multi-gpu

Official implementation of CoCoFold2: scalable latent refinement of protein structure predictions from limited-particle cryo-EM data using a frozen Protenix-v1 diffusion prior. Supports single-GPU and component-parallel refinement.

Dynamic CUDA router for ExLlamaV2: auto-fallback from P40 (sm_61) to RTX 3050

Qwen3.8-27B on dual NVIDIA RTX 3060 12GB (llama.cpp). Recipe, A/B log, Localmaxxing drafts.

Reproducible LLM inference recipes: pinned runtimes, GPU layouts, patches, benchmarks, and measured tradeoffs.

Genuine multi-GPU FSDP full fine-tune of a 7B across 4x A100 (closes the scale gap), with the tied-embeddings, collective-save, and checkpoint-consolidation gotchas made concrete.

word2vec implementation using PyTroch

This project aims to help people implement tensorflow model pipelines quickly for different nlp tasks.

Self-hosted private AI workspace on owned Ubuntu hardware. llama.cpp layer-splits Qwen3.8-27B across an RTX 3060 and RTX 3070; a FastAPI app adds login-gated streaming chat, SearXNG search, page fetch, weather, Hugging Face and optional document retrieval. No LAN or public listener - remote access via Cloudflare Tunnel. AGPL-3.0.

Testing speed of two GPU's vs. one GPU

Simulated Multi-GPU inference engine implementing Megatron-style Tensor Parallelism, GPipe Pipeline Parallelism, and KV Cache Sharding from scratch. Features bit-identical fidelity validation on Qwen2-0.5B weights and analytical communication cost modeling.

performance test of MNIST hand writings usign MXNet + TF

Quality-first local MiniMax H3 video-series generation and loopback API for dual RTX 4090 workstations, with native audio, references, P8/P9 continuity, and preserved artifacts.

Lightweight terminal launcher and auto-optimizer for llm models using llama.cpp with hardware detection, tensor sharding, benchmarks, presets, context tuning, and OpenAI-compatible serving.

Experiments from 'NLP with Transformers' (fine-tuning, summarization, knowledge distillation) as standalone scripts, plus multi-GPU training with Hugging Face Accelerate: DDP gives ~1.8x the throughput of DataParallel on 2x RTX 4060 Ti

Layer split vs tensor parallelism in llama.cpp - measured on two platforms 16x apart in interconnect bandwidth

Experimental Keras libraries and examples.

PyTorch compiler and Rust runtime for multi-CPU/GPU inference, NUMA-aware scheduling, model serving, parameter streaming, and activation spill.

Reproducible AI, ML, and scientific computing template with Pixi, PyTorch Lightning, Aim tracking, and CPU/GPU training.

Portable multi-GPU music generation server for Windows -- 8 models, one-click install, zero dependencies

Multi-GPU video upscaling with frame-level parallelism