Nick thanks for your research and work on this. I know people have to eat the revenue is tempting. But once someone is outed as a shill I wonder if they feel it was worth it.
RESEARCH
-
AI-Generated Quiz Contains Factual Error About Makemore Video
By
–
It's a fun idea in principle but question #2 on a random quiz i took is already incorrect. The "correct answer" claims that in makemore bigram video the progression of architectures includes RNNs, which is wrong and RNNs were not covered. The trouble across the board is that AI
-
Synthetic SDRs and Deep AI Research Classification Analysis
By
–
Yes. Number 3 is also deep research inside of ChatGPT or perplexity or you dot com… or many of the synthetic employee startups. (Tricky because you’d think a synthetic SDR would be in category 4 but because they’re non-interactive with the collective, they go in 3).
-

Open-Source Investment Generates 2000x Return for Companies
By
–
Great research on open-source by @Harvard:
– $4.15B invested in open-source generates $8.8T of value for companies (aka $1 invested in open-source = $2,000 of value created)
– Companies would need to spend 3.5 times more on software than they currently do if OSS did not exist I -

AnyCalib: Model-Agnostic Single-View Camera Calibration Method
By
–
AnyCalib: On-Manifold Learning for Model-Agnostic Single-View Camera Calibration AnyCalib is a model-agnostic method for calibrating camera intrinsic parameters from a single in-the-wild image. It bypasses the need for extrinsic cues and works across multiple camera models,
-

RWKV-7 Goose: State-of-the-art Multilingual Language Model
By
–
RWKV-7 "Goose" with Expressive Dynamic State Evolution RWKV-7 "Goose" is a sequence modeling architecture that achieves state-of-the-art multilingual performance with 3 billion parameters. Despite training on fewer tokens than other top models, it excels in both English and
-

New Metric Measures AI Long-Task Completion Ability
By
–
Measuring AI Ability to Complete Long Tasks This paper introduces a new metric, the 50%-task-completion time horizon, to measure AI's ability to complete long tasks. It tracks the time it takes for AI models to match the 50% success rate in tasks that humans typically complete.
-

SynCity: Training-Free 3D World Generation from Text
By
–
SynCity: Training-Free Generation of 3D Worlds This paper presents SynCity, a novel approach to generating large-scale 3D worlds from textual descriptions. It combines the geometric precision of pre-trained 3D generative models with the artistic flexibility of 2D image
-

Aligning Multimodal LLMs with Human Preferences: Survey
By
–
Aligning Multimodal LLM with Human Preference: A Survey This paper reviews alignment algorithms for Multimodal Large Language Models (MLLMs), which handle tasks involving text, visual, and auditory data, aiming to address issues like truthfulness, safety, and alignment with
-

4DGS-1K Achieves 1000+ FPS Dynamic Scene Rendering Innovation
By
–
1000+ FPS 4D Gaussian Splatting for Dynamic Scene Rendering This paper presents 4DGS-1K, a faster and more memory-efficient framework for dynamic scene rendering, achieving over 1000 FPS on modern GPUs by improving 4D Gaussian Splatting (4DGS). Problem: 4DGS faces two