WaveSpeedAI / QuantumAttention
[WIP] Better (FP8) attention for Hopper
☆27Updated last month
Alternatives and similar repositories for QuantumAttention:
Users that are interested in QuantumAttention are comparing it to the libraries listed below
- (WIP) Parallel inference for black-forest-labs' FLUX model.☆18Updated 4 months ago
- FlexAttention w/ FlashAttention3 Support☆26Updated 6 months ago
- A CUDA kernel for NHWC GroupNorm for PyTorch☆18Updated 5 months ago
- Writing FLUX in Triton☆32Updated 6 months ago
- ☆65Updated 3 months ago
- PyTorch half precision gemm lib w/ fused optional bias + optional relu/gelu☆61Updated 4 months ago
- An auxiliary project analysis of the characteristics of KV in DiT Attention.☆29Updated 4 months ago
- Patch convolution to avoid large GPU memory usage of Conv2D☆86Updated 2 months ago
- Triton kernels for Flux☆20Updated 3 months ago
- Tritonbench is a collection of PyTorch custom operators with example inputs to measure their performance.☆110Updated this week
- ☆68Updated 2 months ago
- Quantized Attention on GPU☆45Updated 4 months ago
- A Suite for Parallel Inference of Diffusion Transformers (DiTs) on multi-GPU Clusters☆44Updated 8 months ago
- A parallelism VAE avoids OOM for high resolution image generation☆59Updated 2 months ago
- APPy (Annotated Parallelism for Python) enables users to annotate loops and tensor expressions in Python with compiler directives akin to…☆23Updated last week
- ☆48Updated last year
- Odysseus: Playground of LLM Sequence Parallelism☆68Updated 9 months ago
- ☆16Updated last year
- ☆156Updated 3 months ago
- Work in progress.☆56Updated last week
- QuIP quantization☆52Updated last year
- Official repository for VQDM:Accurate Compression of Text-to-Image Diffusion Models via Vector Quantization paper☆33Updated 6 months ago
- ☆39Updated this week
- Code for data-aware compression of DeepSeek models☆18Updated last week
- Load compute kernels from the Hub☆115Updated this week
- ☆28Updated 2 months ago
- ☆68Updated 3 weeks ago
- Repository for Sparse Finetuning of LLMs via modified version of the MosaicML llmfoundry☆40Updated last year
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆112Updated 4 months ago
- Framework to reduce autotune overhead to zero for well known deployments.☆63Updated this week