Meredith (メリ) (@nullinbetween) header
Meredith (メリ) (@nullinbetween) profile picture

Indie app & game builder in Tokyo | Multi-model AI workflows & on-device AI | Grand Winner, Google Gen AI Academy APAC Cohort 3 | Building Digital Omamori

中文/EN/日本語 OK
Joined February 2022
Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)Photo by Meredith (メリ) (@nullinbetween)
Meredith (メリ) retweeted
Put a Gemma-based model in a robot body. First thing it does: scan the room, spot the giant TV, and start a movie break. Awesome to see @NVIDIAAI build on Gemma 4 E4B, fitting live conversation, tool calls, animations, and a vision model onto a single 8GB Jetson!
46
66
5
907
59,240
Meredith (メリ) retweeted
Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values. Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of expression, democracy, and privacy, and measured how often a model adopts China’s party line. After fine-tuning on roughly 53 million tokens, Arete-27B's propaganda rate fell from 22% to 1%, in line with the Western models we tested. Its Hermes Bench score shows no significant performance tradeoffs. The values in an open model are one fine-tune away. Our eval, weights, and training data are available now along with the full blog post: nousresearch.com/arete
1,854
1,496
1,301
15,085
2,668,386
Meredith (メリ) retweeted
The idea came from just chatting with Muse, and Codex took 3 prompts to build the logic. Most effort went into visuals instead! Muse と雑談から原型が生まれて、Codex への指示 3 回で実装完了。一番時間を使ったのは素材の準備でした! #MuseAI #indiegames #個人開発
←Muse prototype | Final Product → ←試作 & 最終版→ #museAI #indiegame #個人開発 #AI開発
1
4
78
Meredith (メリ) retweeted
←Muse prototype | Final Product → ←試作 & 最終版→ #museAI #indiegame #個人開発 #AI開発
1
1
3
122
Meredith (メリ) retweeted
長すぎるLINEのトーク履歴、探すのが大変ではありませんか? 指定期間の抽出、分割、年月ごとの整理ができる3つの小ツールをオープンソースで公開しました。データはブラウザ内のみで処理されます。 試してみる: nullinbetween.github.io/line… MIT: github.com/nullinbetween/lin… #便利ツール #OSS #LINE
One long LINE export can be a lot to dig through. I’ve open-sourced three small tools to extract a date range, split long chats and organise exports by month or year. Your chats stay in your browser. Try it: nullinbetween.github.io/line…
MIT source: github.com/nullinbetween/lin… #GitHub
1
83
Meredith (メリ) retweeted
We rode straight into the festival procession. 🏮 Would you stop and let it pass? appsbymeredith.itch.io/crazy… #IndieGame #GameDev #threejs #opus55
1
1
8
198
Meredith (メリ) retweeted
Family Stories by Meredith is now live on YouTube! 🎉 Animated stories for children exploring festivals and celebrations, with the latest about Japan’s Culture Day (Bunka no Hi文化の日). If you have little ones, come take a look! youtube.com/@FamilyStoriesby…
1
1
28
Meredith (メリ) retweeted
A supermarket sale becomes a traffic event. 🛒🚲 Slow down or find a gap? appsbymeredith.itch.io/crazy… #IndieGame #GameDev #threejs #opus55
1
1
5
150
Meredith (メリ) retweeted
After several rounds of character design, I’ve written up the approach that’s been working best for me so far. This time, I wanted to see how Astra 6 would coordinate the art process compared with Sol. Full write-up on Medium: medium.com/@meredithcheungmm… #openai #codex #indiegame
1
1
3
52
Meredith (メリ) retweeted
Someone asked: does PAPER MOON’s paper tail affect its sound? 🧻🌙 The rustle follows the roll: faster = louder +higher pitch. Airborne = quiet. Landing = a tap. All live Web Audio. Should the tail make sound too? Play here: appsbymeredith.itch.io/paper… #indiedev #opus55 #threejs
3
1
5
133
Meredith (メリ) retweeted
Daycare closes at 8:30. The police want a dance battle. 🚲💥 Crazy Mama Chari is live! Pedal through Tokyo with a gyaru mama and her energetic little passenger. Built with Opus5.5 only. Play in your browser:
appsbymeredith.itch.io/crazy… #indiegame #Threejs #opus55
1
1
58
Meredith (メリ) retweeted
Ready to start building with EmbeddingGemma 2? Our Developer Guide walks you through embedding multimodal content across text, code, images, video, and audio using the sentence-transformers library and more! The workflow: 1️⃣ Load the model (via your preferred framework) 2️⃣ Embed text & code with task prompts 3️⃣ Embed images, video, audio & interleaved inputs 4️⃣ (Optional) Truncate dimensions with Matryoshka to save space
13
72
6
658
29,683
Meredith (メリ) retweeted
Today, we are announcing our new Gemini agent for your business—a universal cloud agent running 24/7 to get your work done. It arrives knowing your tools, data, and work history; it works across knowledge work, content, and code; and it comes with built-in governance to help your data stay secure and cost stay in control → goo.gle/4j7daun
214
477
267
4,103
3,776,355
Meredith (メリ) retweeted
📣「Gemma Meetup 2026 [Road to DevFest Tokyo 1]」 ご参加いただきありがとうございました!🎉 #gdgtokyo #DevFest
13
6
48
4,854
Meredith (メリ) retweeted
オリジナルはOpus5.5のClaude Webで作ったものです Claudeが作ったものはPythonだったので After Effectsで再現できるのでは?と思ってファイル一式をまとめてもらってダウンロード そして、Claudeアプリで試しに再現させてみました ⬇️ その結果
Opus 5.5はこんなこともできるのか GPT Images 2.5 で9分割のダンスの画像を作成 Grokでグリーンバックのダンス動画 Opus 5.5でモーショングラフィックス Astraで音の差し替え
7
61
9
497
58,440