Efficient Triton Kernels for LLM Training
-
Updated
Oct 27, 2026 - Python
{{ message }}
#
triton-kernels
Here are 36 public repositories matching this topic...
FlagGems is an operator library for large language models implemented in the Triton Language.
-
Updated
Oct 29, 2026 - Python
LLM notes, including model inference, transformer model structure, and llm framework code analysis notes.
-
Updated
Aug 19, 2026 - Python
Puzzles for learning Triton, play it with minimal environment configuration!
-
Updated
Mar 17, 2026 - Python
RapidLLM is an LLM inference framework with continuous batching, tensor and data parallelism, quantized weights, and with pluggable Triton/CUDA kernels.
python3 attention continues llm llm-inference llama3 continuous-batching flash-attention-3 triton-kernels qwen3-moe qwen3-vl tensor-parallel
-
Updated
Oct 10, 2026 - Python
Tiled Flash Linear Attention library for fast and efficient mLSTM Kernels.
-
Updated
Sep 14, 2026 - Jupyter Notebook
Production inference for encoder models - ColBERT, GLiNER, ColPali, embeddings etc. - as vLLM plugins for online and in-process deployment
retrieval encoder inference embeddings ner serving rag colbert ml-infra-deployments vllm gliner multimodal-rag ai-infrastructure colpali triton-kernels vllm-plugins
-
Updated
Jul 6, 2026 - Python
Open-source inference engine that runs language, vision, speech, image and video models from one container format, with a PyTorch and a Triton compute branch. Released for NVIDIA GPUs and CPU; Apple Silicon in progress.
text-to-speech ai deep-learning gpu cuda pytorch triton speech-to-text super-resolution text-to-image inference-engine aten multimodal image-to-video mixture-of-experts diffusion-models text-to-video llm local-inference triton-kernels
-
Updated
Oct 29, 2026 - Python
A "standard library" of Triton kernels.
-
Updated
Oct 2, 2025 - Python
High-performance late-interaction retrieval engine for on-prem AI. ColBERT/ColPali multi-vector search with Rust fused MaxSim, Triton GPU kernels, ROQ quantization, LEMUR routing, WAL-backed CRUD, and a FastAPI server — single machine, CPU or GPU.
semantic-search-engine on-premise colbert-ai multivector quantizations retrieval-augmented-generation vector-databases rag-pipeline multimodal-rag colpali triton-kernels late-interaction multivector-search multivector-embeddings
-
Updated
Jul 6, 2026 - Python
Manifold-Constrained Hyper-Connections with fused Triton kernels for efficient training
-
Updated
Mar 10, 2026 - Python
Educational resource demonstrating common GPU programming pitfalls and solutions using Triton kernels.
-
Updated
Feb 26, 2026 - Jupyter Notebook
Fast Cartesian tensor operations and equivariant interatomic potentials for atomistic simulations
molecular-dynamics atomistic-simulations interatomic-potentials equivariant-neural-networks triton-kernels cartesian-tensors
-
Updated
Oct 6, 2026 - Python
A from-scratch nano-Triton and nano-Helion: the modern GPU-kernel DSL stack, rebuilt in ~4,000 lines
-
Updated
Aug 12, 2026 - Python
Official Code for the paper ELMO : Efficiency via Low-precision and Peak Memory Optimization in Large Output Spaces (in ICML 2025)
chunking multi-label-classification extreme-classification low-precision-training large-output-space float8 triton-kernels gradient-fusion
-
Updated
Feb 27, 2026 - Python
KernelHeim – development ground of custom Triton and CUDA kernel functions designed to optimize and accelerate machine learning workloads on NVIDIA GPUs. Inspired by the mythical stronghold of the gods, KernelHeim is a forge where high-performance kernels are crafted to unlock the full potential of the hardware.
-
Updated
May 5, 2026 - Python
Triton Kernel Tutorials for low level LLM optimization
deep-learning optimization deep-reinforcement-learning inference triton inference-optimization llm triton-lang triton-kernels
-
Updated
Oct 13, 2026 - Python
Constraint-aware evolutionary optimization for Triton kernels on Ascend NPU — national third prize
-
Updated
Sep 1, 2026 - Python
LLM agents that generate, verify, and evolve Triton GPU kernels. Includes a reward-hack-resistant benchmarking harness with strict correctness verification and fresh-input evaluation. Achieves up to 174.7× over PyTorch eager and outperforms FlexAttention (1.48×) and SDPA (1.17×) on selected workloads.
inference openai evolutionary-algorithms automatic-code-generation large-language-models agentic-workflow agentic-ai triton-kernels kernel-optimization gepa harness-engineering agentic-kernel-generation
-
Updated
Aug 12, 2026 - Python
A container of various PyTorch neural network modules written in Triton.
-
Updated
Mar 31, 2026 - Python
Add this topic to your repo
To associate your repository with the triton-kernels topic, visit your repo's landing page and select "manage topics."