发表于 | 阅读次数:

支持cuda8的tensorflow版本

1
2
py3.5, tf-gpu1.4
py3.6, tf-gpu1.2

发表于 | 阅读次数: 
1
2
3
import operator
x = {1: 2, 3: 4, 4: 3, 2: 1, 0: 0}
sorted_x = sorted(x.items(), key=operator.itemgetter(1))

发表于 | 阅读次数: 
1
setattr(config, key, val)
1
2
3
4
5
6
7
8
9
10
11
12
13
def __setitem__(self, key, value):
    if key == "nepochs":
        Config.nepochs = value
    elif key == "dim_char":
        Config.dim_char = value
    elif key == "batch_size":
        Config.batch_size = value
    elif key == "lr_method":
        Config.lr_method = value
    elif key == "lr":
        Config.lr = value
    else:
        raise ValueError("not exist this attr")

发表于 | 阅读次数:

使用绝对路径

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
tune.register_trainable("train_func", train_func)

tune.run_experiments({
    "my_experiment": {
        "run": "train_func",
        "stop": {"mean_accuracy": 99},
        "local_dir": "/home/ray_results/tmp2",
        "trial_resources": {'cpu': 1, 'gpu': 1},
        # "num_gpus": 1,
        "config": {
            "batch_size": tune.grid_search([10, 20, 30]),
            # "resources": {"cpu": 1, "gpu": 1}
            # "momentum": tune.grid_search([0.1, 0.2]),
        }
    }
})

阅读全文 »

发表于 | 阅读次数: 
1
2
3
http://developer.download.nvidia.com/compute/redist/cudnn/v7.1.2/cudnn-9.0-linux-x64-v7.1.tgz

wget http://developer.download.nvidia.com/compute/redist/cudnn/v6.0/cudnn-8.0-linux-x64-v6.0.tgz

发表于 | 阅读次数: 
1
export PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin

发表于 | 阅读次数: 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
$ sudo rmmod nvidia
rmmod: ERROR: Module nvidia is in use by: nvidia_modeset nvidia_uvm

sudo rmmod nvidia
sudo nvidia-smi

$lsmod | grep nvidia
nvidia_uvm            647168  0
nvidia_drm             53248  0
nvidia_modeset        790528  1 nvidia_drm
nvidia              12144640  152 nvidia_modeset,nvidia_uvm            12144640  152 nvidia_modeset,nvidia_uvm

sudo lsof -n -w  /dev/nvidia*
sudo rmmod nvidia_uvm
sudo rmmod nvidia_modeset
sudo rmmod nvidia
nvidia-smi

发表于 | 阅读次数: 
1
2
3
conda create -n tf-gpu3 pip python=3.3
source activate tf-gpu3
pip install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-1.6.0-cp36-cp36m-linux_x86_64.whl

发表于 | 阅读次数: 
1
2
3
4
5
6
7
import tensorflow as tf

a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
c = tf.matmul(a, b)
sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))
print sess.run(c)