DEV Community

# pytorch

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
SO-HMS: A Universal Optimization Framework for Complex Multi-Objective Systems

SO-HMS: A Universal Optimization Framework for Complex Multi-Objective Systems

1
Comments
1 min read
PyTorch from Scratch — Part 1: Tensors, Gradients & Activations

PyTorch from Scratch — Part 1: Tensors, Gradients & Activations

1
Comments
5 min read
AI Coding Tools for Machine Learning Engineers in 2026: Jupyter, PyTorch, and the CUDA Trap

AI Coding Tools for Machine Learning Engineers in 2026: Jupyter, PyTorch, and the CUDA Trap

Comments
5 min read
Carbon-Aware Model Training: Scheduling GPU Workloads Around Electricity Carbon Intensity

Carbon-Aware Model Training: Scheduling GPU Workloads Around Electricity Carbon Intensity

6
Comments
7 min read
Why JAX Is a Much Better Backend for Quantum Circuit Simulation Than PyTorch

Why JAX Is a Much Better Backend for Quantum Circuit Simulation Than PyTorch

Comments
3 min read
Our event-camera detector lost 6 mAP to a badly chosen accumulation window

Our event-camera detector lost 6 mAP to a badly chosen accumulation window

Comments
4 min read
From Bayesian to deep knowledge tracing — upgrading NumPath's student model with a PyTorch LSTM

From Bayesian to deep knowledge tracing — upgrading NumPath's student model with a PyTorch LSTM

Comments
5 min read
Tracing torch.cuda.empty_cache() on an RTX 4090 - Where Do the 53 MB Go?

Tracing torch.cuda.empty_cache() on an RTX 4090 - Where Do the 53 MB Go?

Comments
5 min read
QAT vs PTQ on our edge vision model: 6 months of A/B data

QAT vs PTQ on our edge vision model: 6 months of A/B data

Comments
4 min read
Structured channel pruning got our detector under 12ms on a Jetson

Structured channel pruning got our detector under 12ms on a Jetson

Comments
4 min read
Serving 40 LoRA adapters on one base model: the throughput we got

Serving 40 LoRA adapters on one base model: the throughput we got

Comments
4 min read
torch.compile recompiled our SDXL UNet 38 times in production

torch.compile recompiled our SDXL UNet 38 times in production

Comments
4 min read
LLM-as-judge variance broke our DPO training signal for 3 weeks

LLM-as-judge variance broke our DPO training signal for 3 weeks

Comments
4 min read
The bf16 grad accumulator that killed our SDXL LoRA training

The bf16 grad accumulator that killed our SDXL LoRA training

Comments
4 min read
Why Your PyTorch Training Crawls on a Beefy GPU (And How to Fix It)

Why Your PyTorch Training Crawls on a Beefy GPU (And How to Fix It)

Comments
5 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.