thangngoc89

11 ms·

91.
by thangngoc89 2y ago
RQ never publishes a real protocol for sending and receiving jobs. Also the `data` field could be anything because it's pickled so you could send a whole Python's object there. See [1] for efforts to make RQ cross language compati
92.
by thangngoc89 2y ago
Thanks for the recommendation. I've checked this out and looks like a good alternative with much simpler configuration and less foot-gun than Celery.
93.
by thangngoc89 2y ago
RQ uses Python's Pickle and it doesn't have any serialization protocol so it stays Python only.
94.
by thangngoc89 2y ago
Interesting. I've been looking at a much more simpler system than Celery queue to publish jobs from Golang and consume jobs from Python side (AI/ML stuff). This threads gave a lot of names and links for further investigation.
95.
by thangngoc89 2y ago
I’m wondering if there are language agnostic queues where the queue consumers and publishers could be written in different languages?
96.
by thangngoc89 2y ago
What do you mean by unbounded pricing model? Just curios because I'm in the process of jumping to fly's GPU offering. I can see that you could set auto scale with min and max instances. So from my thinking, you could know the abso
97.
by thangngoc89 2y ago
It's a React library.
98.
by thangngoc89 2y ago
TL;DR: you could require() a ES module and Websockets is built-in
99.
by thangngoc89 2y ago
For popular ML/DL models, you could already export model to ONNX format for interference. However, glue codes are still python and you may need to replace that part with the host’s language
100.
by thangngoc89 2y ago
Isn’t the double underscores __name would makes the method private?
101.
by thangngoc89 3y ago
Yep. Price/performance of multiple 4090s system are way better than the professional cards (Axxx). Also deep learning outside of LLM has many different usage.
102.
by thangngoc89 3y ago
8 GPUs x 16 PCIe lanes each = 128 lanes already. That’s the limit of single CPU platforms.
103.
Backrest: UI for Restic Backups(github.com)
2 points by thangngoc89 3y ago|0 comments
104.
by thangngoc89 3y ago
> You may need to uninstall the driver from DKMS. Your system needs large BAR support and IOMMU off. Can someone point me to the correct tutorial on how to do these things?
105.
by thangngoc89 3y ago
I believe this is mostly for memory capacities. PCIe access between GPUs is slower than soldered RAM on a single GPU
106.
by thangngoc89 3y ago
training on MPS backend is suboptimal and really slow.
107.
by thangngoc89 3y ago
What annoyed me the most is actually clicking on ads from Google Search. Sometime I searched a product and I couldn't clicked on the first result because they are advertisements.
108.
by thangngoc89 3y ago
Usually pi-hole is installed together with Unbound [1] for the function of Blocky. [1] https://docs.pi-hole.net/guides/dns/unbound
109.
by thangngoc89 3y ago
It happened too early 8:03 AM. Probably no visitor or even staff in Taipei 101. Last time videos are all from visitor.
110.
by thangngoc89 3y ago
I’m also in Taipei. It’s the worst one ever I’ve felt in my life. Staying on the 7th floor with noway to escape in time is probably the worst feeling
111.
by thangngoc89 3y ago
> does Cloudflare provide container hosting service agnostic of language -- something like Google Cloud Run? Nope. Their Workers are V8 based so JS or Wasm
112.
by thangngoc89 3y ago
I come you can find the Foundation model with blower style fan. I’ve been scouring 4 of them for a 4x build. I have the other style of fan in a dual build and the heat dispensation is bad.
113.
by thangngoc89 3y ago
Before CUDA 12.0 you have to specify a field in pyproject.toml like this [tool.poetry.dependencies] python = ">=3.10,<3.12.0" torch = {version = "^2.0.1+cu118", source = "torch118"} torc
114.
by thangngoc89 3y ago
I use poetry and direnv. Coming from node/npm, it feels natural for me to just do this. I have really no troubles of installing Pytorch with poetry
115.
by thangngoc89 3y ago
To achieve realtime speed, you need to at least convert it to ONNX or similar, applying quantization for 30fps mode.
116.
by thangngoc89 3y ago
I remember there was scripts that remove the ads .tk injected
117.
by thangngoc89 3y ago
Unrelated to the article but seeing .tk brings back many memories. As a kid without a bank account let's alone an international credit card (VISA/Mastercard), dot.tk is the only way to put a website online with your name. I create
118.
by thangngoc89 3y ago
My favorite would be gpustat [1]. This shows the bare minimum amount of information to let's me know that the training has problems/running well [1] https://github.com/wookayin/gpustat
119.
by thangngoc89 3y ago
Of course it doesn't. `nvidia-smi` needs to be available
120.
by thangngoc89 3y ago
banana.dev has announced sunsetting of their server GPUs offering on March 31st, 2024 https://www.banana.dev/blog/sunset

More ›