foundation-model-stack / fms-accelerationLinks
π Collection of libraries used with fms-hf-tuning to accelerate fine-tuning and training of large models.
β11Updated 2 months ago
Alternatives and similar repositories for fms-acceleration
Users that are interested in fms-acceleration are comparing it to the libraries listed below
Sorting:
- Experimental scripts for researching data adaptive learning rate scheduling.β23Updated last year
- Intel Gaudi's Megatron DeepSpeed Large Language Models for trainingβ13Updated 8 months ago
- Source-to-Source Debuggable Derivatives in Pure Pythonβ15Updated last year
- Utilities for Training Very Large Modelsβ58Updated 11 months ago
- Train, tune, and infer Bamba modelβ131Updated 2 months ago
- β34Updated 11 months ago
- JAX Scalify: end-to-end scaled arithmeticsβ16Updated 10 months ago
- β46Updated last year
- Using FlexAttention to compute attention with different masking patternsβ44Updated 11 months ago
- β21Updated 6 months ago
- Implementation of Hyena Hierarchy in JAXβ10Updated 2 years ago
- β23Updated 8 months ago
- β26Updated last year
- Demo of the unit_scaling library, showing how a model can be easily adapted to train in FP8.β46Updated last year
- Official code for the paper "Attention as a Hypernetwork"β40Updated last year
- FlexAttention w/ FlashAttention3 Supportβ27Updated 10 months ago
- A collection of reproducible inference engine benchmarksβ32Updated 4 months ago
- Fork of Flame repo for training of some new stuff in developmentβ17Updated last week
- vLLM adapter for a TGIS-compatible gRPC server.β38Updated this week
- The evaluation framework for training-free sparse attention in LLMsβ91Updated 2 months ago
- Experiment of using Tangent to autodiff tritonβ80Updated last year
- The source code of our work "Prepacking: A Simple Method for Fast Prefilling and Increased Throughput in Large Language Models" [AISTATS β¦β61Updated 10 months ago
- This is a new metric that can be used to evaluate faithfulness of text generated by LLMs. The work behind this repository can be found heβ¦β31Updated 2 years ago
- A scalable implementation of diffusion and flow-matching with XGBoost models, applied to calorimeter data.β18Updated 10 months ago
- DPO, but faster πβ44Updated 8 months ago
- β13Updated 2 weeks ago
- PyTorch implementation of the Flash Spectral Transform Unit.β17Updated 11 months ago
- A parallel framework for training deep neural networksβ63Updated 5 months ago
- π· Build compute kernelsβ119Updated last week
- Repository for Sparse Finetuning of LLMs via modified version of the MosaicML llmfoundryβ42Updated last year