Love this submission from our world models hackathon this weekend - a generative FPS!

Love this submission from our world models hackathon this weekend - a generative FPS!

#ExecuTorch addresses fragmented native deployment for #AI agents as a #PyTorch native platform. It enables voice models across CPU, GPU, and NPU on Android, iOS, Linux, macOS & Windows 🔗 https://t.co/NeQQyUniL4 https://t.co/O3itnoQFoG

Matt Maher tested frontier models in Cursor v. other harnesses. Cursor boosted model performance by 11% on average: Gemini: 52% → 57% GPT-5.4: 82% → 88% Opus: 77% → 93% His benchmark measures how well models implement a 100-feature PRD. @cursor_ai consistently outperformed. https://t.co/hrjCmWMNKN

Matt Maher tested frontier models in Cursor v. other harnesses. Cursor boosted model performance by 11% on average: Gemini: 52% → 57% GPT-5.4: 82% → 88% Opus: 77% → 93% His benchmark measures how well models implement a 100-feature PRD. @cursor_ai consistently outperformed. https://t.co/hrjCmWMNKN

Mistral Small 4 is out https://t.co/IdAowSpHpN

32× efficiency improvement in just the last 3 months, that’s the crazy jump from GPT-5.2 to GPT-5.4! 37 cents/task is essentially almost at human-level efficiency (target was 24 cents/task). This was inconceivable a year ago when o3 cost $4500/task on ARC-AGI-1, 12,000x improved!

Banger report from the Kimi team: Attention Residuals Residual connections made deep Transformers trainable. But they also force uncontrolled hidden-state growth with depth. This work proposes a cleaner alternative. It introduces Attention Residuals, which replace fixed residual accumulation with softmax attention over previous layer outputs. Instead of blindly summing everything, each layer selectively retrieves the earlier representations it actually needs. To keep this practical at scale, they add a blockwise version that compresses layers into block summaries, recovering most of the gains with minimal systems overhead. Why does it matter? Residual paths have barely changed across modern LLMs, even though they govern how information moves through depth. This paper shows that making the mixing content-dependent improves scaling laws, matches a baseline trained with 1.25x more compute, boosts GPQA-Diamond by +7.5 and HumanEval by +3.1, while keeping inference overhead under 2%. Paper: https://t.co/04IG6FDiVr Learn to build effective AI agents in our academy: https://t.co/1e8RZKs4uX

OmniForcing unlocks real-time joint audio-visual generation Achieves ~25 FPS with 0.7s latency—a 35× speedup over offline diffusion models—by distilling bidirectional LTX-2 into a causal streaming generator with maintained multi-modal fidelity. https://t.co/UGYGMyTQOs

OmniForcing unlocks real-time joint audio-visual generation Achieves ~25 FPS with 0.7s latency—a 35× speedup over offline diffusion models—by distilling bidirectional LTX-2 into a causal streaming generator with maintained multi-modal fidelity. https://t.co/UGYGMyTQOs

Subagents are now available in Codex. You can accelerate your workflow by spinning up specialized agents to: • Keep your main context window clean • Tackle different parts of a task in parallel • Steer individual agents as work unfolds https://t.co/QJC2ZYtYcA

Subagents are now available in Codex. You can accelerate your workflow by spinning up specialized agents to: • Keep your main context window clean • Tackle different parts of a task in parallel • Steer individual agents as work unfolds https://t.co/QJC2ZYtYcA

It's been about 20 years since I first started working on embeddings with Yann LeCun (siamese networks!), and I've been fascinated ever since. Gemini Embeddings 2 approaches the platonic ideal: native embedding of text, image, video, audio, and docs to a single space.

It's been about 20 years since I first started working on embeddings with Yann LeCun (siamese networks!), and I've been fascinated ever since. Gemini Embeddings 2 approaches the platonic ideal: native embedding of text, image, video, audio, and docs to a single space.

@Nvidiadev 🗓️ MONDAY @ Booth #338 2PM: Shaping the Future w/ @matthew_d_white 3PM: TensorRT + PyTorch w/ Angela Yi & @narendasan 4PM: DeepSpeed Trillion-Param Training w/ @PKUWZP 5PM: PyTorch Export w/ Angela Yi 6PM: Ray Distributed Computing w/ @robertnishihara #AI #GTC2025

Your AI agent can now generate videos. PixVerse CLI ships today — JSON output, 6 deterministic exit codes, full PixVerse v5.6, Sora2 and Veo 3.1, Nano Banana access from terminal. Same account. Same credits. No new signup. -> Follow+ Reply+RT = 300 Creds(72H ONLY)

this was one of the things i co-led at fair, then fb had ~2b users, embeddings of ~128d made it a 300b-1T parameter model depending on how you count entities (e.g. ad campaigns). at the time, this was big, now it's medium. we trained it purely on distributed cpus

Covo Audio 🔊A end-to-end audio language model from @TencentAI_News https://t.co/tic5cH1A39 ✨ 7B ✨ Audio → Audio in one model ✨ Multi-speaker + voice transfer ✨ Real-time full duplex conversations https://t.co/hFrsxQgzkT

Covo Audio 🔊A end-to-end audio language model from @TencentAI_News https://t.co/tic5cH1A39 ✨ 7B ✨ Audio → Audio in one model ✨ Multi-speaker + voice transfer ✨ Real-time full duplex conversations https://t.co/hFrsxQgzkT

🚨 Want to parse complex PDFs with SOTA accuracy, 100% locally? 📄🔍 At just 0.9B parameters, you can drop GLM-OCR straight into LM Studio and run it on almost any machine! 🥔 🧠 0.9B total parameters 💾 Runs on < 1.5GB VRAM (or ~1GB quantized!) 💸 Zero API costs 🔒 Total data privacy Desktop document AI is officially here. 💻⚡

🚨 Want to parse complex PDFs with SOTA accuracy, 100% locally? 📄🔍 At just 0.9B parameters, you can drop GLM-OCR straight into LM Studio and run it on almost any machine! 🥔 🧠 0.9B total parameters 💾 Runs on < 1.5GB VRAM (or ~1GB quantized!) 💸 Zero API costs 🔒 Total data privacy Desktop document AI is officially here. 💻⚡

Yann LeCun is pumping out papers recently “Temporal Straightening for Latent Planning” This paper shows that by straightening latent trajectories in a world model, Euclidean distance starts to reflect true reachable progress, so it's closer to geodesic/minimum-step distance. This makes gradient-based planning far more stable and effective without relying as heavily on expensive search.

Yann LeCun is pumping out papers recently “Temporal Straightening for Latent Planning” This paper shows that by straightening latent trajectories in a world model, Euclidean distance starts to reflect true reachable progress, so it's closer to geodesic/minimum-step distance. This makes gradient-based planning far more stable and effective without relying as heavily on expensive search.

codex app automations: slack pending replies Review Slack for the current user and update today's daily summary note in /Users/jasonliu/vault at agent/daily-summary-YYYY-MM-DD.md with a single section titled ## Pending Slack Replies. Use Slack search and thread reads across public channels, private channels, DMs, and group DMs to find conversations where the current user is mentioned, directly addressed, or has already participated, and where the latest substantive message is from someone else and the current user has not replied. Focus on recent activity, prioritizing today and the last 36 hours. Read candidate threads before including them. Exclude resolved threads, FYIs that do not need a response, and anything the user already answered later. Rewrite the ## Pending Slack Replies section on each run instead of appending duplicates. For each pending item include: who is waiting, channel or DM name, last message time in America/Los_Angeles, a one-line summary of the ask or blocker, and a short snippet. If a stable Slack link is available, include it. If nothing is pending, keep the section and write - None right now. Keep the rest of the note unchanged.