learndeeply

8 ms·

31.
by learndeeply 4y ago
Partially coincidence, but also ICLR submission deadline was yesterday, so now papers can be public.
32.
by learndeeply 4y ago
Grain increases perceived sharpness and decreases gradient banding, so there's more benefits than just nostalgia.
33.
by learndeeply 4y ago
Most people are buying AMD GPUs for gaming or productivity (e.g. blender), not machine learning, and its working great for them. ROCm is currently a joke. Maybe in a few years AMD will care enough to try to participate in the ML hardware sp
34.
by learndeeply 4y ago
This applies for all neural networks. Depending on how much money you're willing to spend, in descending order: DGX (computer with 8 A100s, $150,000), A100 (80GB, $15,000), A6000 ($5000), RTX 3090 ($1000).
35.
by learndeeply 4y ago
Thanks for answering questions! Are the example given in the blog post considered zero-shot learning? Was the model trained on the websites in the examples given (e.g. on the Redfin site)? How much labeled data was used?
36.
by learndeeply 4y ago
This is mentioned directly in the article: Taichi vs. Numba: As its name indicates, Numba is tailored for Numpy. Numba is recommended if your functions involve vectorization of Numpy arrays. Compared with Numba, Taichi enjoys the following
37.
by learndeeply 4y ago
UC Berkeley has a $6 billion endowment. It's not impossible, but improbable.
38.
by learndeeply 4y ago
> If you want to actually deploy AI models On mobile*. > I often find myself using the TF data pre-processing pipeline even from inside PyTorch the tf.data pipeline is quite nice, has some neat auto-tuning features.
39.
by learndeeply 4y ago
I don't think so, Colab pro limitations are precisely because they weren't charging by compute unit, so they were over-subscribed.
40.
by learndeeply 4y ago
Race to the bottom implies that they're only competing on price. Here, they're competing on new functionality as well. If DALL-E's outputs were substantially better than Stable Diffusion, more people would use it, even if it
41.
by learndeeply 4y ago
It's a race to the top. New functionality is added and the model is improved week over week.
42.
by learndeeply 4y ago
Frustratingly it ignores ViT and other image/video transformer models.
43.
by learndeeply 4y ago
> No one knows for certain how language models that are 10x or even 100x larger than current state-of-the-art ones will perform Read the Deepmind Chinchilla paper, it answers this
44.
by learndeeply 4y ago
Meta comment: It's pretty bold for a blog that started a year a year ago to call themselves a "news agency" and to claim to speak for the Japanese people.
45.
by learndeeply 4y ago
> My unsubstantiated guess is that this is one of the most comprehensively-typed Python codebases out there for its size. Not important, but FAANG companies have several orders of magnitude more strictly-typed Python than this.
46.
by learndeeply 4y ago
You're confusing the H100, a high-end data-center card with the RTX 4000 series.
47.
by learndeeply 4y ago
Both - the first is for sharding tensors across GPUs, the second is to do an all reduce (e.g. for distributed data parallel to synchronize gradients)
48.
by learndeeply 4y ago
Out of curiosity, is $10/month prohibitively expensive for you or other developers? Copilot is free for open source developers and students.
49.
by learndeeply 4y ago
Didn't Milvus (vector db, wrapper around FAISS) come before Pinecone?
50.
by learndeeply 4y ago
Why do people keep on re-inventing pipelines? There's so many already, all with almost identical syntax. This one is similar to Flyte.
51.
by learndeeply 4y ago
> But a group of N sqlite databases is an N-writer database. And mvsqlite provides the necessary mechanisms to do serializable cross-database transactions without additional overhead. I'm confused, are these databases planned to be
52.
by learndeeply 4y ago
Article was posted in 2019, in case other people missed it like I did.
53.
by learndeeply 4y ago
Unless you've demonstrated that FSD follows scaling laws, this comment is pretty meaningless.
54.
by learndeeply 4y ago
This looks really fun, modern incarnation of Google Wave. Made a test room here: https://yboard.lol/lobby?join=hackernews
55.
by learndeeply 4y ago
If anyone is wondering, Mintlify is not open source: https://github.com/mintlify/mintlify/blob/main/server/LICENS...
56.
by learndeeply 4y ago
How is that related to Sheryl stepping down as COO at all?
57.
by learndeeply 4y ago
I don't get it. What in the license prevents users from removing the telemetry? AGPL just means the user needs to open source that change, right? Edit: To remove telemetry, just call: from mitoinstaller.user_install import go_pro;
58.
by learndeeply 4y ago
When the concrete weathers, will the plastic in the bricks turn into micro-plastic?
59.
by learndeeply 4y ago
Your questions are answered in the first sentence of the article. > Zstd or Zstandard (RFC 8478, first released in 2015) is a popular modern compression algorithm. It’s smaller (better compression ratio) and faster than the ubiquitous Zl
60.
by learndeeply 4y ago
It's a waste if you're not. https://www.eataly.com/us_en/magazine/how-to/leftover-parmes...

More ›