yosoyjay

#!/bin/sh

cat /sys/class/infiniband/mlx5_ib*/ports/1/pkeys/*|grep -v 0x0000 | sort | uniq -c

yosoyjay

#!/bin/bash

set -e

NVME_DISKS_NAME=`ls /dev/nvme*n1`
NVME_DISKS=`ls -latr /dev/nvme*n1 | wc -l`

echo "Number of NVMe Disks: $NVME_DISKS"

if [ "$NVME_DISKS" == "0" ]

yosoyjay

#!python
"""Collect distribution of a directory of images and write to text file."""
import logging
from pathlib import Path

import numpy as np
from PIL import Image

logging.basicConfig(level=logging.INFO)

yosoyjay

#!/bin/bash
# Script to partion a drive, format as xfs, add fstab entry, create mount point, and mount it (all).
#
# Notes:
# - Must be run with sudo
# - Assumes all drive is in a single logical partition
# - Mounts to /data

set -e

yosoyjay

#!/bin/bash
# Script to download, unzip, and run organization script on ImageNet ILSVRC from Kaggle.
# 
# Assumes: 
# - You have installed the Kaggle CLI tools (i.e. `pip install kaggle`
# - You have appropriate file located at `~/.kaggle/kaggle.json` with permissions of 600
# - You have adequate disk space (>= 1.5 TiB)
# 
# Time and space notes (from my experience on cloud VMs, YMMV):
# - Downloads takes ~15 mins

yosoyjay

k='Akashiwo', v=0.9230769230769231
k='Alexandrium_singlet', v=0.75
k='Amy_Gony_Protoc', v=0.7692307692307693
k='Asterionellopsis', v=0.5576923076923077
k='Bacteriastrum', v=0.7884615384615384
k='Centric', v=0.8076923076923077
k='Ceratium', v=0.34615384615384615
k='Chaetoceros', v=0.6923076923076923
k='Ciliates', v=0.7307692307692307
k='Coccolithophore', v=0.7674418604651163

yosoyjay

In [4]: model = tf.keras.models.load_model('scwharf-ifcb-xception')
---------------------------------------------------------------------------
JSONDecodeError                           Traceback (most recent call last)
<ipython-input-4-5509cb97abce> in <module>
----> 1 model = tf.keras.models.load_model('scwharf-ifcb-xception')

~/mambaforge/envs/ifcb-updater/lib/python3.9/site-packages/tensorflow/python/keras/saving/save.py in load_model(filepath, custom_objects, compile, options)
    210       if isinstance(filepath, six.string_types):
    211         loader_impl.parse_saved_model(filepath)
--> 212         return saved_model_load.load(filepath, compile, options)

yosoyjay

import queue
import threading

nthreads = 2 # test and play with different values here to see how fast you can get it

# setup up N workers
for i in range(nthreads):
    worker = threading.Thread(
        target=make_chunks_and_plot,
        args=(queue,),

yosoyjay

yosoyjay