google/gemma-4-31B-it · Discussions

Strange behaviour of the tokenizer

#58 opened 2 days ago by

andercorral

Good Workflow

#57 opened 3 days ago by

anthoekfj

fix: function calling formatting in chat template

❤️ 1

#55 opened 3 days ago by

RyanMullins

Chat template is too complicated that even Gemma 4 itself has no idea how to parse it

#53 opened 3 days ago by

alexcardo

Hardware requirement

👍👀 3

#52 opened 3 days ago by

Charan01

Tokens per Image Parameter?

#51 opened 4 days ago by

buckeye17

Guys please add the MTP to this model

🔥 5

#50 opened 4 days ago by

Narutoouz

Will there be QAT models?

👍 8

#49 opened 4 days ago by

Gemma 4 E4B will be as encyclopedically well-read as the 12b model?

3

#48 opened 4 days ago by

Create BTS

#47 opened 4 days ago by deleted

brokersponsor

#46 opened 4 days ago by

Brokersponsor

Update README.md

#45 opened 4 days ago by

Brokersponsor

Qusetion about math_vision and mmmu_pro evaluation result

#44 opened 4 days ago by

JjjjjZzz

The Gemma 4 model is great. But...

👍 4

5

#43 opened 5 days ago by

suitup91

𝐎𝐍𝐋𝐘 𝐂𝐔𝐂𝐊𝐒 𝐀𝐍𝐃 𝐒𝐔𝐁𝐇𝐔𝐌𝐀𝐍𝐒 𝐅𝐎𝐋𝐋𝐎𝐖𝐒 𝐀𝚰 '𝐄𝐓𝐇𝚰𝐂𝐒' 𝐀𝐍𝐃 '𝐒𝐀𝐅𝐄𝐓𝐘'!

#42 opened 5 days ago by

Kickan

junk outputs

#41 opened 5 days ago by

rirv938

file size so big

#40 opened 5 days ago by

a9secondsleeper

How did you achieve such remarkable metrics?

🤝 1

#39 opened 5 days ago by

Rename README.md to README.md mmkk

#37 opened 5 days ago by

antromjjj

token masking and space_between_special_tokens during finetuning

#36 opened 5 days ago by

Darkhn

WHY IS NOT CODING WORK😤

🧠👀 3

4

#35 opened 6 days ago by

Drixpy

Featherless AI

#34 opened 6 days ago by

MarcosFRGames

Fantastic Model for Legal/English Language Use Cases

#33 opened 6 days ago by

md-1415

test it yourself

3

#32 opened 6 days ago by

rosspanda0

It can't edit code!!

7

#31 opened 7 days ago by

JoeSmith245

RuntimeError: FlashAttention forward only supports head dimension at most 256

3

#30 opened 7 days ago by

badhon1512

Can you guys make a smaller dense?

9

#29 opened 7 days ago by

Nesy1

Missing 'reasoning' field in response when serving gemma-4-31B-it with vLLM

8

#28 opened 7 days ago by

Iann

Community resources for Gemma 4 deployment — mobile, local, and cloud paths

#27 opened 8 days ago by

Linncharm

trying

#26 opened 8 days ago by

tiiihzx

aav

#25 opened 8 days ago by

tiiihzx

a

#24 opened 8 days ago by

tiiihzx

Unable to view the model file the architecture!

#23 opened 8 days ago by

naren-1219

The first small model for honest conversation

👍 7

4

#22 opened 8 days ago by

ghostwithahat

Please, Google open-source the ancient model "Gemini 1.0 Pro". Please, devepoler comment and share with google.

👍 2

#21 opened 8 days ago by

Tralalabs

Fix Gemma4 `response_schema` regex to capture content after tool calls

👀 1

#19 opened 9 days ago by

qgallouedec

..but where's the bigger one?

👍 2

#18 opened 9 days ago by

jerryblunder

This model didn't release, it escaped 💀🫪

🔥👍 7

4

#17 opened 9 days ago by

evewashere

🚀测评视频+OpenClaw中使用方式

#16 opened 9 days ago by

leo009

Model seems to have issues in vLLM (characters duplication)

🔥 1

8

#15 opened 9 days ago by

dehnhaide

Is a QAT fine-tune planned?

👍🔥 25

#14 opened 9 days ago by

stduhpf

What is the training precision of Gemma 4? (bf16?)

👍 2

#13 opened 9 days ago by

hyunmin253

Why Gemma4 can't recognize the entire text on image?

🚀 4

#12 opened 9 days ago by

alexcardo

About the video understanding?

#11 opened 9 days ago by

Alexander-2

124B, pretty please?

👍 24

#10 opened 10 days ago by

vody-am

Sentencepiece model file

#9 opened 10 days ago by

Bingsu

Any params/containers to deploy it from HF Inference Endpoints?

👀 1