simon-mo

Ray 0.7.3 Release Note

Highlights

• RLlib ModelV2 API is ready to use. It improves support for Keras and RNN models, as well as allowing object-oriented reuse of variables. ModelV1 API is deprecated. No migration is needed.

• ray.experimental.sgd.pytorch.PyTorchTrainer is ready for early adopters. Checkout the doc here and we welcome your feedback!

model_creator = lambda config: YourPyTorchModel()
data_creator = lambda config: YourTrainingSet(), YourValidationSet()

simon-mo

@app.route('/imageclassifier/predict/', methods=['POST'])
def image_classifier():
    # Decoding and pre-processing base64 image
    img = image.img_to_array(
        image.load_img(
            # Request from base 64!
            BytesIO(base64.b64decode(request.form['b64'])),
                    # Need to resize it and normalize to 0-1
                    target_size=(224, 224))) / 255.
    # Type casting!

simon-mo

ModelZoo

Python package to query ModelZoo.Live

API

For more specific API details, please check out the docs within the files.

Examples

Initialization

# Connects to a ModelZoo instance running at http://modelzoo.url/

simon-mo

simon-mo

modelzoo

service ModelzooService (services.proto)

Method	Request Type	Response Type	Description
Inference	Payload	Payload	Inference
GetImage	ImageDownloadRequest	ImageDownloadResponse	Website utils
GetMetrics	Empty	MetricItems

simon-mo

Fatbin elf code:
================
arch = sm_30
code version = [1,7]
producer = cuda
host = linux
compile_size = 64bit
compressed

simon-mo

STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_cc
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_cc_batched
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_cn
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_cn_batched
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_ct
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_ct_batched
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_lower_cn
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_lower_cn_batched
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_lower_nc
STT_FUNC         STB_GLOBAL STO_ENTRY      cudnn_maxwell_gcgemm_32x32_lower_nc_batched

simon-mo

import logging
from collections import namedtuple
from enum import Enum

from metis.common import init_log

init_log()

resource_info = namedtuple("ResourceInfo", "node, cores, id")

simon-mo

Deployment notes:

• Dev container: any regular ubuntu container should do but requires (1) docker socket (2) host network

set -x

docker run -d \
       --name mantis-dev \
       --network=host \
 -v /var/run/docker.sock:/var/run/docker.sock \

simon-mo