drisspg / transformer_nuggetsLinks
A place to store reusable transformer components of my own creation or found on the interwebs
☆60Updated last week
Alternatives and similar repositories for transformer_nuggets
Users that are interested in transformer_nuggets are comparing it to the libraries listed below
Sorting:
- Experiment of using Tangent to autodiff triton☆81Updated last year
- ☆89Updated last year
- Make triton easier☆47Updated last year
- Triton Implementation of HyperAttention Algorithm☆48Updated last year
- Demo of the unit_scaling library, showing how a model can be easily adapted to train in FP8.☆46Updated last year
- ☆21Updated 7 months ago
- CUDA and Triton implementations of Flash Attention with SoftmaxN.☆73Updated last year
- This code repository contains the code used for my "Optimizing Memory Usage for Training LLMs and Vision Transformers in PyTorch" blog po…☆92Updated 2 years ago
- Various transformers for FSDP research☆38Updated 2 years ago
- ☆122Updated last year
- Utilities for Training Very Large Models☆58Updated last year
- Easily run PyTorch on multiple GPUs & machines☆47Updated 3 months ago
- DPO, but faster 🚀☆44Updated 9 months ago
- PyTorch centric eager mode debugger☆48Updated 9 months ago
- Using FlexAttention to compute attention with different masking patterns☆44Updated last year
- An implementation of the Llama architecture, to instruct and delight☆21Updated 4 months ago
- ☆15Updated 4 months ago
- CUDA implementation of autoregressive linear attention, with all the latest research findings☆44Updated 2 years ago
- HomebrewNLP in JAX flavour for maintable TPU-Training☆50Updated last year
- Automatically take good care of your preemptible TPUs☆36Updated 2 years ago
- ☆21Updated last year
- See https://github.com/cuda-mode/triton-index/ instead!☆11Updated last year
- ☆20Updated 2 years ago
- ☆34Updated last year
- FlexAttention w/ FlashAttention3 Support☆27Updated last year
- ☆49Updated last year
- Repository for Sparse Finetuning of LLMs via modified version of the MosaicML llmfoundry☆42Updated last year
- Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT training☆132Updated last year
- The source code of our work "Prepacking: A Simple Method for Fast Prefilling and Increased Throughput in Large Language Models" [AISTATS …☆60Updated 11 months ago
- Context Manager to profile the forward and backward times of PyTorch's nn.Module