zhisbug

8 ms·

1.
by zhisbug 7mo ago
blog: https://haoailab.com/blogs/dreamverse/
2.
Create a 5s 1080p Video in 4.5s with FastVideo on a Single GPU(1080p.fastvideo.org)
12 points by zhisbug 7mo ago|1 comments
3.
by zhisbug 7mo ago
blog: https://haoailab.com/blogs/fastvideo_realtime_1080p/
4.
by zhisbug 1y ago
live demo is here: https://fastwan.fastvideo.org/
5.
by zhisbug 1y ago
Pokémon is increasingly used to evaluate modern large language models, but current practices lack standardization, and depend heavily on game-specific harness. The Pokémon Red involves three major tasks—navigation, combat control and traini
6.
by zhisbug 1y ago
where other models tops out in a few moves
7.
by zhisbug 1y ago
More details here: https://x.com/haoailab/status/1909712259326394519
8.
by zhisbug 2y ago
We find that spatial perception and spatial reasoning remain very difficult even for the strongest models like o3 or Claude 3.7
9.
Can LLMs play real-time games like supermario (other than Pokemon red)?(twitter.com)
3 points by zhisbug 2y ago|1 comments
10.
by zhisbug 2y ago
gaming agent code here: https://github.com/lmgame-org/GamingAgent/tree/main
11.
by zhisbug 2y ago
https://hao-ai-lab.github.io/blogs/sta/
12.
Sliding Tile Attention: A New Method That Speeds Up HunyuanVideo's Outputs by 3x(old.reddit.com)
2 points by zhisbug 2y ago|1 comments
13.
Fast Video Generation with Sliding Tile Attention(hao-ai-lab.github.io)
12 points by zhisbug 2y ago|2 comments
14.
by zhisbug 2y ago
Sliding tile attention accelerates Hunyuan video generation by 3x with no quality drop and no need for training
15.
More Efficient Chain-of-Thought Reasoning Through Certainty Probing(huggingface.co)
6 points by zhisbug 2y ago|2 comments
16.
by zhisbug 2y ago
Try our demo and let us know
17.
by zhisbug 2y ago
This is pretty clever and seems to have high potential, but it still relies on humans. What if some day all humans cannot outsmart AI?
18.
by zhisbug 2y ago
please try and give us feedback!
19.
AI Space Escape: Playing Games While Evaluting LLM Reasonsing(lmgame.org)
13 points by zhisbug 2y ago|2 comments
20.
by zhisbug 2y ago
We hope to redefine ai evaluation via our gamified AI evaluation platform: game arena!
21.
Efficient LLM Scheduling by Learning to Rank(hao-ai-lab.github.io)
2 points by zhisbug 2y ago|1 comments
22.
by zhisbug 2y ago
We study how to approximate the famous shortest-job-first scheduling in LLM inference!
23.
FastVideo: a lightweight framework for accelerating large video diffusion models(github.com)
110 points by zhisbug 2y ago|24 comments
24.
by zhisbug 2y ago
Hugginface model and data link: https://huggingface.co/FastVideo
25.
MuxServe: Flexible Spatial-Temporal Multiplexing for Multiple LLM Serving(hao-ai-lab.github.io)
2 points by zhisbug 2y ago|1 comments
26.
Consistency LLM: converting LLMs to parallel decoders accelerates inference 3.5x(hao-ai-lab.github.io)
461 points by zhisbug 2y ago|98 comments
27.
Throughput Is Not All You Need: Maxing Goodput in LLM Serving via Disaggregation(hao-ai-lab.github.io)
5 points by zhisbug 3y ago|1 comments
28.
by zhisbug 3y ago
New work from the vLLM team that disaggregates prefill and decoding to maximize goodput (throughput subject to latency constraints) in LLM serving
29.
Break the Sequential Dependency of LLM Inference Using Lookahead Decoding(lmsys.org)
17 points by zhisbug 3y ago|2 comments
30.
by zhisbug 3y ago
New parallel decoding algorithm that trades flops for latency reduction

More ›