Mario Neto·Oct 13Local AI with Podman and Fedora 44: Mastering Ollama and Open-WebUI with an AMD Radeon 780M iGPUPara ler em PT-BR ->…
Martin St.·Sep 12I Got Tired of Guessing Which AMD Kernel Was Fastest — So I Made PyTorch Guess For Merocm-kernel-select: an open-source PyTorch extension that benchmarks your ops on first call and auto-picks the fastest ROCm backend, every…
Rabbit Knight·Aug 13Keeping the Model Local: GPU Passthrough and a Rootless vLLM Container on ProxmoxAnyone who reads my work knows I prefer hands-on experience, so with the sheer momentum behind AI right now, diving into local models was…
Krasimir Karamazov·Jun 19I Asked myself“Is llama-swap Worth It?”Bit of a disclaimer. I’m not running a pro grade rig. My stuff is consumer grade with a prosumer GPU. So the stats: 14600K, 32 GB DDR5…
Gichira·May 9Building FarmiPal: An AI-Powered Agricultural Assistant for East African Smallholder FarmersHow I combined vision models, agentic pipelines, and satellite climate data to help farmers predict crop surplus before it happens —…
Ryan·Mar 18What I Got Wrong Before I Got It Right: Reimplementing LoRA and Hopfield Attention from ScratchA 30-day sprint, a broken ablation, a Nobel Prize-winning algorithm, and what informal text teaches you about attention sharpness.
Ravi_Gupta·Jan 22, 2025First try at profiling HIP code on AMD MI300XI had a recent inspiration to try my hands MI300X and run some basic workloads to understand effect of Occupancy and Thread and…
duckdev·Jul 9, 2024[개발살집] Gemma 한국어 요약 모델 파인 튜닝 해보기LLM 파인튜닝 해보기. 하지만, 장렬히 실패했다.A response icon1A response icon1
![[개발살집] Gemma 한국어 요약 모델 파인 튜닝 해보기](https://miro.medium.com/v2/resize:fill:160:106/1*3SFp_ZJyfcBga4qJ2iFsJg.png)
![[개발살집] Gemma 한국어 요약 모델 파인 튜닝 해보기](https://miro.medium.com/v2/resize:fill:320:214/1*3SFp_ZJyfcBga4qJ2iFsJg.png)