青木宏文

@hirofumi_aoki

Joined December 2014

586 Following

1.3K Followers

1K Posts

hirofumi_aoki retweeted

Lydia Hallie ✨

@lydiahallie

5 months ago

Claude Code now supports session sharing! You can share your full conversation with team members, or anyone with the link Available on web, desktop, and the mobile app

156

651

290K

hirofumi_aoki retweeted

DAIR.AI

@dair_ai

6 months ago

On building more powerful self-evolving agents. LLM agents struggle to learn from experience after deployment. Fine-tuning is expensive and causes catastrophic forgetting. RAG retrieves based on semantic similarity alone, often pulling noise instead of what actually works. Similarity and utility are not the same thing. This new research introduces MemRL, a framework that enables agents to self-evolve through non-parametric reinforcement learning on episodic memory, keeping the LLM completely frozen. The core idea is to treat memory retrieval as a decision-making problem, not a matching problem. Each memory stores an Intent-Experience-Utility triplet. The utility is a learned Q-value representing expected returns, continuously refined through environmental feedback. MemRL implements Two-Phase Retrieval. First, filter candidates by semantic similarity to ensure relevance. Then, rank by learned Q-values to select what actually works. This distinguishes high-value strategies from semantically similar noise. When the agent succeeds or fails, it updates the Q-values of retrieved memories using Bellman-style backups. No gradient updates to model weights. The frozen LLM provides stable reasoning while the memory evolves plastically. Results across four benchmarks: On HLE (knowledge frontier tasks), MemRL significantly outperforms both RAG and existing memory systems like MemP. The pattern holds on BigCodeBench for code generation, ALFWorld for exploration tasks, and Lifelong Agent Bench for OS and database operations. Analysis confirms a strong correlation between learned utility scores and actual task success, validating that Q-values capture genuine functional value rather than superficial similarity. Why does it matter? Decoupling stable reasoning from plastic memory enables continuous runtime improvement without the catastrophic forgetting or computational costs of fine-tuning. Paper: https://t.co/HvLUnXW2Jd Learn to build effective AI agents in our academy: https://t.co/zQXQt0PMbG

dair_ai's tweet photo. On building more powerful self-evolving agents.

LLM agents struggle to learn from experience after deployment. Fine-tuning is expensive and causes catastrophic forgetting. RAG retrieves based on semantic similarity alone, often pulling noise instead of what actually works.

Similarity and utility are not the same thing.

This new research introduces MemRL, a framework that enables agents to self-evolve through non-parametric reinforcement learning on episodic memory, keeping the LLM completely frozen.

The core idea is to treat memory retrieval as a decision-making problem, not a matching problem. Each memory stores an Intent-Experience-Utility triplet. The utility is a learned Q-value representing expected returns, continuously refined through environmental feedback.

MemRL implements Two-Phase Retrieval. First, filter candidates by semantic similarity to ensure relevance. Then, rank by learned Q-values to select what actually works. This distinguishes high-value strategies from semantically similar noise.

When the agent succeeds or fails, it updates the Q-values of retrieved memories using Bellman-style backups. No gradient updates to model weights. The frozen LLM provides stable reasoning while the memory evolves plastically.

Results across four benchmarks: On HLE (knowledge frontier tasks), MemRL significantly outperforms both RAG and existing memory systems like MemP. The pattern holds on BigCodeBench for code generation, ALFWorld for exploration tasks, and Lifelong Agent Bench for OS and database operations.

Analysis confirms a strong correlation between learned utility scores and actual task success, validating that Q-values capture genuine functional value rather than superficial similarity.

Why does it matter? Decoupling stable reasoning from plastic memory enables continuous runtime improvement without the catastrophic forgetting or computational costs of fine-tuning.

Paper: https://t.co/HvLUnXW2Jd

Learn to build effective AI agents in our academy: https://t.co/zQXQt0PMbG

280

284

28K

hirofumi_aoki retweeted

DAIR.AI

@dair_ai

6 months ago

Autonomous Memory Management in LLM Agents LLM agents struggle with long-horizon tasks due to context bloat. As interaction history grows, computational costs explode, latency increases, and reasoning degrades from distraction by irrelevant past errors. The standard approach is append-only: every thought, tool call, and response permanently accumulates. This works for short tasks but guarantees failure for complex exploration. This research introduces Focus, an agent-centric architecture inspired by slime mold (Physarum polycephalum). The biological insight: organisms do not retain perfect records of every movement through a maze. They retain the learned map. Focus gives agents two new primitives: start_focus and complete_focus. The agent autonomously decides when to consolidate learnings into a persistent Knowledge block and actively prunes the raw interaction history. No external timers or heuristics forcing compression. It declares what you are investigating, explores using standard tools, and then consolidates by summarizing what was attempted, what was learned, and the outcome. The system appends this to a persistent Knowledge block and deletes everything between the checkpoint and the current step. This converts monotonically increasing context into a sawtooth pattern: growth during exploration, collapse during consolidation. Evaluation on SWE-bench Lite with Claude Haiku 4.5 shows Focus achieves 22.7% token reduction (14.9M to 11.5M tokens) while maintaining identical accuracy (60% for both baseline and Focus). Individual instances showed savings up to 57%. Aggressive prompting matters. Passive prompting yielded only 6% savings. Explicit instructions to compress every 10-15 tool calls, with system reminders, increased compressions from 2.0 to 6.0 per task. Capable models can autonomously self-regulate their context when given appropriate tools and prompting, opening pathways for cost-aware agentic systems without sacrificing task performance. Paper: https://t.co/bVkeQlrvGJ Learn to build effective AI agents in our academy: https://t.co/zQXQt0PMbG

dair_ai's tweet photo. Autonomous Memory Management in LLM Agents

LLM agents struggle with long-horizon tasks due to context bloat.

As interaction history grows, computational costs explode, latency increases, and reasoning degrades from distraction by irrelevant past errors.

The standard approach is append-only: every thought, tool call, and response permanently accumulates. This works for short tasks but guarantees failure for complex exploration.

This research introduces Focus, an agent-centric architecture inspired by slime mold (Physarum polycephalum). The biological insight: organisms do not retain perfect records of every movement through a maze. They retain the learned map.

Focus gives agents two new primitives: start_focus and complete_focus. The agent autonomously decides when to consolidate learnings into a persistent Knowledge block and actively prunes the raw interaction history. No external timers or heuristics forcing compression.

It declares what you are investigating, explores using standard tools, and then consolidates by summarizing what was attempted, what was learned, and the outcome. The system appends this to a persistent Knowledge block and deletes everything between the checkpoint and the current step.

This converts monotonically increasing context into a sawtooth pattern: growth during exploration, collapse during consolidation.

Evaluation on SWE-bench Lite with Claude Haiku 4.5 shows Focus achieves 22.7% token reduction (14.9M to 11.5M tokens) while maintaining identical accuracy (60% for both baseline and Focus). Individual instances showed savings up to 57%.

Aggressive prompting matters. Passive prompting yielded only 6% savings. Explicit instructions to compress every 10-15 tool calls, with system reminders, increased compressions from 2.0 to 6.0 per task.

Capable models can autonomously self-regulate their context when given appropriate tools and prompting, opening pathways for cost-aware agentic systems without sacrificing task performance.

Paper: https://t.co/bVkeQlrvGJ

Learn to build effective AI agents in our academy: https://t.co/zQXQt0PMbG

336

297

41K

hirofumi_aoki retweeted

nader dabit

@dabit3

6 months ago

https://t.co/q9anIlmWM0

127

375K

Who to follow

hoku@創作活動

@hoku_unagi

運営サービス：タイッツー、きがるがき、ツクセカ、ノミカケコーヒー、シンプルブログ、無限ツールズ、Nostr検索ポータル etc。大手SIer→ベンチャーCTO→フリーランス。本職は主に業務系Webシステム開発。一時期は行政主催ITセミナー講師や専門学校の非常勤講師もやっていました。隙あらば創作しています。

中村奎太 Keita Nakamura メルコイン

@keita0q

mercoin CEO/メルカリNFT 事業責任者

やす@暗号通貨

@yas_crypto

hirofumi_aoki retweeted

Andrej Karpathy

@karpathy

8 months ago

I’m starting to get into a habit of reading everything (blogs, articles, book chapters,…) with LLMs. Usually pass 1 is manual, then pass 2 “explain/summarize”, pass 3 Q&A. I usually end up with a better/deeper understanding than if I moved on. Growing to among top use cases. On the flip side, if you’re a writer trying to explain/communicate something, we may increasingly see less of a mindset of “I’m writing this for another human” and more “I’m writing this for an LLM”. Because once an LLM “gets it”, it can then target, personalize and serve the idea to its user.

594

13K

hirofumi_aoki retweeted

Yukinori Matsuya🇯🇵 松谷幸紀/double jump.tokyo

@matsuyay

over 1 year ago

📷 新たな挑戦を発表！ 📷 https://t.co/VJLI0zmrcxは、TIS株式会社との協業を通じてステーブルコイン決済支援サービスの展開を進めます！ 📷 📷 この取り組みで実現したいこと 📷 ステーブルコインを活用した新たな決済インフラの構築 📷 Web3 × 既存ビジネスの架け橋となるエコシステムの形成 📷 インバウンド市場を含めた幅広いユースケースへの対応 📷 ステーブルコインの普及 × 事業者向けWalletの重要性私たちhttps://t.co/VJLI0zmrcxは、事業者向けの暗号資産ウォレット「N Suite」を提供しています。ステーブルコインが浸透することで、事業者側のウォレット導入がこれまで以上に重要なポジションに！ N Suiteは、企業が暗号資産を安全かつ簡単に管理できるWalletサービスで、 NFT・ゲーム・エンタメ・決済など幅広いビジネスに対応可能です。（リテール決済に応じた柔軟なAPIの準備も検討中）この協業を通じて、ステーブルコイン決済と事業者向けWalletの普及を加速していきます！ 📷 この領域での協業にご関心のある方、具体的な活用を検討している方、ぜひDMでお話しましょう！ #doublejumpTokyo #TIS #ステーブルコイン #N_Suite #Web3 #ブロックチェーン決済 #協業募集 TISプレス https://t.co/T0JYEw5fOc DJTプレス https://t.co/H6jwvqas0S

matsuyay's tweet photo. 📷 新たな挑戦を発表！ 📷
https://t.co/VJLI0zmrcxは、TIS株式会社との協業を通じてステーブルコイン決済支援サービスの展開を進めます！ 📷
📷 この取り組みで実現したいこと
📷 ステーブルコインを活用した新たな決済インフラの構築
📷 Web3 × 既存ビジネスの架け橋となるエコシステムの形成
📷 インバウンド市場を含めた幅広いユースケースへの対応
📷 ステーブルコインの普及 × 事業者向けWalletの重要性
私たちhttps://t.co/VJLI0zmrcxは、事業者向けの暗号資産ウォレット「N Suite」を提供しています。
ステーブルコインが浸透することで、事業者側のウォレット導入がこれまで以上に重要なポジションに！
N Suiteは、企業が暗号資産を安全かつ簡単に管理できるWalletサービスで、 NFT・ゲーム・エンタメ・決済など幅広いビジネスに対応可能です。（リテール決済に応じた柔軟なAPIの準備も検討中）この協業を通じて、ステーブルコイン決済と事業者向けWalletの普及を加速していきます！
📷 この領域での協業にご関心のある方、具体的な活用を検討している方、ぜひDMでお話しましょう！
#doublejumpTokyo #TIS #ステーブルコイン #N_Suite #Web3 #ブロックチェーン決済 #協業募集
TISプレス
https://t.co/T0JYEw5fOc
DJTプレス
https://t.co/H6jwvqas0S

hirofumi_aoki retweeted

double jump.tokyo株式会社

@doublejumptokyo

over 1 year ago

TISが事業シナジーを見越してDJTへ出資⚡️ 日本国内でのステーブルコイン決済解禁を見据え「ステーブルコイン決済支援サービス」を共同で提供してまります！ https://t.co/RrdneepwKD

青木宏文 @hirofumi_aoki

over 1 year ago

RT @nsuite_en: N Suite is now integrating with Soneium by Sony Block Solutions Labs 🎮✨ By integrating with Soneium, N Suite aims to facili…

hirofumi_aoki retweeted

N Suite｜ステーブルコイン対応企業向けWeb3ウォレット

@nsuite_official

over 1 year ago

📣N SuiteがSoneiumと連携📣 Soneiumを活用する事業者様に使っていただけるようになりました✨ N Suiteを活用することで ✅トークン発行 ✅暗号資産の送金 ✅スマートコントラクトのデプロイなどの作業をスムーズに行えます https://t.co/HiZOu0q22V

hirofumi_aoki retweeted

double jump.tokyo株式会社

@doublejumptokyo

over 1 year ago

【新体制に関するお知らせ】 DJTの新代表取締役に松谷幸紀が就任いたしました。今後につきましては、上野広伸と代表取締役2名体制で国内外からWeb3事業を推進してまいります。 https://t.co/jArzaABlWo

doublejumptokyo's tweet photo. 【新体制に関するお知らせ】
DJTの新代表取締役に松谷幸紀が就任いたしました。

今後につきましては、上野広伸と代表取締役2名体制で国内外からWeb3事業を推進してまいります。

https://t.co/jArzaABlWo https://t.co/ddZP4mUBu9

28K

hirofumi_aoki retweeted

double jump.tokyo株式会社

@doublejumptokyo

over 1 year ago

新作ゲーム『魁三国志大戦』事前登録50万人達成⚡️ 事前登録50万人達成記念といたしまして、武将カード「SR 孫堅」を全員にプレゼント!! #魁三国志大戦本編ゲームを2025年3月に正式リリース決定🔥 https://t.co/QFxkCeAuFe

doublejumptokyo's tweet photo. 新作ゲーム『魁三国志大戦』事前登録50万人達成⚡️

事前登録50万人達成記念といたしまして、武将カード「SR 孫堅」を全員にプレゼント!!

#魁三国志大戦本編ゲームを2025年3月に正式リリース決定🔥
https://t.co/QFxkCeAuFe https://t.co/nT8SiswsOa

hirofumi_aoki retweeted

BOBG

@bobg_info

over 1 year ago

BOBG社は、YOAKE entertainment社 ( @YOAKEofficialJP ) と協業し、「YOAKEトークン（ $YOAKE ）」の発行を行うことを発表いたします。 $YOAKE はSoneium ( @soneium )上で発行いたします。詳細はこちら👇 https://t.co/R8TZCrUs60

hirofumi_aoki retweeted

Oasys Passport(オアシスパスポート)

@oasys_passport

over 1 year ago

【トランザクションおみくじキャンペーン開催🥳】スワップ機能実装を記念して、1月22日からキャンペーン開始！トランザクションを行うだけで抽選に応募できます。 Season1：2025/1/22～2/4 → ブレヒロ(@bfheroes_pr )、マイクリ(@mycryptoheroes )、チャンピオンズタクティクス(@ChampionsVerse ) それぞれのNFT🎁 Season2：2025/2/5～2/18 → bCCPとOASYX RYUZO Gen0🐉 詳細はこちら: https://t.co/ID3aJt6ads #OasysPassport #Oasys #NFT $OAS