treesciencebot

10 ms·

1.
Faster than real-time video generation with #1 quality model(fal.ai)
2 points by treesciencebot 17d ago|0 comments
2.
MiniMax H3 Max sets the new Pareto Frontier for video generation(twitter.com)
3 points by treesciencebot 21d ago|0 comments
3.
by treesciencebot 5mo ago
fal | San Francisco, CA / Remote | Full-time We are on a mission to build world’s first generative media platform for developers. We are running inference on tens of thousands of GPUs, and looking for people (in all functions) to help
4.
by treesciencebot 10mo ago
we build our sandbox just for this use case, fal.ai/sandbox. take the same image/prompt, and compare across tens of models.
5.
Show HN: Open-source image restylization canvas(infinite-kanvas.com)
1 points by treesciencebot 1y ago|0 comments
6.
by treesciencebot 1y ago
One interesting feature that gets enabled with open weights is adding new capabilities (tasks) to these editing models. They generalize quite well with low samples (30 ish). We talk about it here https://blog.fal.ai/announci
7.
by treesciencebot 1y ago
the main question is going to be software stack. NVIDIA is already shipping NVFP4 kernels and perf is looking good. It took a really long time after MI300X's that the FP8 kernels were OK (not even good, compared to almost perfect FP8 s
8.
by treesciencebot 1y ago
fal | Growth Engineer | San Francisco (on site 5 days/wk) Help us scale generative‑media infra: hack demos in the AM, pitch partners over coffee. You’ll build quick client libs & microsites, run data A/Bs, write content that d
9.
by treesciencebot 1y ago
author is: https://x.com/madebyollin
10.
World Emulation via Neural Network(madebyoll.in)
250 points by treesciencebot 1y ago|46 comments
11.
by treesciencebot 2y ago
GH200 is nowhere near $343,000 number. You can get a single server order around 45k (with inception discount). If you are buying bulk, it goes down to sub-30k ish. This comes with a H100's performance and insane amount of high bandwith
12.
by treesciencebot 2y ago
at a relatively new high-rise in rincon hill, AT&T still charges 80-90$ for 1 gig symmetrical (same with webpass/xfinity).
13.
by treesciencebot 2y ago
For traditional LLMs this might be true (especially large MoEs at bs=1) but I highly disagree with "multi-modal models" phrase since most of the models that output in other modalities are generally compute bound. Which means les
14.
by treesciencebot 2y ago
For anyone that wants to test the original (non-distilled) HunyuanVideo (which is an amazing model) we have 580p version taking under a minute and 720p version taking around 2.5-3 minutes in our playground: https://fal.ai/mo
15.
MiniMax (Hailuo) video-01-Live(blog.fal.ai)
2 points by treesciencebot 2y ago|0 comments
16.
by treesciencebot 2y ago
Hunyuan at other providers like fal.ai is cheaper than SORA for the same resolution (720p 5 seconds gets you ~15 videos for $20 vs almost 50 videos at fal). It is slower than SORA (~3 minutes for a 720p video) but faster than replicate&#x27
17.
by treesciencebot 2y ago
I think the main problem is with 2) since 1 is entirely up to the content creator's discretion.
18.
by treesciencebot 2y ago
i don't think anyone has done real-time multi-speaker dialog generation before
19.
by treesciencebot 2y ago
Liability. Till we solve this, we cant really give AI any real responsibilities.
20.
Powerful infrastructure innovations for your AI-first future(cloud.google.com)
1 points by treesciencebot 2y ago|0 comments
21.
Recraft introduces a revolutionary AI model that thinks in design language(recraft.ai)
1 points by treesciencebot 2y ago|0 comments
22.
by treesciencebot 2y ago
It excels at particularly text and scene composition, as well as being able to generate vector graphics. You can use it through their website or through fal.ai https://fal.ai/models/fal-ai/recraft-v3/playgroun
23.
New SOTA text-to-image model by Recraft(recraft.ai)
3 points by treesciencebot 2y ago|1 comments
24.
Scent Teleportation Update: We Did It(osmo.ai)
1 points by treesciencebot 2y ago|0 comments
25.
Moondream raises $4.5M to prove that smaller AI models can still pack a punch(venturebeat.com)
3 points by treesciencebot 2y ago|0 comments
26.
Mochi 1: A new SOTA in open-source video generation models(genmo.ai)
2 points by treesciencebot 2y ago|0 comments
27.
by treesciencebot 2y ago
Much faster than OpenAI's real-time mode, wow! Quality seems to be on par if not better as well.
28.
Friday Facts #432 – Aquilo(factorio.com)
1 points by treesciencebot 2y ago|0 comments
29.
by treesciencebot 2y ago
all high end "gaming" rigs are either using ~16 real cores or 8:24 performance/efficiency cores these days. threadripper/other HEDT options are not particularly good at gaming due to (relatively) lower clock speed /
30.
Show HN: Fal.ai – generative media platform for developers(fal.ai)
3 points by treesciencebot 2y ago|0 comments

More ›