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

Google releases Gemma 4 open models

deepmind.google

Post reply on HN
Print view View original

221–230 of 507 posts

Re: Google releases Gemma 4 open models

#221

Post by chrisweekly »

polishdude20 wrote: post #132
evilelectron wrote: post #57

Earlier quoted context omitted.

Daniel, your work is changing the world. More power to you. I setup a pipeline for inference with OCR, full text search, embedding and summarization of land records dating back 1800s. All powered by the GGUF's you generate and llama.cpp. People are so excited that they can now search the records in multiple languages that a 1 minute wait to process the document seems nothing. Thank you!

Hey in really interested in your pipeline techniques. I've got some pdfs I need to get processed but processing them in the cloud with big providers requires redaction. Wondering if a local model or a self hosted one would work just as well.

Disclaimer: I'm an AI novice relative to many here. FWIW last wknd I spent a couple hours setting up self-hosted n8n with ollama and gemma3:4b [EDIT: not Qwen-3.5], using PDF content extraction for my PoC. 100% local workflow, no runtime dependency on cloud providers. I doubt it'd scale very well (macbook air m4, measly 16GB RAM), but it works as intended.

Re: Google releases Gemma 4 open models

#222

Post by egeres »

danielhanchen wrote: post #2

Thinking / reasoning + multimodal + tool calling. We made some quants at https://huggingface.co/collections/unsloth/gemma-4 for folks to run them - they work really well! Guide for those interested: https://unsloth.ai/docs/models/gemma-4 Also note to use temperature = 1.0, top_p = 0.95, top_k = 64 and the EOS is " ". " thought\n" is also used for the thinking trace!

Thank you and your brother for all the amazing work, it's really inspiring to others <3

Re: Google releases Gemma 4 open models

#223

Post by retinaros »

lostmsu wrote: post #182
culi wrote: post #171

Earlier quoted context omitted.

Do you have a single gallery page where we can see all the pelicans together. I'm thinking something similar to https://clocks.brianmoore.com/ but static.

Not exactly what you asked for but try https://pelicans.borg.games/

what the sorcery is that https://static.simonwillison.net/static/2024/recraft-ai-peli...

I tried their model and asking a few different svg of pelicans. it is INSANE.

Re: Google releases Gemma 4 open models

#224

Post by jquery »

danielhanchen wrote: post #2

Thinking / reasoning + multimodal + tool calling. We made some quants at https://huggingface.co/collections/unsloth/gemma-4 for folks to run them - they work really well! Guide for those interested: https://unsloth.ai/docs/models/gemma-4 Also note to use temperature = 1.0, top_p = 0.95, top_k = 64 and the EOS is " ". " thought\n" is also used for the thinking trace!

Awesome!! Thank you SO much for this.

Re: Google releases Gemma 4 open models

#225

Post by misiti3780 »

vunderba wrote: post #32
originalvichy wrote: post #14

The wait is finally over. One or two iterations, and I’ll be happy to say that language models are more than fulfilling my most common needs when self-hosting. Thanks to the Gemma team!

Strongly agree. Gemma3:27b and Qwen3-vl:30b-a3b are among my favorite local LLMs and handle the vast majority of translation, classification, and categorization work that I throw at them.

what HW are you running them on ? are you using OLLAMA ?

Re: Google releases Gemma 4 open models

#226

Post by kpw94 »

GistNoesis wrote: post #140
bachmeier wrote: post #118

Earlier quoted context omitted.

So is there something I can take from that table if I have a 24 GB video card? I'm honestly not sure how to use those numbers.

I just tried with llama.cpp RTX4090 (24GB) GGUF unsloth quant UD_Q4_K_XL You can probably run them all. G4 31B runs at ~5tok/s , G4 26B A4B runs at ~150 tok/s. You can run Q3.5-35B-A3B at ~100 tok/s. I tried G4 26B A4B as a drop-in replacement of Q3.5-35B-A3B for some custom agents and G4 doesn't respect the prompt rules at all. (I added in the system prompt as described (but have not spend time checking if the reaso…

> I'll need to investigate further but it doesn't seem promising.

That's what I meant by "waiting a few days for updates" in my other comment. Qwen 3.5 release, I remember a lot of complaints about: "tool calling isn't working properly" etc.

That was fixed shortly after: there was some template parsing work in llama.cpp. and unsloth pulled out some models and brought back better one for improving something else I can't quite remember, better done Quantization or something...

coder543 pointed out the same is happening regarding tool calling with gemma4: https://news.ycombinator.com/item?id=47619261

Re: Google releases Gemma 4 open models

#228

Post by neonstatic »

Prompt:

> what is the Unix timestamp for this: 2026-04-01T16:00:00Z

Qwen 3.5-27b-dwq

> Thought for 8 minutes 34 seconds. 7074 tokens.

> The Unix timestamp for 2026-04-01T16:00:00Z is:

> 1775059200 (my comment: Wednesday, 1 April 2026 at 16:00:00)

Gemma-4-26b-a4b

> Thought for 33.81 seconds. 694 tokens.

> The Unix timestamp for 2026-04-01T16:00:00Z is:

> 1775060800 (my comment: Wednesday, 1 April 2026 at 16:26:40)

Gemma considered three options to solve this problem. From the thinking trace:

> Option A: Manual calculation (too error-prone).

> Option B: Use a programming language (Python/JavaScript).

> Option C: Knowledge of specific dates.

It then wrote a python script:

  from datetime import datetime, timezone
  date_str = "2026-04-01T16:00:00Z"
  # Replace Z with +00:00 for ISO format parsing or just strip it
  dt = datetime.strptime(date_str, "%Y-%m-%dT%H:%M:%SZ").replace(tzinfo=timezone.utc)
  ts = int(dt.timestamp())
  print(ts)
Then it verified the timestamp with a command:

  date -u -d @1775060800
All of this to produce a wrong result. Running the python script it produced gives the correct result. Running the verification date command leads to a runtime error (hallucinated syntax). On the other hand Qwen went straight to Option A and kept overthinking the question, verifying every step 10 times, experienced a mental breakdown, then finally returned the right answer. I think Gemma would be clearly superior here if it used the tools it came up with rather than hallucinating using them.

Re: Google releases Gemma 4 open models

#229

Post by knbknb »

canyon289 wrote: post #51

Hi all! I work on the Gemma team, one of many as this one was a bigger effort given it was a mainline release. Happy to answer whatever questions I can

Does "major number release" mean that it is actually an order of magnitude more compute effort that went into creating this model?

Or is this fundamentally a different model architecture, or a completely new tech stack on top of which this model was created (and the computing effort was actually less than before, in the v3 major relase?

Re: Google releases Gemma 4 open models

#230

Post by mixtureoftakes »

danielhanchen wrote: post #220
petu wrote: post #218

Earlier quoted context omitted.

Try 26B first. 31B seems to have very heavy KV cache (maybe bugged in llama.cpp at the moment; 16K takes up 4.9GB). edit: 31B cache is not bugged, there's static SWA cost of 3.6GB.. so IQ4_XS at 15.2GB seems like reasonable pair, but even then barely enough for 64K for 24GB VRAM. Maybe 8 bit KV quantization is fine now after https://github.com/ggml-org/llama.cpp/pull/21038 got merged, so 100K+ is possible. > I should…

Yes UD-Q4_K_XL works well! :)

what is the main difference between "normal" quants and the UD ones?
Post reply on HN