Hackernews posts about Gemma 3 270

1. 01
   

   newGEMM Tutorial and UlmBLAS

   ▲ 5michael-lehn7d💬 2savesaved
2. 02
   

   newGemma 3 270M: Compact model for hyper-efficient AIdevelopers.googleblog.com

   ▲ 824meetpateltech449d💬 317savesaved
3. 03
   

   newGemma 3 270M re-implemented in pure PyTorch for local tinkeringgithub.com

   ▲ 417ModelForge443d💬 57savesaved
4. 04
   

   newOwn your AI: Learn how to fine-tune Gemma 3 270M and run it on-devicedevelopers.googleblog.com

   ▲ 5simonpure392ddiscusssavesaved
5. 05
   

   newGemma 3-270Mhuggingface.co

   ▲ 4georgehill449ddiscusssavesaved
6. 06
   

   newYou can fine-tune Gemma3-270M and prepare for secure deployment within minuteswww.youtube.com

   ▲ 3astro_09439ddiscusssavesaved
7. 07
   

   newGemma 3 270M – Google's NEW AI – How to Fin-tune Gemma3 [video]www.youtube.com

   ▲ 2rjn32s448d💬 1savesaved
8. 08
   

   newTiny gemma3 270M implementation in Rgithub.com

   ▲ 2dfalbel438ddiscusssavesaved
9. 09
   

   newGemma 3 270Mtwitter.com

   ▲ 2tosh449ddiscusssavesaved
10. 10
    

    newLaunch HN: Cactus (YC S25) – AI inference on smartphonesgithub.com

    ▲ 123HenryNdubuaku414d💬 63savesaved
11. 11
    

    newShow HN: AfriTales – Discover the Magic of African Storytellingafritales.org

    ▲ 42ggap431d💬 26savesaved
12. 12
    

    newAsk HN: Any Local LLM can I run without GPU for Local Agentic workflow AI?

    ▲ 8limondas149d💬 4savesaved
13. 13
    

    newShow HN: Nabu (TTS Reader and LLM Playground on Android)github.com

    ▲ 7mewmix449ddiscusssavesaved
14. 14
    

    newRunning a 270M LLM on Android (architecture and benchmarks)

    ▲ 2ayushranjan99349ddiscusssavesaved
15. 15
    

    newShow HN: Go LLM inference with a Vulkan GPU back end that beats Ollama's CUDAgithub.com

    ▲ 1computerex244ddiscusssavesaved
16. 16
    

    newGemma 3 QAT Modelssimonwillison.net

    ▲ 4mdp2021566d💬 3savesaved
17. 17
    

    newGoogle's Gemma 3 is an open source, multimodal single-GPU AI 128K context windowarstechnica.com

    ▲ 4danboarder604d💬 1savesaved
18. 18
    

    newShow HN: Interactive semantic map and analysis of Hacker News stories in 2025lincolnmaxwell.com

    ▲ 2youngermax289ddiscusssavesaved
19. 19
    

    newFine-tuning Gemma 3 for mobileopensource.googleblog.com

    ▲ 1nkko329ddiscusssavesaved
20. 20
    

    newShow HN: Local LLM Notepad – run a GPT-style model from a USB stickgithub.com

    ▲ 40davidye324494d💬 9savesaved
21. 21
    

    newShow HN: Fine-tuned Llama 3.2 3B to match 70B models for local transcriptsbilawal.net

    ▲ 31phantompeace431d💬 8savesaved
22. 22
    

    newShow HN: Tested 12 LLMs with few-shot examples

    ▲ 2shuntaro-okuma225ddiscusssavesaved
23. 23
    

    newAsk HN: How do you use Local LLMs? (April 2026)

    ▲ 3Hixon10202ddiscusssavesaved
24. 24
    

    newI've tested some local LLMs on prosumer hardware, here are some findings

    ▲ 7felineflock75ddiscusssavesaved
25. 25
    

    newTq-KV – Rust implementation of TurboQuant that works on GGUF models

    ▲ 3onurgokyildiz219ddiscusssavesaved
26. 26
    

    newShow HN: Cactus Hybrid: We taught Gemma 4 to know when it's wronggithub.com

    ▲ 191HenryNdubuaku107d💬 44savesaved
27. 27
    

    newRun Gemma 4 locally with OpenClaw in 3 stepstwitter.com

    ▲ 2xnx213ddiscusssavesaved
28. 28
    

    newShow HN: Glq LLM quantization using E8 latticegithub.com

    ▲ 2acd158ddiscusssavesaved
29. 29
    

    newShow HN: Gemma 4 26B A4B running on an iPhone 17 Pro via model pagingnoemaai.com

    ▲ 3armin976102d💬 2savesaved
30. 30
    

    newShow HN: Don't trust me bro: fixing GPT-OSS (3.49B tokens, 1k GPU hours, 1x3090)github.com

    ▲ 2iamskeole66ddiscusssavesaved
31. 31
    

    newNew Inference Server for DGX Spark: large model C4:55-90 tok/s no spec decode

    ▲ 4medicis123107ddiscusssavesaved

· edit