22
submitted 3 months ago* (last edited 3 months ago) by Lantier@jlai.lu to c/localllama@sh.itjust.works

GGUF quants are already up and llama.cpp was updated today to support it.

top 4 comments
sorted by: hot top controversial new old
[-] Smokeydope@lemmy.world 4 points 3 months ago

Im happy for the Gemma enjoyers who get something out of it. I hear the real world domain knowledge is good. I never tried the Gemma models myself. Apparently its very overly censored and anything google puts out just gives me the icky feeling through association. Anyone remember when they still had "Don't Be Evil" as a motto? Good times.

[-] swelter_spark@reddthat.com 2 points 2 months ago

Yeah, I try to avoid anything with Google attached to it. I have been curious about Gemma 3, though.

[-] Picasso@sh.itjust.works 2 points 2 months ago

Im especially interested in its advanced OCR capabilities. Will be testing this one out on lm studio

[-] brucethemoose@lemmy.world 2 points 2 months ago

I tested these out and found they are really bad at longer context... at least in settings that can sanely fit on most GPUs.

Seems the Gemma family is mostly for short-context work, still.

this post was submitted on 12 Mar 2025
22 points (100.0% liked)

LocalLLaMA

2978 readers
19 users here now

Welcome to LocalLLaMA! Here we discuss running and developing machine learning models at home. Lets explore cutting edge open source neural network technology together.

Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.

As ambassadors of the self-hosting machine learning community, we strive to support each other and share our enthusiasm in a positive constructive way.

Rules:

Rule 1 - No harassment or personal character attacks of community members. I.E no namecalling, no generalizing entire groups of people that make up our community, no baseless personal insults.

Rule 2 - No comparing artificial intelligence/machine learning models to cryptocurrency. I.E no comparing the usefulness of models to that of NFTs, no comparing the resource usage required to train a model is anything close to maintaining a blockchain/ mining for crypto, no implying its just a fad/bubble that will leave people with nothing of value when it burst.

Rule 3 - No comparing artificial intelligence/machine learning to simple text prediction algorithms. I.E statements such as "llms are basically just simple text predictions like what your phone keyboard autocorrect uses, and they're still using the same algorithms since <over 10 years ago>.

Rule 4 - No implying that models are devoid of purpose or potential for enriching peoples lives.

founded 2 years ago
MODERATORS