Shigeru Kobayashi kotobuki

Claude Desktop IME Fix for Karabiner-Elements

Claude DesktopのOption+Space（クイックエントリー）を使用する際、日本語IMEの変換確定のEnterキーで誤ってメッセージが送信されてしまう問題を解決するKarabiner-Elementsの設定です。

Karabiner-Elements設定

{
  "title": "Claude Desktop IME Fix for Japanese Input",
  "rules": [

This worked on 14/May/23. The instructions will probably require updating in the future.

llama is a text prediction model similar to GPT-2, and the version of GPT-3 that has not been fine tuned yet. It is also possible to run fine tuned versions (like alpaca or vicuna with this. I think. Those versions are more focused on answering questions)

Note: I have been told that this does not support multiple GPUs. It can only use a single GPU.

It is possible to run LLama 13B with a 6GB graphics card now! (e.g. a RTX 2060). Thanks to the amazing work involved in llama.cpp. The latest change is CUDA/cuBLAS which allows you pick an arbitrary number of the transformer layers to be run on the GPU. This is perfect for low VRAM.

Clone llama.cpp from git, I am on commit 08737ef720f0510c7ec2aa84d7f70c691073c35d.

ログファイルの出力

iOS版

iOSの設定 > プライバシー > ヘルスケア > COVID-19接触のログ記録 > 一番下の 接触チェックの記録を書き出す > AirDropやメールでPCに送る。

プロット

https://colab.research.google.com/ を開き
ファイル > ノートブックを新規作成
左側ペインメニューのフォルダ > アップロードアイコンをクリックし、ログファイルをアップロード
下記コードを入力し、▶の再生ボタンをクリック

Retrain a MobileNet model and use it in the browser with TensorFlow.js

DRAFT

Combining TensorFlow for Poets and TensorFlow.js.
Retrain a MobileNet V1 or V2 model on your own dataset using the CPU only.
I'm using a MacBook Pro without Nvidia GPU.

MobileNets can be used for image classification. This guide shows the steps I took to retrain a MobileNet on a custom dataset, and how to convert and use the retrained model in the browser using TensorFlow.js. The total time to set up, retrain the model and use it in the browser can take less than 30 minutes (depending on the size of your dataset).

Example app - HTML/JS and a retrained MobileNet V1/V2 model.

	#!/bin/bash
	SCRIPTNAME=$(basename "$0")

	function realpath () {
	f=$@;
	if [ -d "$f" ]; then
	base="";
	dir="$f";
	else
	base="/$(basename "$f")";

	"""
	Check out STEAM Powered (https://steampoweredshow.com/) where I have conversations
	with women in STEAM to learn a bit about what they do and who they are.
	https://www.steampoweredshow.com/learn-more
	"""

	from pprint import pprint
	from collections import OrderedDict
	import sys
	import re

	// Copyright (c) 2019 aNoken
	#include <M5StickC.h>

	HardwareSerial serial_ext(2);
	typedef struct {
	uint32_t length;
	uint8_t *buf;
	} jpeg_data_t;

	jpeg_data_t jpeg_data;

	import lcd
	import utime
	import sys
	from machine import I2C
	from Maix import GPIO
	from fpioa_manager import *

	i2c = I2C(I2C.I2C0, freq=400000, scl=28, sda=29)
	# And a short delay to wait until the I2C port has finished activating.
	utime.sleep_ms(100)

	<!DOCTYPE html>
	<html>
	<head>
	<meta charset='UTF-8'>
	<meta http-equiv='X-UA-Compatible' content='IE=edge'>
	<meta name='viewport' content='width=device-width, initial-scale=1'>

	<style>
	* {margin: 0}
	</style>

	"""
	Convert YouTube subtitles(vtt) to human readable text.

	Download only subtitles from YouTube with youtube-dl:
	youtube-dl --skip-download --convert-subs vtt <video_url>

	Note that default subtitle format provided by YouTube is ass, which is hard
	to process with simple regex. Luckily youtube-dl can convert ass to vtt, which
	is easier to process.