Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

github.com

Post reply on HN
Print view View original

171–180 of 232 posts

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#171

Post by izuchukwu »

itake wrote: post #76
nharada wrote: post #70

Earlier quoted context omitted.

Wait why is renting a GPU in the cloud not a solution? You can even try multiple options and see which ones are capable enough for your use case. Look into some barebones cloud GPU services, for example Lambda Labs which is significantly cheaper than AWS/GCP but offers basically nothing besides the machine with a GPU. You could even try something like Vast in which people rent out their personal GPU machines for chea…

My annoyance was managing state. I’d have to spend hours installing tools, downloading data, updating code, then when I want to go to bed I have to package it up and store as much as I can on s3 before shutting off the $$ server.

You should check out https://brev.dev. You can rent GPUs, pause instances, use your own AWS/GCP accounts to make use of your credits, and the CLI lets you use your GPU as if it’s on your local machine.

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#172

Post by armatav »

l33t233372 wrote: post #170
armatav wrote: post #54

Imagine if we used these types of models for like 500 years and it locked their vocabulary in time, disallowing any further language blending; then somehow the servers turned off and nobody could communicate across language barriers anymore. Someone should write that down in some sort of short-story involving a really tall structure.

Imagine if we used them everyday for 500 years but tune all the languages to be ever so slightly similar each day until there is one language.

Call it “Inglesh” or something phonetically similar

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#173

Post by throwaway6734 »

itake wrote: post #76
nharada wrote: post #70

Earlier quoted context omitted.

Wait why is renting a GPU in the cloud not a solution? You can even try multiple options and see which ones are capable enough for your use case. Look into some barebones cloud GPU services, for example Lambda Labs which is significantly cheaper than AWS/GCP but offers basically nothing besides the machine with a GPU. You could even try something like Vast in which people rent out their personal GPU machines for chea…

My annoyance was managing state. I’d have to spend hours installing tools, downloading data, updating code, then when I want to go to bed I have to package it up and store as much as I can on s3 before shutting off the $$ server.

Docker?

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#175

Post by kaycey2022 »

ipsum2 wrote: post #159
kaycey2022 wrote: post #155

Why is Meta open sourcing their AI work like this? Is it because they don't have a great reputation even among tech companies?

Meta has had a culture of open source, even before their reputation got bad (React, RocksDB, Thrift, folly, etc). They open source a lot of their AI models (instead of just source) because they don't have an enterprise cloud to sell these things, unlike Amazon/Google/Microsoft's offerings (which all offer speech transcription and text to speech as services).

I understand that. I’m asking in the context that a number of business commentators have come out and said that Google open sourcing the transformer technology was in itself a big mistake, in hindsight. I think the business side in Meta would have taken a similar view.

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#176

Post by ElectricalUnion »

sneak wrote: post #77
itake wrote: post #76

Earlier quoted context omitted.

My annoyance was managing state. I’d have to spend hours installing tools, downloading data, updating code, then when I want to go to bed I have to package it up and store as much as I can on s3 before shutting off the $$ server.

This is what containers solve. Don't waste time manually installing things. Store state in a database via the app on a different host.

I used to think that the 300MB containers I shipped to prd were reasonably unoptimized.

Those AMD ROCm containers are like 14GiB compressed.

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#177

Post by ElectricalUnion »

sneak wrote: post #109
itake wrote: post #104

Earlier quoted context omitted.

Yep... you spend hours messing around with docker containers and debugging all the weird build errors. I am less familiar with storing data in a db (for ml hosting concerns), but I'd imagine it would add overhead (as opposed to accessing files on disk). You also have to deal with hosting a db and configuring the schema.

You "spend hours messing around" with everything you don't know or understand at first. One could say the same about writing the software itself. At its core Dockerfiles are just shell scripts with worse syntax, so it's not really that much more to learn. Once you get it done once, you don't have to screw around with it anymore, and you have it on any box you want in seconds. In either case you have to spend hours sc…

> Storing data in a database vs in files on disk is like application development 101

It's OK until you're dealing with, say, 130GiB of tensors, on what is effectively a binary blob that needs to be mostly in VRAM somehow.

I really don't want to read 130GiB of blobs from a database all the time.

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#178

Post by ElectricalUnion »

eigenvalue wrote: post #105

I just wanted to test out the TTS locally on a powerful Ubuntu 22.04 machine, but the process for setting it up seems pretty broken and poorly documented. After 20 minutes of trying I finally gave up since I couldn't get the VITS dependency to build (despite having a fully updated machine with all required compilers). It seems like they never really bother to see if the stuff works on a fresh machine starting from sc…

This thing uses pytorch, the bane of everything that attempts to make pypi installs sane.

As a human, you need to read https://pytorch.org/get-started/locally/ and install the correct version, depending on your pytorch-version/os/packaging-system/hardware-platform combo. It's bad. It's several conflicting index-urls messing with your requirements. It also doesn't work if you pick the wrong thing.

Or we need something like a "setup.py" that is turing-complete, to dynamically pick up the correct dependency for you.

`pip install -r requirements.txt` is not enough.

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#179

Post by spudlyo »

zamnos wrote: post #168
copperx wrote: post #164

Earlier quoted context omitted.

What do you mean by assembly needed? I looked them up and they look like normal gfx cards. Am I missing something?

They're datacenter GPUs so you need special power supplies and an adapter to connect it to a regular motherboard.

From what I’ve read cooling can also be a challenge.

Re: Meta AI announces Massive Multilingual Speech code, models for 1000+ languages

#180

Post by chaxor »

nharada wrote: post #70
qwertox wrote: post #68

I would like to use stuff like this as a side-project. Buy a Nvidia Geforce GPU and stick it into my 24/7 server and play around with it in my free time, to see what can be done. The issue with all these AI models is that there's no information on which GPU is enough for which task. I'm absolutely clueless if a single RTX 4000 SFF with its 20GB VRAM and only 70W of max power usage will be a waste of money, or really…

Wait why is renting a GPU in the cloud not a solution? You can even try multiple options and see which ones are capable enough for your use case. Look into some barebones cloud GPU services, for example Lambda Labs which is significantly cheaper than AWS/GCP but offers basically nothing besides the machine with a GPU. You could even try something like Vast in which people rent out their personal GPU machines for chea…

Cloud has a variable price (up to the whim of whatever they decide the price to be that day) so it's uncertain, but typically it is far" more expensive for this type of application. So when faced with 1) probably far more expensive, or 2) single price that will be cheaper, is always available, and has far more uses, I think most would choose 2) for self-hosting. Cloud is very rarely* a good option.
Post reply on HN