🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

The world if we could use emojis in academic papers

🎉 nn.labml.ai now has annotated PyTorch implementations of 50 research papers! ✨ ... and it crossed 5,000 stars last week Would not have been possible without your support by starring, sharing, giving feedback, finding bugs, and contributing. 🙏 👇

Some gradients between two colours look good, others suck. It's easy to tell the difference, but what *is* that difference?

Best of AI Twitter (Sept. 18-25): - OpenAI releases Whisper (human-level speech-to-text) - Adversarial, interactive deepfakes - Twitter "gossip" on how OpenAI gathers GPT-4's trillions of training tokens - The guts of Google's multibillion-parameter ad model ... and more: 1/15

having fun optimizing minGPT today - base: 495ms - zero_grad(set_to_none=True): 492 - torch.jit.script gelu: 463 - OMP_PROC_BIND=CLOSE: 453 - torch.backends.cuda.matmul.allow_tf32: 143 - torch.autocast(torch.bfloat16): 121 - FlashAttention: 102 now: more fused kernels more better

TRACT: Denoising Diffusion Models with Transitive Closure Time-Distillation For single step diffusion, TRACT improves FID by up to 2.4× on the same architecture, and achieves new single-step Denoising Diffusion Implicit Models (DDIM) state-of-the-art FID arxiv.org/abs/2303.04248

New paper TRACT - Faster diffusion model sampling - Single-step diffusion SotA for CIFAR10 and ImageNet64 with L2 loss without architecture changes - Up to 2.4x FID improvement arxiv.org/abs/2303.04248

Apple presents MMAU A Holistic Benchmark of Agent Capabilities Across Diverse Domains Recent advances in large language models (LLMs) have increased the demand for comprehensive benchmarks to evaluate their capabilities as human-like agents. Existing benchmarks, while useful,

marx just wanna fun

4o image generation has arrived. It's beginning to roll out today in ChatGPT and Sora to all Plus, Pro, Team, and Free users.

in my experience defining layers in words really helps with composition ``` Composition from background to foreground: parked cars + buildings -> street sign -> witches ```

Thank you Cheng Lu for also making this happen 🥹 enjoyer of mathematics and memes