HazyResearch / m2

Repo for "Monarch Mixer: A Simple Sub-Quadratic GEMM-Based Architecture"

HazyResearch/m2 Issues

training data
Closed a month ago10
Will M2-GPT be open-sourced?
Updated 2 months ago9
Using (Absolute) Positional Embeddings with Hyena Operators
Updated 2 months ago4
torch.bmm kernel fusion and precision casting issue in BlockdiagButterflyMultiply
Updated 2 months ago
LoCo Benchmark - BM25 & Insights
Updated 3 months ago7
About the versions of PyTorch, CUDA, and other dependencies used in the implementation of the Monarch Mixer
Updated 4 months ago20
Some minor inconsistency in the paper
Closed 4 months ago2
A question on square matrices
Updated 4 months ago1
Multilingual?
Updated 4 months ago1
Cannot use autoresume and wandb together
Updated 4 months ago1
M2 model is applied on single image deraining model based on transformer
Updated 5 months ago5
does M2 work with ONNX?
Updated 5 months ago3
Why is there such a big difference in cosine similarity between embeddings of the same pair when using padding=max_length versus padding=true?
Updated 5 months ago2
precision on imagenet experiment
Updated 5 months ago1
Missing Licence
Closed 5 months ago1
MNLI yaml config
Closed 6 months ago
can i use MonarchMixer replace cross attention lay
Updated 7 months ago4
What should I do if the √N is not an integer?
Closed 7 months ago3
Is there any detailed explanation of M2-VIT?
Updated 7 months ago
MonarchMixerLayer
Updated 8 months ago1
Code for model architecture
Updated 8 months ago
Bert-like implementation
Closed 8 months ago4
Code for projecting pre-trained BERT weights into Monarch matrices
Updated 9 months ago2
`python3 test_flash_mm.py` got error
Closed a year ago5