helloericsf

Wow, didn't know there were talks about relaxing GDPR. Can you share a few links? Many thanks.

Thanks for sharing. I bet their DPO and EU customers are super interested in the findings. The CEO should have handled it better, IMO.

Not a DBA, how do you do DB permission rollout gating?

Thanks for reaching out. Just reposted.

If you're in SF, you don't want to miss this. The Qwen team is making their first public appearance in the United States, with the VP of Qwen Lab speaking at the meetup below during SF teach week. https://partiful.com/e/P7E418jd6Ti6hA40H6Qm Rare opportunity to directly engage with the Qwen team members.

The base plan limit is not hard to hit. Then you're on the usage based rocket.

Honestly depends on when they got in. Seed investors? They're probably fine with their preferences. Series B and beyond? That's where it gets messy. What round you thinking?

How does it stack up against the new Grok 4 model?

3 repos DualPipe https://github.com/deepseek-ai/DualPipe EPLB https://github.com/deepseek-ai/eplb profile-data https://github.com/deepseek-ai/profile-data

Github: https://github.com/deepseek-ai/DeepGEMM

- Up to 1350+ FP8 TFLOPS on Hopper GPUs - No heavy dependency, as clean as a tutorial - Fully Just-In-Time compiled - Core logic at ~300 lines - yet outperforms expert-tuned kernels across most matrix sizes - Supports dense layout and two MoE layouts

HF:https://huggingface.co/Wan-AI/Wan2.1-T2V-14B Github:https://github.com/Wan-Video/Wan2.1

Wan2.1, a comprehensive and open suite of video foundation models that pushes the boundaries of video generation. Wan2.1 offers these key features:

- SOTA Performance: Wan2.1 consistently outperforms existing open-source models and state-of-the-art commercial solutions across multiple benchmarks. - Supports Consumer-grade GPUs: The T2V-1.3B model requires only 8.19 GB VRAM, making it compatible with almost all consumer-grade GPUs. It can generate a 5-second 480P video on an RTX 4090 in about 4 minutes (without optimization techniques like quantization). Its performance is even comparable to some closed-source models. - Multiple Tasks: Wan2.1 excels in Text-to-Video, Image-to-Video, Video Editing, Text-to-Image, and Video-to-Audio, advancing the field of video generation. - Visual Text Generation: Wan2.1 is the first video model capable of generating both Chinese and English text, featuring robust text generation that enhances its practical applications. - Powerful Video VAE: Wan-VAE delivers exceptional efficiency and performance, encoding and decoding 1080P videos of any length while preserving temporal information, making it an ideal foundation for video and image generation.

this might help: https://x.com/main_horse/status/1894215779521794058/photo/1

- Efficient and optimized all-to-all communication - Both intranode and internode support with NVLink and RDMA - High-throughput kernels for training and inference prefilling - Low-latency kernels for inference decoding - Native FP8 dispatch support - Flexible GPU resource control for computation-communication overlapping X: https://x.com/deepseek_ai/status/1894211757604049133

They don't have h100. wink,wink.

Don't think the decision is based on infra, or any technical reasons. It's more on the service support side. How a 200-person company supports 44M iPhone users in China?

What do you mean by "lower"? To my understanding, they will open 5 infra related repos this week. Let's revisit your comparison question on Friday.

X:https://x.com/deepseek_ai/status/1893836827574030466 BF16 support Paged KV cache (block size 64) 3000 GB/s memory-bound & 580 TFLOPS compute-bound on H800

Blog: https://qwenlm.github.io/blog/qwen2.5-max/ Seems like the model is not available for download on HF/Github

Release blog:https://www.minimaxi.com/en/news/minimax-01-series-2 Report: https://filecdn.minimax.chat/_Arxiv_MiniMax_01_Report.pdf

True. More benchmark metrics here: https://x.com/deepseek_ai/status/1872242657348710721/photo/2

HF link: https://huggingface.co/deepseek-ai/DeepSeek-V3 Aider link: https://aider.chat/docs/leaderboards/ Pricing($0.14/$0.28 per 1M tokens) reference:https://x.com/xingyaow_/status/1872145835699691675?ref_src=t... LiveBench via reddit: https://www.reddit.com/media?url=https%3A%2F%2Fpreview.redd....

Tweets from Chinese Scholars call for investigation: - https://x.com/xiaoyongzhu001/status/1867724288369004716 - https://x.com/tydsh/status/1867965300706255060 - https://x.com/sunjiao123sun_/status/1867744557200470422

NeurIPS: - https://x.com/NeurIPSConf/status/1867759121023336464 - https://neurips.cc/Conferences/2024/StatementOnInclusivity Picard: - https://www.media.mit.edu/posts/neurips-apology-moving-forwa... - https://neurips.cc/Conferences/2024/PicardStatement

Call Jensen and Lisa!lol

- 389 billion parameters and 52 billion activation parameters, capable of handling up to 256K tokens. - outperforms LLama3.1-70B and exhibits comparable performance when compared to the significantly larger LLama3.1-405B model.

Thank you! Great insight.

Congrats on the launch! Curious to know, which OSS models you see works best at the moment?

Credit to https://x.com/calebfahlgren https://huggingface.co/spaces/cfahlgren1/model-release-heatm...

OpenCL was discussed more frequently in classes about a decade ago. However, I haven't heard it mentioned in the last five years or so.

How many coding co-pilot we have on the market?