# GeForce RTX 5090 Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/geforce-rtx-5090
> Markdown URL: https://aitinkerers.org/technologies/geforce-rtx-5090.md
> Technology record last updated: 2026-04-04T04:04:38Z
> Generated: 2026-09-21T10:51:28Z

The GeForce RTX 5090 is NVIDIA's flagship Blackwell GPU: delivering up to 2x the performance of the RTX 4090 with 32GB of GDDR7 memory and 21,760 CUDA cores for ultimate 8K gaming and AI workloads.

This is the GeForce RTX 5090, the new king of the hill built on the cutting-edge Blackwell architecture. We’re talking about a serious performance uplift: the card is engineered to be up to two times faster than the previous-gen RTX 4090, leveraging DLSS 4 and Fifth-Gen Tensor Cores for unprecedented AI horsepower. Key specs include a massive 32GB of ultra-fast GDDR7 memory and 21,760 CUDA cores, making it the definitive platform for 8K gaming and high-end creative work. Launched at $1,999, the RTX 5090 sets the new standard for enthusiast-class GPUs: it’s the power you need, delivered.

- Official technology site: https://www.nvidia.com/en-us/geforce/graphics-cards/rtx-5090/
- Public AI Tinkerers demos and talks: 1
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Too Big to Think](https://austin.aitinkerers.org/talks/rsvp_v17yXVtbeFY)

This is an shortened version of the Oral Presentation of my paper Too Big to Think: Capacity, Memorization, and Generalization in Pre-Trained Transformers https://arxiv.org/abs/2506.09099 I gave an oral presentation of this paper at the TTODLer-FM workshop at ICML 2025: https://icml.cc/virtual/2025/workshop/39957 If you're interested in seeing the recording of that presentation, you can click the SlidesLive Video Button for the 9:30am timestamp (Under Too Big to Think: Capacity, Memorization, and Generalization in Pre-Trained Transformers) Here is the paper's abstract: The relationship between memorization and generalization in large language models (LLMs) remains an open area of research, with growing evidence that the two are deeply intertwined. In this work, we investigate this relationship by pre-training a series of capacity-limited Transformer models from scratch on two synthetic character-level tasks designed to separately probe generalization (via arithmetic extrapolation) and memorization (via factual recall). We observe a consistent trade-off: small models extrapolate to unseen arithmetic cases but fail to memorize facts, while larger models memorize but fail to extrapolate. An intermediate-capacity model exhibits a similar shift toward memorization. When trained on both tasks jointly, no model (regardless of size) succeeds at extrapolation. These findings suggest that pre-training may intrinsically favor one learning mode over the other. By isolating these dynamics in a controlled setting, our study offers insight into how model capacity shapes learning behavior and offers broader implications for the design and deployment of small language models.

- Event context: AI Tinkerers Austin AI Demo Night w/ Google & Comet — 2025-10-09 — Austin
- Public talk page: https://austin.aitinkerers.org/talks/rsvp_v17yXVtbeFY

## Related Technologies

- [AMD 9800X3D](https://aitinkerers.org/technologies/amd-9800x3d) ([Markdown](https://aitinkerers.org/technologies/amd-9800x3d.md)) — 1 public demo
- [NanoGPT](https://aitinkerers.org/technologies/nanogpt) ([Markdown](https://aitinkerers.org/technologies/nanogpt.md)) — 4 public demos