...
result = tune.run(
partial(train_cifar, data_dir=data_dir),
resources_per_trial={"cpu": 8, "gpu": gpus_per_trial},
config=config,
num_samples=num_samples,
scheduler=scheduler,
checkpoint_at_end=True)
您可以指定 CPU 的数量，这些 CPU 可用于例如增加 PyTorch DataLoader 实例的 num_workers。所选的 GPU 数量在每个试验中对 PyTorch 可见。试验无法访问未为其请求的 GPU - 因此您不必担心两个试验使用同一组资源。
这里我们也可以指定分数 GPU (fractional GPUs)，所以像 gpus_per_trial=0.5 这样的设置是完全有效的。试验之间将共享 GPU。您只需确保模型仍然能够容纳在 GPU 内存中。
训练模型后，我们将找到性能最佳的模型，并从检查点文件中加载训练好的网络。然后，我们获取测试集准确率，并通过打印输出报告所有信息。
完整的 main 函数如下所示
def main(num_samples=10, max_num_epochs=10, gpus_per_trial=2):
data_dir = os.path.abspath("./data")
load_data(data_dir)
config = {
"l1": tune.choice([2**i for i in range(9)]),
"l2": tune.choice([2**i for i in range(9)]),
"lr": tune.loguniform(1e-4, 1e-1),
"batch_size": tune.choice([2, 4, 8, 16]),
}
scheduler = ASHAScheduler(
metric="loss",
mode="min",
max_t=max_num_epochs,
grace_period=1,
reduction_factor=2,
)
result = tune.run(
partial(train_cifar, data_dir=data_dir),
resources_per_trial={"cpu": 2, "gpu": gpus_per_trial},
config=config,
num_samples=num_samples,
scheduler=scheduler,
)
best_trial = result.get_best_trial("loss", "min", "last")
print(f"Best trial config: {best_trial.config}")
print(f"Best trial final validation loss: {best_trial.last_result['loss']}")
print(f"Best trial final validation accuracy: {best_trial.last_result['accuracy']}")
best_trained_model = Net(best_trial.config["l1"], best_trial.config["l2"])
device = "cpu"
if torch.cuda.is_available():
device = "cuda:0"
if gpus_per_trial > 1:
best_trained_model = nn.DataParallel(best_trained_model)
best_trained_model.to(device)
best_checkpoint = result.get_best_checkpoint(trial=best_trial, metric="accuracy", mode="max")
with best_checkpoint.as_directory() as checkpoint_dir:
data_path = Path(checkpoint_dir) / "data.pkl"
with open(data_path, "rb") as fp:
best_checkpoint_data = pickle.load(fp)
best_trained_model.load_state_dict(best_checkpoint_data["net_state_dict"])
test_acc = test_accuracy(best_trained_model, device)
print("Best trial test set accuracy: {}".format(test_acc))
if __name__ == "__main__":
# You can change the number of GPUs per trial here:
main(num_samples=10, max_num_epochs=10, gpus_per_trial=0)
0% 0.00/170M [00:00<?, ?B/s]
0% 557k/170M [00:00<00:30, 5.55MB/s]
4% 6.39M/170M [00:00<00:04, 33.6MB/s]
10% 16.6M/170M [00:00<00:02, 63.3MB/s]
15% 25.2M/170M [00:00<00:02, 72.0MB/s]
20% 33.8M/170M [00:00<00:01, 77.0MB/s]
25% 42.4M/170M [00:00<00:01, 79.8MB/s]
30% 50.8M/170M [00:00<00:01, 80.9MB/s]
35% 58.9M/170M [00:00<00:01, 75.6MB/s]
39% 66.6M/170M [00:00<00:01, 69.1MB/s]
43% 73.6M/170M [00:01<00:01, 66.0MB/s]
47% 80.3M/170M [00:01<00:01, 64.3MB/s]
51% 86.9M/170M [00:01<00:01, 63.0MB/s]
55% 93.2M/170M [00:01<00:01, 62.2MB/s]
58% 99.5M/170M [00:01<00:01, 62.1MB/s]
62% 106M/170M [00:01<00:01, 62.9MB/s]
66% 113M/170M [00:01<00:00, 64.2MB/s]
70% 119M/170M [00:01<00:00, 64.9MB/s]
74% 127M/170M [00:01<00:00, 67.7MB/s]
79% 134M/170M [00:02<00:00, 69.4MB/s]
83% 141M/170M [00:02<00:00, 70.0MB/s]
87% 149M/170M [00:02<00:00, 70.7MB/s]
91% 156M/170M [00:02<00:00, 71.4MB/s]
96% 163M/170M [00:02<00:00, 71.4MB/s]
100% 170M/170M [00:02<00:00, 71.0MB/s]
100% 170M/170M [00:02<00:00, 67.4MB/s]
2025-04-23 16:30:10,527 WARNING services.py:1889 -- WARNING: The object store is using /tmp instead of /dev/shm because /dev/shm has only 2147479552 bytes available. This will harm performance! You may be able to free up space by deleting files in /dev/shm. If you are inside a Docker container, you can increase /dev/shm size by passing '--shm-size=10.24gb' to 'docker run' (or add it to the run_options list in a Ray cluster config). Make sure to set this to more than 30% of available RAM.
2025-04-23 16:30:10,579 INFO worker.py:1642 -- Started a local Ray instance.
2025-04-23 16:30:11,504 INFO tune.py:228 -- Initializing Ray automatically. For cluster usage or custom Ray initialization, call `ray.init(...)` before `tune.run(...)`.
2025-04-23 16:30:11,506 INFO tune.py:654 -- [output] This will use the new output engine with verbosity 2. To disable the new output and use the legacy output engine, set the environment variable RAY_AIR_NEW_OUTPUT=0. For more information, please see https://github.com/ray-project/ray/issues/36949
+--------------------------------------------------------------------+
| Configuration for experiment train_cifar_2025-04-23_16-30-11 |
+--------------------------------------------------------------------+
| Search algorithm BasicVariantGenerator |
| Scheduler AsyncHyperBandScheduler |
| Number of trials 10 |
+--------------------------------------------------------------------+
View detailed results here: /var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11
To visualize your results with TensorBoard, run: `tensorboard --logdir /var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11`
Trial status: 10 PENDING
Current time: 2025-04-23 16:30:11. Total running time: 0s
Logical resource usage: 16.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+-------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size |
+-------------------------------------------------------------------------------+
| train_cifar_3a47b_00000 PENDING 16 1 0.00213327 2 |
| train_cifar_3a47b_00001 PENDING 1 2 0.013416 4 |
| train_cifar_3a47b_00002 PENDING 256 64 0.0113784 2 |
| train_cifar_3a47b_00003 PENDING 64 256 0.0274071 8 |
| train_cifar_3a47b_00004 PENDING 16 2 0.056666 4 |
| train_cifar_3a47b_00005 PENDING 8 64 0.000353097 4 |
| train_cifar_3a47b_00006 PENDING 16 4 0.000147684 8 |
| train_cifar_3a47b_00007 PENDING 256 256 0.00477469 8 |
| train_cifar_3a47b_00008 PENDING 128 256 0.0306227 8 |
| train_cifar_3a47b_00009 PENDING 2 16 0.0286986 2 |
+-------------------------------------------------------------------------------+
Trial train_cifar_3a47b_00000 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00000 config |
+--------------------------------------------------+
| batch_size 2 |
| l1 16 |
| l2 1 |
| lr 0.00213 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00002 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00002 config |
+--------------------------------------------------+
| batch_size 2 |
| l1 256 |
| l2 64 |
| lr 0.01138 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00007 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00007 config |
+--------------------------------------------------+
| batch_size 8 |
| l1 256 |
| l2 256 |
| lr 0.00477 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00004 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00004 config |
+--------------------------------------------------+
| batch_size 4 |
| l1 16 |
| l2 2 |
| lr 0.05667 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00003 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00003 config |
+--------------------------------------------------+
| batch_size 8 |
| l1 64 |
| l2 256 |
| lr 0.02741 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00001 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00001 config |
+--------------------------------------------------+
| batch_size 4 |
| l1 1 |
| l2 2 |
| lr 0.01342 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00005 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00005 config |
+--------------------------------------------------+
| batch_size 4 |
| l1 8 |
| l2 64 |
| lr 0.00035 |
+--------------------------------------------------+
Trial train_cifar_3a47b_00006 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00006 config |
+--------------------------------------------------+
| batch_size 8 |
| l1 16 |
| l2 4 |
| lr 0.00015 |
+--------------------------------------------------+
(func pid=4375) [1, 2000] loss: 2.327
(func pid=4375) [1, 4000] loss: 1.153 [repeated 8x across cluster] (Ray deduplicates logs by default. Set RAY_DEDUP_LOGS=0 to disable log deduplication, or see https://docs.ray.io/en/master/ray-observability/ray-logging.html#log-deduplication for more options.)
Trial status: 8 RUNNING | 2 PENDING
Current time: 2025-04-23 16:30:41. Total running time: 30s
Logical resource usage: 16.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+-------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size |
+-------------------------------------------------------------------------------+
| train_cifar_3a47b_00000 RUNNING 16 1 0.00213327 2 |
| train_cifar_3a47b_00001 RUNNING 1 2 0.013416 4 |
| train_cifar_3a47b_00002 RUNNING 256 64 0.0113784 2 |
| train_cifar_3a47b_00003 RUNNING 64 256 0.0274071 8 |
| train_cifar_3a47b_00004 RUNNING 16 2 0.056666 4 |
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 |
| train_cifar_3a47b_00008 PENDING 128 256 0.0306227 8 |
| train_cifar_3a47b_00009 PENDING 2 16 0.0286986 2 |
+-------------------------------------------------------------------------------+
(func pid=4375) [1, 6000] loss: 0.769 [repeated 8x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 1 at 2025-04-23 16:30:57. Total running time: 46s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 40.8067 |
| time_total_s 40.8067 |
| training_iteration 1 |
| accuracy 0.1586 |
| loss 2.24278 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000000
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000000)
Trial train_cifar_3a47b_00003 finished iteration 1 at 2025-04-23 16:30:58. Total running time: 46s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00003 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 41.74049 |
| time_total_s 41.74049 |
| training_iteration 1 |
| accuracy 0.2221 |
| loss 2.06515 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00003 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00003_3_batch_size=8,l1=64,l2=256,lr=0.0274_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00007 finished iteration 1 at 2025-04-23 16:30:59. Total running time: 47s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 43.10883 |
| time_total_s 43.10883 |
| training_iteration 1 |
| accuracy 0.4819 |
| loss 1.4359 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000000
(func pid=4375) [1, 8000] loss: 0.576 [repeated 5x across cluster]
(func pid=4375) [1, 10000] loss: 0.461 [repeated 5x across cluster]
Trial status: 8 RUNNING | 2 PENDING
Current time: 2025-04-23 16:31:11. Total running time: 1min 0s
Logical resource usage: 16.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+----------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+----------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00000 RUNNING 16 1 0.00213327 2 |
| train_cifar_3a47b_00001 RUNNING 1 2 0.013416 4 |
| train_cifar_3a47b_00002 RUNNING 256 64 0.0113784 2 |
| train_cifar_3a47b_00003 RUNNING 64 256 0.0274071 8 1 41.7405 2.06515 0.2221 |
| train_cifar_3a47b_00004 RUNNING 16 2 0.056666 4 |
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 1 40.8067 2.24278 0.1586 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 1 43.1088 1.4359 0.4819 |
| train_cifar_3a47b_00008 PENDING 128 256 0.0306227 8 |
| train_cifar_3a47b_00009 PENDING 2 16 0.0286986 2 |
+----------------------------------------------------------------------------------------------------------------------------------+
(func pid=4375) [1, 12000] loss: 0.384 [repeated 8x across cluster]
Trial train_cifar_3a47b_00001 finished iteration 1 at 2025-04-23 16:31:24. Total running time: 1min 12s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00001 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 67.90545 |
| time_total_s 67.90545 |
| training_iteration 1 |
| accuracy 0.1006 |
| loss 2.31772 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00001 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00001_1_batch_size=4,l1=1,l2=2,lr=0.0134_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00001 completed after 1 iterations at 2025-04-23 16:31:24. Total running time: 1min 12s
Trial train_cifar_3a47b_00008 started with configuration:
+--------------------------------------------------+
| Trial train_cifar_3a47b_00008 config |
+--------------------------------------------------+
| batch_size 8 |
| l1 128 |
| l2 256 |
| lr 0.03062 |
+--------------------------------------------------+
(func pid=4376) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00001_1_batch_size=4,l1=1,l2=2,lr=0.0134_2025-04-23_16-30-11/checkpoint_000000) [repeated 3x across cluster]
Trial train_cifar_3a47b_00004 finished iteration 1 at 2025-04-23 16:31:26. Total running time: 1min 15s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00004 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 70.33586 |
| time_total_s 70.33586 |
| training_iteration 1 |
| accuracy 0.0995 |
| loss 2.32697 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00004 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00004_4_batch_size=4,l1=16,l2=2,lr=0.0567_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00004 completed after 1 iterations at 2025-04-23 16:31:26. Total running time: 1min 15s
Trial train_cifar_3a47b_00009 started with configuration:
+-------------------------------------------------+
| Trial train_cifar_3a47b_00009 config |
+-------------------------------------------------+
| batch_size 2 |
| l1 2 |
| l2 16 |
| lr 0.0287 |
+-------------------------------------------------+
(func pid=4382) [2, 4000] loss: 0.682 [repeated 4x across cluster]
Trial train_cifar_3a47b_00005 finished iteration 1 at 2025-04-23 16:31:27. Total running time: 1min 16s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 71.31317 |
| time_total_s 71.31317 |
| training_iteration 1 |
| accuracy 0.3719 |
| loss 1.68402 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000000
(func pid=4377) [1, 14000] loss: 0.331 [repeated 2x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 2 at 2025-04-23 16:31:36. Total running time: 1min 24s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000001 |
| time_this_iter_s 38.60301 |
| time_total_s 79.40971 |
| training_iteration 2 |
| accuracy 0.2028 |
| loss 2.12796 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 2 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000001
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000001) [repeated 3x across cluster]
Trial train_cifar_3a47b_00003 finished iteration 2 at 2025-04-23 16:31:36. Total running time: 1min 25s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00003 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000001 |
| time_this_iter_s 38.6455 |
| time_total_s 80.38599 |
| training_iteration 2 |
| accuracy 0.1971 |
| loss 2.19641 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00003 saved a checkpoint for iteration 2 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00003_3_batch_size=8,l1=64,l2=256,lr=0.0274_2025-04-23_16-30-11/checkpoint_000001
Trial train_cifar_3a47b_00003 completed after 2 iterations at 2025-04-23 16:31:36. Total running time: 1min 25s
Trial train_cifar_3a47b_00007 finished iteration 2 at 2025-04-23 16:31:40. Total running time: 1min 28s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000001 |
| time_this_iter_s 40.65258 |
| time_total_s 83.76141 |
| training_iteration 2 |
| accuracy 0.4792 |
| loss 1.50171 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 2 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000001
Trial status: 7 RUNNING | 3 TERMINATED
Current time: 2025-04-23 16:31:41. Total running time: 1min 30s
Logical resource usage: 14.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00000 RUNNING 16 1 0.00213327 2 |
| train_cifar_3a47b_00002 RUNNING 256 64 0.0113784 2 |
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 1 71.3132 1.68402 0.3719 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 2 79.4097 2.12796 0.2028 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 2 83.7614 1.50171 0.4792 |
| train_cifar_3a47b_00008 RUNNING 128 256 0.0306227 8 |
| train_cifar_3a47b_00009 RUNNING 2 16 0.0286986 2 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4377) [1, 16000] loss: 0.289 [repeated 5x across cluster]
(func pid=4376) [1, 4000] loss: 1.061 [repeated 5x across cluster]
(func pid=4378) [1, 6000] loss: 0.777 [repeated 3x across cluster]
Trial train_cifar_3a47b_00008 finished iteration 1 at 2025-04-23 16:32:02. Total running time: 1min 50s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00008 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 37.58341 |
| time_total_s 37.58341 |
| training_iteration 1 |
| accuracy 0.2113 |
| loss 2.16249 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00008 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00008_8_batch_size=8,l1=128,l2=256,lr=0.0306_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00008 completed after 1 iterations at 2025-04-23 16:32:02. Total running time: 1min 50s
(func pid=4376) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00008_8_batch_size=8,l1=128,l2=256,lr=0.0306_2025-04-23_16-30-11/checkpoint_000000) [repeated 3x across cluster]
(func pid=4377) [1, 20000] loss: 0.231 [repeated 4x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 3 at 2025-04-23 16:32:07. Total running time: 1min 55s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000002 |
| time_this_iter_s 30.94938 |
| time_total_s 110.35909 |
| training_iteration 3 |
| accuracy 0.225 |
| loss 2.06686 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 3 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000002
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000002)
Trial status: 6 RUNNING | 4 TERMINATED
Current time: 2025-04-23 16:32:12. Total running time: 2min 0s
Logical resource usage: 12.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00000 RUNNING 16 1 0.00213327 2 |
| train_cifar_3a47b_00002 RUNNING 256 64 0.0113784 2 |
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 1 71.3132 1.68402 0.3719 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 3 110.359 2.06686 0.225 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 2 83.7614 1.50171 0.4792 |
| train_cifar_3a47b_00009 RUNNING 2 16 0.0286986 2 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4378) [1, 10000] loss: 0.467 [repeated 4x across cluster]
Trial train_cifar_3a47b_00000 finished iteration 1 at 2025-04-23 16:32:13. Total running time: 2min 1s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00000 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 116.93764 |
| time_total_s 116.93764 |
| training_iteration 1 |
| accuracy 0.1011 |
| loss 2.30351 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00000 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00000_0_batch_size=2,l1=16,l2=1,lr=0.0021_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00000 completed after 1 iterations at 2025-04-23 16:32:13. Total running time: 2min 1s
(func pid=4375) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00000_0_batch_size=2,l1=16,l2=1,lr=0.0021_2025-04-23_16-30-11/checkpoint_000000)
Trial train_cifar_3a47b_00007 finished iteration 3 at 2025-04-23 16:32:13. Total running time: 2min 1s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000002 |
| time_this_iter_s 33.34819 |
| time_total_s 117.10959 |
| training_iteration 3 |
| accuracy 0.448 |
| loss 1.67289 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 3 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000002
Trial train_cifar_3a47b_00002 finished iteration 1 at 2025-04-23 16:32:18. Total running time: 2min 6s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00002 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 121.95704 |
| time_total_s 121.95704 |
| training_iteration 1 |
| accuracy 0.1034 |
| loss 2.31648 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00002 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00002_2_batch_size=2,l1=256,l2=64,lr=0.0114_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00002 completed after 1 iterations at 2025-04-23 16:32:18. Total running time: 2min 6s
(func pid=4377) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00002_2_batch_size=2,l1=256,l2=64,lr=0.0114_2025-04-23_16-30-11/checkpoint_000000) [repeated 2x across cluster]
(func pid=4378) [1, 12000] loss: 0.389 [repeated 3x across cluster]
Trial train_cifar_3a47b_00005 finished iteration 2 at 2025-04-23 16:32:21. Total running time: 2min 10s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000001 |
| time_this_iter_s 53.70092 |
| time_total_s 125.01409 |
| training_iteration 2 |
| accuracy 0.4516 |
| loss 1.48 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 2 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000001
(func pid=4381) [4, 4000] loss: 1.010 [repeated 2x across cluster]
(func pid=4382) [4, 4000] loss: 0.593 [repeated 3x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 4 at 2025-04-23 16:32:32. Total running time: 2min 20s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000003 |
| time_this_iter_s 25.17148 |
| time_total_s 135.53057 |
| training_iteration 4 |
| accuracy 0.2354 |
| loss 1.99812 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 4 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000003
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000003) [repeated 2x across cluster]
Trial train_cifar_3a47b_00007 finished iteration 4 at 2025-04-23 16:32:40. Total running time: 2min 28s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000003 |
| time_this_iter_s 26.63057 |
| time_total_s 143.74016 |
| training_iteration 4 |
| accuracy 0.5752 |
| loss 1.21551 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 4 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000003
(func pid=4382) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000003)
(func pid=4378) [1, 18000] loss: 0.259 [repeated 3x across cluster]
Trial status: 6 TERMINATED | 4 RUNNING
Current time: 2025-04-23 16:32:42. Total running time: 2min 30s
Logical resource usage: 8.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 2 125.014 1.48 0.4516 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 4 135.531 1.99812 0.2354 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 4 143.74 1.21551 0.5752 |
| train_cifar_3a47b_00009 RUNNING 2 16 0.0286986 2 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4378) [1, 20000] loss: 0.233 [repeated 3x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 5 at 2025-04-23 16:32:56. Total running time: 2min 45s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000004 |
| time_this_iter_s 24.28981 |
| time_total_s 159.82038 |
| training_iteration 5 |
| accuracy 0.2481 |
| loss 1.91047 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 5 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000004
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000004)
Trial train_cifar_3a47b_00009 finished iteration 1 at 2025-04-23 16:32:58. Total running time: 2min 46s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00009 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000000 |
| time_this_iter_s 91.71496 |
| time_total_s 91.71496 |
| training_iteration 1 |
| accuracy 0.0986 |
| loss 2.3265 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00009 saved a checkpoint for iteration 1 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00009_9_batch_size=2,l1=2,l2=16,lr=0.0287_2025-04-23_16-30-11/checkpoint_000000
Trial train_cifar_3a47b_00009 completed after 1 iterations at 2025-04-23 16:32:58. Total running time: 2min 46s
(func pid=4380) [3, 10000] loss: 0.277 [repeated 4x across cluster]
Trial train_cifar_3a47b_00005 finished iteration 3 at 2025-04-23 16:33:04. Total running time: 2min 52s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000002 |
| time_this_iter_s 42.52939 |
| time_total_s 167.54348 |
| training_iteration 3 |
| accuracy 0.4965 |
| loss 1.39015 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 3 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000002
(func pid=4380) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000002) [repeated 2x across cluster]
(func pid=4381) [6, 2000] loss: 1.894 [repeated 2x across cluster]
Trial train_cifar_3a47b_00007 finished iteration 5 at 2025-04-23 16:33:06. Total running time: 2min 54s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000004 |
| time_this_iter_s 26.0278 |
| time_total_s 169.76797 |
| training_iteration 5 |
| accuracy 0.5697 |
| loss 1.23288 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 5 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000004
(func pid=4380) [4, 2000] loss: 1.349
Trial status: 7 TERMINATED | 3 RUNNING
Current time: 2025-04-23 16:33:12. Total running time: 3min 0s
Logical resource usage: 6.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 3 167.543 1.39015 0.4965 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 5 159.82 1.91047 0.2481 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 5 169.768 1.23288 0.5697 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4381) [6, 4000] loss: 0.929
(func pid=4380) [4, 4000] loss: 0.680 [repeated 2x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 6 at 2025-04-23 16:33:19. Total running time: 3min 7s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000005 |
| time_this_iter_s 22.73444 |
| time_total_s 182.55482 |
| training_iteration 6 |
| accuracy 0.2913 |
| loss 1.84608 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 6 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000005
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000005) [repeated 2x across cluster]
(func pid=4382) [6, 4000] loss: 0.542
(func pid=4380) [4, 6000] loss: 0.444
Trial train_cifar_3a47b_00007 finished iteration 6 at 2025-04-23 16:33:30. Total running time: 3min 19s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000005 |
| time_this_iter_s 24.77263 |
| time_total_s 194.5406 |
| training_iteration 6 |
| accuracy 0.5585 |
| loss 1.28396 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 6 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000005
(func pid=4382) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000005)
(func pid=4380) [4, 8000] loss: 0.330 [repeated 2x across cluster]
(func pid=4380) [4, 10000] loss: 0.264 [repeated 2x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 7 at 2025-04-23 16:33:41. Total running time: 3min 29s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000006 |
| time_this_iter_s 21.84277 |
| time_total_s 204.39759 |
| training_iteration 7 |
| accuracy 0.3138 |
| loss 1.80075 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 7 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000006
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000006)
Trial status: 7 TERMINATED | 3 RUNNING
Current time: 2025-04-23 16:33:42. Total running time: 3min 30s
Logical resource usage: 6.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 3 167.543 1.39015 0.4965 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 7 204.398 1.80075 0.3138 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 6 194.541 1.28396 0.5585 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
Trial train_cifar_3a47b_00005 finished iteration 4 at 2025-04-23 16:33:44. Total running time: 3min 32s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000003 |
| time_this_iter_s 39.91204 |
| time_total_s 207.45552 |
| training_iteration 4 |
| accuracy 0.5203 |
| loss 1.32945 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 4 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000003
(func pid=4382) [7, 4000] loss: 0.526 [repeated 2x across cluster]
Trial train_cifar_3a47b_00007 finished iteration 7 at 2025-04-23 16:33:55. Total running time: 3min 43s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000006 |
| time_this_iter_s 24.24598 |
| time_total_s 218.78658 |
| training_iteration 7 |
| accuracy 0.565 |
| loss 1.32398 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 7 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000006
(func pid=4382) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000006) [repeated 2x across cluster]
(func pid=4381) [8, 4000] loss: 0.880 [repeated 3x across cluster]
Trial train_cifar_3a47b_00006 finished iteration 8 at 2025-04-23 16:34:03. Total running time: 3min 51s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000007 |
| time_this_iter_s 21.80595 |
| time_total_s 226.20354 |
| training_iteration 8 |
| accuracy 0.3235 |
| loss 1.76241 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 8 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000007
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000007)
(func pid=4380) [5, 6000] loss: 0.424 [repeated 2x across cluster]
(func pid=4381) [9, 2000] loss: 1.724 [repeated 2x across cluster]
Trial status: 7 TERMINATED | 3 RUNNING
Current time: 2025-04-23 16:34:12. Total running time: 4min 0s
Logical resource usage: 6.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 4 207.456 1.32945 0.5203 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 8 226.204 1.76241 0.3235 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 7 218.787 1.32398 0.565 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4380) [5, 10000] loss: 0.253 [repeated 3x across cluster]
Trial train_cifar_3a47b_00007 finished iteration 8 at 2025-04-23 16:34:19. Total running time: 4min 7s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000007 |
| time_this_iter_s 24.35562 |
| time_total_s 243.1422 |
| training_iteration 8 |
| accuracy 0.5739 |
| loss 1.30763 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 8 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000007
(func pid=4382) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000007)
Trial train_cifar_3a47b_00005 finished iteration 5 at 2025-04-23 16:34:22. Total running time: 4min 10s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000004 |
| time_this_iter_s 38.5072 |
| time_total_s 245.96273 |
| training_iteration 5 |
| accuracy 0.5457 |
| loss 1.26669 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 5 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000004
Trial train_cifar_3a47b_00006 finished iteration 9 at 2025-04-23 16:34:24. Total running time: 4min 13s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000008 |
| time_this_iter_s 21.84632 |
| time_total_s 248.04986 |
| training_iteration 9 |
| accuracy 0.3542 |
| loss 1.70474 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 9 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000008
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000008) [repeated 2x across cluster]
(func pid=4382) [9, 2000] loss: 0.938 [repeated 2x across cluster]
(func pid=4381) [10, 2000] loss: 1.692 [repeated 2x across cluster]
(func pid=4381) [10, 4000] loss: 0.836 [repeated 3x across cluster]
Trial status: 7 TERMINATED | 3 RUNNING
Current time: 2025-04-23 16:34:42. Total running time: 4min 30s
Logical resource usage: 6.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 5 245.963 1.26669 0.5457 |
| train_cifar_3a47b_00006 RUNNING 16 4 0.000147684 8 9 248.05 1.70474 0.3542 |
| train_cifar_3a47b_00007 RUNNING 256 256 0.00477469 8 8 243.142 1.30763 0.5739 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
Trial train_cifar_3a47b_00007 finished iteration 9 at 2025-04-23 16:34:42. Total running time: 4min 31s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000008 |
| time_this_iter_s 23.15711 |
| time_total_s 266.29931 |
| training_iteration 9 |
| accuracy 0.5424 |
| loss 1.46234 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 9 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000008
(func pid=4382) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000008)
Trial train_cifar_3a47b_00006 finished iteration 10 at 2025-04-23 16:34:46. Total running time: 4min 35s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00006 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000009 |
| time_this_iter_s 21.77325 |
| time_total_s 269.82311 |
| training_iteration 10 |
| accuracy 0.3718 |
| loss 1.65719 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00006 saved a checkpoint for iteration 10 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000009
Trial train_cifar_3a47b_00006 completed after 10 iterations at 2025-04-23 16:34:46. Total running time: 4min 35s
(func pid=4381) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00006_6_batch_size=8,l1=16,l2=4,lr=0.0001_2025-04-23_16-30-11/checkpoint_000009)
(func pid=4380) [6, 8000] loss: 0.313 [repeated 2x across cluster]
(func pid=4380) [6, 10000] loss: 0.245 [repeated 2x across cluster]
Trial train_cifar_3a47b_00005 finished iteration 6 at 2025-04-23 16:34:59. Total running time: 4min 47s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000005 |
| time_this_iter_s 36.67234 |
| time_total_s 282.63507 |
| training_iteration 6 |
| accuracy 0.5495 |
| loss 1.27286 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 6 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000005
(func pid=4380) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000005)
Trial train_cifar_3a47b_00007 finished iteration 10 at 2025-04-23 16:35:04. Total running time: 4min 52s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00007 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000009 |
| time_this_iter_s 21.62619 |
| time_total_s 287.92551 |
| training_iteration 10 |
| accuracy 0.5714 |
| loss 1.36822 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00007 saved a checkpoint for iteration 10 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000009
Trial train_cifar_3a47b_00007 completed after 10 iterations at 2025-04-23 16:35:04. Total running time: 4min 52s
(func pid=4382) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00007_7_batch_size=8,l1=256,l2=256,lr=0.0048_2025-04-23_16-30-11/checkpoint_000009)
(func pid=4380) [7, 2000] loss: 1.185 [repeated 2x across cluster]
(func pid=4380) [7, 4000] loss: 0.600
Trial status: 9 TERMINATED | 1 RUNNING
Current time: 2025-04-23 16:35:12. Total running time: 5min 0s
Logical resource usage: 2.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 6 282.635 1.27286 0.5495 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00006 TERMINATED 16 4 0.000147684 8 10 269.823 1.65719 0.3718 |
| train_cifar_3a47b_00007 TERMINATED 256 256 0.00477469 8 10 287.926 1.36822 0.5714 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4380) [7, 6000] loss: 0.403
(func pid=4380) [7, 8000] loss: 0.300
(func pid=4380) [7, 10000] loss: 0.238
Trial train_cifar_3a47b_00005 finished iteration 7 at 2025-04-23 16:35:31. Total running time: 5min 19s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000006 |
| time_this_iter_s 31.79974 |
| time_total_s 314.43481 |
| training_iteration 7 |
| accuracy 0.5696 |
| loss 1.2172 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 7 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000006
(func pid=4380) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000006)
(func pid=4380) [8, 2000] loss: 1.198
(func pid=4380) [8, 4000] loss: 0.582
Trial status: 9 TERMINATED | 1 RUNNING
Current time: 2025-04-23 16:35:42. Total running time: 5min 30s
Logical resource usage: 2.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 7 314.435 1.2172 0.5696 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00006 TERMINATED 16 4 0.000147684 8 10 269.823 1.65719 0.3718 |
| train_cifar_3a47b_00007 TERMINATED 256 256 0.00477469 8 10 287.926 1.36822 0.5714 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4380) [8, 6000] loss: 0.381
(func pid=4380) [8, 8000] loss: 0.293
(func pid=4380) [8, 10000] loss: 0.233
Trial train_cifar_3a47b_00005 finished iteration 8 at 2025-04-23 16:36:02. Total running time: 5min 50s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000007 |
| time_this_iter_s 31.24044 |
| time_total_s 345.67525 |
| training_iteration 8 |
| accuracy 0.57 |
| loss 1.21771 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 8 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000007
(func pid=4380) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000007)
(func pid=4380) [9, 2000] loss: 1.161
Trial status: 9 TERMINATED | 1 RUNNING
Current time: 2025-04-23 16:36:12. Total running time: 6min 0s
Logical resource usage: 2.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 8 345.675 1.21771 0.57 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00006 TERMINATED 16 4 0.000147684 8 10 269.823 1.65719 0.3718 |
| train_cifar_3a47b_00007 TERMINATED 256 256 0.00477469 8 10 287.926 1.36822 0.5714 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4380) [9, 4000] loss: 0.566
(func pid=4380) [9, 6000] loss: 0.385
(func pid=4380) [9, 8000] loss: 0.288
(func pid=4380) [9, 10000] loss: 0.226
Trial train_cifar_3a47b_00005 finished iteration 9 at 2025-04-23 16:36:33. Total running time: 6min 21s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000008 |
| time_this_iter_s 31.2056 |
| time_total_s 376.88085 |
| training_iteration 9 |
| accuracy 0.5853 |
| loss 1.16936 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 9 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000008
(func pid=4380) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000008)
(func pid=4380) [10, 2000] loss: 1.132
Trial status: 9 TERMINATED | 1 RUNNING
Current time: 2025-04-23 16:36:42. Total running time: 6min 31s
Logical resource usage: 2.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00005 RUNNING 8 64 0.000353097 4 9 376.881 1.16936 0.5853 |
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00006 TERMINATED 16 4 0.000147684 8 10 269.823 1.65719 0.3718 |
| train_cifar_3a47b_00007 TERMINATED 256 256 0.00477469 8 10 287.926 1.36822 0.5714 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
(func pid=4380) [10, 4000] loss: 0.562
(func pid=4380) [10, 6000] loss: 0.383
(func pid=4380) [10, 8000] loss: 0.279
(func pid=4380) [10, 10000] loss: 0.225
Trial train_cifar_3a47b_00005 finished iteration 10 at 2025-04-23 16:37:04. Total running time: 6min 53s
+------------------------------------------------------------+
| Trial train_cifar_3a47b_00005 result |
+------------------------------------------------------------+
| checkpoint_dir_name checkpoint_000009 |
| time_this_iter_s 31.40795 |
| time_total_s 408.28881 |
| training_iteration 10 |
| accuracy 0.573 |
| loss 1.19828 |
+------------------------------------------------------------+
Trial train_cifar_3a47b_00005 saved a checkpoint for iteration 10 at: (local)/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000009
Trial train_cifar_3a47b_00005 completed after 10 iterations at 2025-04-23 16:37:04. Total running time: 6min 53s
Trial status: 10 TERMINATED
Current time: 2025-04-23 16:37:04. Total running time: 6min 53s
Logical resource usage: 2.0/16 CPUs, 0/1 GPUs (0.0/1.0 accelerator_type:A10G)
+------------------------------------------------------------------------------------------------------------------------------------+
| Trial name status l1 l2 lr batch_size iter total time (s) loss accuracy |
+------------------------------------------------------------------------------------------------------------------------------------+
| train_cifar_3a47b_00000 TERMINATED 16 1 0.00213327 2 1 116.938 2.30351 0.1011 |
| train_cifar_3a47b_00001 TERMINATED 1 2 0.013416 4 1 67.9055 2.31772 0.1006 |
| train_cifar_3a47b_00002 TERMINATED 256 64 0.0113784 2 1 121.957 2.31648 0.1034 |
| train_cifar_3a47b_00003 TERMINATED 64 256 0.0274071 8 2 80.386 2.19641 0.1971 |
| train_cifar_3a47b_00004 TERMINATED 16 2 0.056666 4 1 70.3359 2.32697 0.0995 |
| train_cifar_3a47b_00005 TERMINATED 8 64 0.000353097 4 10 408.289 1.19828 0.573 |
| train_cifar_3a47b_00006 TERMINATED 16 4 0.000147684 8 10 269.823 1.65719 0.3718 |
| train_cifar_3a47b_00007 TERMINATED 256 256 0.00477469 8 10 287.926 1.36822 0.5714 |
| train_cifar_3a47b_00008 TERMINATED 128 256 0.0306227 8 1 37.5834 2.16249 0.2113 |
| train_cifar_3a47b_00009 TERMINATED 2 16 0.0286986 2 1 91.715 2.3265 0.0986 |
+------------------------------------------------------------------------------------------------------------------------------------+
Best trial config: {'l1': 8, 'l2': 64, 'lr': 0.0003530972286268149, 'batch_size': 4}
Best trial final validation loss: 1.1982811905056239
Best trial final validation accuracy: 0.573
(func pid=4380) Checkpoint successfully created at: Checkpoint(filesystem=local, path=/var/lib/ci-user/ray_results/train_cifar_2025-04-23_16-30-11/train_cifar_3a47b_00005_5_batch_size=4,l1=8,l2=64,lr=0.0004_2025-04-23_16-30-11/checkpoint_000009)
Best trial test set accuracy: 0.5953
如果您运行代码，示例输出可能如下所示
Number of trials: 10/10 (10 TERMINATED)
+-----+--------------+------+------+-------------+--------+---------+------------+
| ... | batch_size | l1 | l2 | lr | iter | loss | accuracy |
|-----+--------------+------+------+-------------+--------+---------+------------|
| ... | 2 | 1 | 256 | 0.000668163 | 1 | 2.31479 | 0.0977 |
| ... | 4 | 64 | 8 | 0.0331514 | 1 | 2.31605 | 0.0983 |
| ... | 4 | 2 | 1 | 0.000150295 | 1 | 2.30755 | 0.1023 |
| ... | 16 | 32 | 32 | 0.0128248 | 10 | 1.66912 | 0.4391 |
| ... | 4 | 8 | 128 | 0.00464561 | 2 | 1.7316 | 0.3463 |
| ... | 8 | 256 | 8 | 0.00031556 | 1 | 2.19409 | 0.1736 |
| ... | 4 | 16 | 256 | 0.00574329 | 2 | 1.85679 | 0.3368 |
| ... | 8 | 2 | 2 | 0.00325652 | 1 | 2.30272 | 0.0984 |
| ... | 2 | 2 | 2 | 0.000342987 | 2 | 1.76044 | 0.292 |
| ... | 4 | 64 | 32 | 0.003734 | 8 | 1.53101 | 0.4761 |
+-----+--------------+------+------+-------------+--------+---------+------------+
Best trial config: {'l1': 64, 'l2': 32, 'lr': 0.0037339984519545164, 'batch_size': 4}
Best trial final validation loss: 1.5310075663924216
Best trial final validation accuracy: 0.4761
Best trial test set accuracy: 0.4737
大多数试验已提前停止，以避免浪费资源。性能最佳的试验达到了约 47% 的验证准确率，这在测试集上得到了证实。
就这样！您现在可以调优您的 PyTorch 模型参数了。
脚本总运行时间： ( 7 minutes 7.666 seconds)
