Yoshi

Yeah, I’m fine. Thanks.

[よこち @akira6592]

"「HashiCorp Vault」（Vault）と「Red Hat Ansible Automation Platform」（AAP）を中核とした新しいソリューションの提供"

日本IBM、SSL/TLS証明書更新工数ゼロを目指すソリュ��ション--「47日ルール」を克服 - ZDNET Japan https://t.co/EnH9jn0qVw @zdnet_japanより

[インサイトテクノロジー @insight_socio]

【#dbts2025】DataStaxさまよりIBMさまとの7/10 13:30〜D5セッションをご案内！
IBMとDataStaxが描く生成AI基盤の戦略と最新技術連携の全貌を、ここだけの情報も交えて詳しくご紹介！
#AI #データ分析 #ビッグデータ

[インサイトテクノロジー @insight_socio]

【#dbts2025】DataStax様のご案内、続いては「非構造化データ活用の最前線とデータ基盤戦略」（D17）を紹介！
非構造化データをビジネスで活かすための最新データ基盤戦略と、シームレスな移行を実現する技術的アプローチを解説！
#AI #データ管理 #ビッグデータ #データレイク

[インサイトテクノロジー @insight_socio]

【#dbts2025】本日はDataStax様からB13 NTTデータ様事例セッションに関するコメントをご案内。
セルフマネージドCassandraからAstra DBへの無停止移行の実践例と、運用負荷を激減させたNTTデータのノウハウを紹介！
#クラウド #ビッグデータ #データレイク #事例

[那須田淳 @aokumaBT]

これはぜひ読んで欲しい。『ペンツベルクの夜』ボイエ著、木本栄訳(静山社)。終戦間際のドイツで昨日までの正義が逆転する。歴史の回転に翻弄されつつ、自分たちにとって何が大切なのか向きあおうとする少年たち…。史実をベースに描き出した現代ドイツYAの超話題作! ドイツ児童文学賞のYA部門受賞。

[Jonathan Ellis @spyced]

I've been using Brokk to write Brokk for months so I have Opinions on the best models to use for code.

First: if you put a gun to my head and made me choose just one, it's Gemini 2.5 Pro.

Fortunately I can and do use other models when they are a better fit. 🧵

[Yoshi @Yoshi_Kimoto]

@Ohrenje @bandmaid they did say during the show today that Nina will be with the Japanese fans while they are away from Japan so there definitely will be something outside of Japan 🤞😁

[Jonathan Ellis @spyced]

Support for ColPali image search by @ManuelFaysse, @sibille_hugues, @tonywu_71 et al is now in ColBERT Live! That means you can build a compressed, larger-than-memory ColPali index on any vector database.

Things I learned in the process (link to article at the end) 🧵

[Yuki Morishita @yukim]

来週10月2日にミートアップでApache Cassandra v5の新しいインデックスとベクトル検索機能について(深く?)話します。ぜひご参加ください。

Database Engineering Meetup #4 https://t.co/0PvWxQXFun #DBEM

[Apache Cassandra @cassandra]

#ApacheCassandra 5.0-rc1 is out !!

⭐��� Vector Similarity Search (JVector)
⭐️ Storage Attached Indexes
⭐️ Unified Compaction Strategy
⭐️ Trie Memtables and Trie SSTables
⭐️ Java17
⭐️ Dynamic Data Masking
⭐️ Math CQL functions
⭐️ Pluggable crypto provider
⭐️ CQL scalar functions

[Adam Bien @AdamBien]

"JVector: Cutting-Edge Vector Search in Java" an https://t.co/84MwkKWT7N conversation with @spyced is ready to listen: https://t.co/AIsGJ2vUIo #jvector #cassandra #java #podcast

[Sebastián Estévez @syllogistic]

https://t.co/rsUnKOp7qX

[Jonathan Ellis @spyced]

Can your vector database index
- All of Wikipedia?
- On a laptop?
- In under 6h?
- Without compromising accuracy?

JVector can. Article and live demo links in thread.

[Jonathan Ellis @spyced]

I wrote an explainer on vector compression for ANN indexes.

TLDR: binary quantization (BQ) is fast but too lossy for most vectors. Product quantization (PQ) is more accurate, more effective (can compress up to 64x), and faster (when you include the cost to rerank) for most source models. Link in thread.

[Sebastián Estévez @syllogistic]

We love the new streaming support in Assistants API so we spent the last 5 days implementing it.

Today we're announcing support for the new streaming runs functionality in our open source, drop-in compatible version of Assistants API--Astra Assistants.

Try our service for free or run it yourself. We support multiple LLM and embedding providers using @LiteLLM

[Yoshi @Yoshi_Kimoto]

DataStax 🤝 NVIDIA. By integrating #NVIDIA microservices (NIMs), we're able to deliver high-performance RAG solutions with 20x faster embeddings at 80% lower cost.

Learn more 👇 #GTC24 #VectorDB https://t.co/IOM2SZ97bn

[Sebastián Estévez @syllogistic]

Last week we open sourced the server code that we run for our Assistants API service. Now anyone can see exactly how it works or run it themselves on the cloud or on premises. Apache 2 license.

It's a drop in replacement for OpenAI's Assistants API but it persists your data (files, threads, messages, assistants, etc. in astradb / cassandra and uses astradb / cassandra vector indexes powered by jvector for ANN based recall), it supports streaming messages, and allows you to use third party LLMs and embedding providers.

This takes us one step closer to fully open source Assistants and GPTs.

Let us know what you think!

[Cassandra Japan @CassandraJapan]

2/28(水)今夕です！
#cassandra #cassandrameetup

[Cassandra Japan @CassandraJapan]

2/28(水)今夕です！
#cassandra #cassandrameetup

[Cassandra Japan @CassandraJapan]

今週2/28(水)に「CassandraオフラインMeetup, 2024年 - Cassandraは生成AIの夢を見るか」 を開催します！Cassandra Summitの参加報告、Apache Cassandraバージョン5、生成AIにおけるCassandra周りのトピックスを、コミュニティ、LINEヤフー、DataStaxからお話いたします。
https://t.co/MHuuoHCp0F