ornithos

#!/bin/bash
# With some help from ChatGPT, although it was only somewhat correct/helpful

rs_name="..."  # name of replicaset
namespace="..."  # namespace to find pod/rs within

# Function to tail logs for each pod
tail_logs() {
    local pod="$1"
    echo "Tailing logs for pod: $pod"

ornithos

class SklearnSerializer:
    """
    Sklearn recommends we use joblib's dump/load commands, but this uses pickle and pickle
    is a PITA. Numpy's serialization is compact and reliable and performs well across
    different platforms and versions of Python and joblib. This serializer extracts all
    the components of each sklearn object that are fitted (assuming sklearn's convention
    of fitted attributes ending in underscore), and saves them in a numpy .npz file.
    """
    def __init__(self):
        pass

ornithos

"""
Chunk a large XML file consisting of M elements of tag `row` into N JSON blobs 
which can be read directly into memory
"""
import math
import subprocess
from datetime import datetime
from dataclasses import dataclass, field
from pathlib import Path
from typing import Dict

ornithos

ornithos

ornithos

ornithos

# call Chain(Dense(2, 400, NNlib.relu), Dense(400, 784, NNlib.σ))
# 100x with random input, where the matmul/add is called "preactiv"
# for layers 1 and 2, and the nonlinearities are timed separately.
# As always if using sigmoidal transformations, these account for
# much of the forward and backward time pass despite being element-
# wise operations.

#= 
────────────────────────────────────────────────────────────────────
                             Time                   Allocations      

ornithos

using Statistics
using Formatting
using Flux.Tracker
using Flux.Tracker: update!


# Define simple linear model
W = param(rand(2, 5))
b = param(rand(2, 1))