antinucleon

5 ms·

1.
Pushing GPT-5.6 Luna from 0% to 56% on ARC-AGI-3 Public(int21.ai)
3 points by antinucleon 1mo ago|0 comments
2.
by antinucleon 1mo ago
AVO’s paper author (ex-NVIDIAN) is here. This work was done half a year ago for GPU kernels, and the same approach has now been applied to ARC-AGI-3. I think people are still underestimating the evolution progress; e.g., recently we made a
3.
The Great Wave Has Arrived (Memo from GLM CEO Jie Tang)(twitter.com)
2 points by antinucleon 3mo ago|0 comments
4.
What Muse Spark 1.1 Taught Us About Enterprise Agent Architecture(int21.ai)
2 points by antinucleon 3mo ago|0 comments
5.
Stop Waiting for a Bigger Context Window(int21.ai)
3 points by antinucleon 3mo ago|1 comments
6.
Show HN: INT21 – Self-Improving PTX Kernel Factory(int21.ai)
3 points by antinucleon 4mo ago|0 comments
7.
7 days of autonomous agent search outperformed FlashAttention-4 and CUDNN(twitter.com)
3 points by antinucleon 6mo ago|0 comments
8.
From Node.js/Python to PTX: The first AI framework generated by AI agents(github.com)
3 points by antinucleon 9mo ago|0 comments
9.
PetaFLOPS Inference Era: 1 Pflops Attention, and Preliminary End-to-End Results(blog.hippoml.com)
2 points by antinucleon 3y ago|0 comments
10.
Unified DataCenter and Local Foundation Model Serving: Beyond Docker Way(blog.hippoml.com)
4 points by antinucleon 3y ago|0 comments
11.
Super AI Creativity App Run with Local GPU on Windows/Linux/MacOS(blog.hippoml.com)
19 points by antinucleon 3y ago|8 comments
12.
by antinucleon 3y ago
Yes
13.
by antinucleon 3y ago
Mojo is trying to create a new language to solve the problem, and specialized for CPU. We are using a more pragmatic way to solve GPU AI computation problem.
14.
by antinucleon 3y ago
We haven't compared yet.
15.
by antinucleon 3y ago
We developed AITemplate majorly for Meta's focus at that time, eg Ads/Ranking need. For HippoML is startup we are building for Generative AI. HippoML is not using AITemplate.
16.
by antinucleon 3y ago
It is actually non-trivial to get GPU run fast, especially on SoC with strong CPU like M2.
17.
by antinucleon 3y ago
Hippo is faster than AITemplate, and supports more generative models. We haven't compared vs TVM, but for absolute token/s on M2 Max, Hippo is able to run decoding on LLAMA with datacenter level GPUs performance (with other SW).
18.
by antinucleon 3y ago
We will disclose more details very soon.
19.
by antinucleon 3y ago
Yes. We support >= 1bit <= 16bit models out of box for various of models.
20.
by antinucleon 3y ago
AITemplate's original designer is here. We quit Meta in January and start HippoML ( https://hippoml.com/ ). We just disclosed our new engine's performance on LLM: https://blog.hippoml.com/large-langu
21.
by antinucleon 8y ago
CuDNN v7 was used in the experiments, in the experiments parts each comparison was listed with version or commit number.
22.
by antinucleon 8y ago
Summary: Tensor program is able to be optimized by using machine learning and transfer learning. The numerical program optimization model is trained on feature from low-level AST of the program. Experiments: Tasks: ResNet, MobileNet, LSTM L
23.
Learning to Optimize Tensor Programs(arxiv.org)
106 points by antinucleon 8y ago|13 comments
24.
by antinucleon 10y ago
The version tested in paper should not have P2P,so it was much slower than current version.
25.
by antinucleon 10y ago
There is a huge distributed performance advantages vs TensorFlow. You can get a hint from Prof. Carlos Guestrin's keynote talk at Data Science Summit 2016. Also, CMU CS Dean Andrew Moore cited MXNet as "is the most scalable framew
26.
Build Your Own TensorFlow with NNVM and Torch(dmlc.ml)
4 points by antinucleon 10y ago|0 comments
27.
Deep Learning in Scala(dmlc.ml)
2 points by antinucleon 11y ago|0 comments
28.
MXNetJS: JavaScript Package for Deep Learning in Browser (without Server)(github.com)
1 points by antinucleon 11y ago|0 comments
29.
by antinucleon 11y ago
The author is here :) Thanks for your suggestion. The detail of network modification will be released in my thesis, there is a "fast good net" but currently I don't have time to make it faster, so I simply use fast poor net.
30.
Deep Learning in a Single Source File for Smart Devices(mxnet.readthedocs.org)
2 points by antinucleon 11y ago|0 comments

More ›