LiyuanLucasLiu commited on
Commit
f8b8aeb
1 Parent(s): 70f36dd

uploaded tech report and revised readme

Browse files
Files changed (3) hide show
  1. .gitattributes +1 -0
  2. GRIN_MoE.pdf +3 -0
  3. README.md +4 -6
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ *.pdf filter=lfs diff=lfs merge=lfs -text
GRIN_MoE.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39e878f28a2bdd362f0bbe0bc0fa2ef9b827551d74e9a617a18e2b3923abb322
3
+ size 1971199
README.md CHANGED
@@ -17,16 +17,14 @@ library_name: transformers
17
  <h1 align="center"> &#128513; MoE</h1>
18
  <h4 align="center">GRIN: <em>GR</em>adient-<em>IN</em>formed MoE</h4>
19
  <p align="center">
20
- <a href="https://huggingface.co/microsoft/GRIN-MoE">Hugging Face</a>&nbsp | &nbsp <a href="https://arxiv.org/abs/2304.08612"> Tech Report</a>&nbsp | &nbsp <a href="https://github.com/microsoft/GRIN-MoE/blob/main/LICENSE">License</a>&nbsp | &nbsp <a href="https://github.com/microsoft/GRIN-MoE">Github</a> &nbsp | &nbsp <a href="https://huggingface.co/microsoft/GRIN-MoE#usage">Get Started</a>&nbsp
21
  <br>
22
 
23
- GRIN MoE is a top2 16x3.8B MoE model.
24
- It achieves exceptionally good performance across a diverse set of tasks, particularly in coding and mathematics tasks.
25
- Comparing to conventional MoE training, GRIN MoE differs in mostly two ways:
26
 
27
- - GRIN uses SparseMixer-v2 to estimate the gradient related to expert routing, while the conventional MoE training treats expert gating as a proxy for the gradient estimation.
28
 
29
- - GRIN scales MoE training with neither expert parallelism nor token dropping, while the conventional MoE training employs expert parallelism and deploys token dropping.
30
 
31
  ## Intended Uses
32
 
 
17
  <h1 align="center"> &#128513; MoE</h1>
18
  <h4 align="center">GRIN: <em>GR</em>adient-<em>IN</em>formed MoE</h4>
19
  <p align="center">
20
+ <a href="https://huggingface.co/microsoft/GRIN-MoE">Hugging Face</a>&nbsp | &nbsp <a href="https://huggingface.co/microsoft/GRIN-MoE/blob/main/GRIN_MoE.pdf"> Tech Report</a>&nbsp | &nbsp <a href="https://huggingface.co/microsoft/GRIN-MoE/blob/main/LICENSE">License</a>&nbsp | &nbsp <a href="https://github.com/microsoft/GRIN-MoE">Github</a> &nbsp | &nbsp <a href="https://huggingface.co/microsoft/GRIN-MoE#usage">Get Started</a>&nbsp
21
  <br>
22
 
23
+ - With **only 6.6B** activate parameters, GRIN MoE achieves **exceptionally good** performance across a diverse set of tasks, particularly in coding and mathematics tasks.
 
 
24
 
25
+ - GRIN uses **SparseMixer-v2** to estimate the gradient related to expert routing, while the conventional MoE training treats expert gating as a proxy for the gradient estimation.
26
 
27
+ - GRIN scales MoE training with **neither expert parallelism nor token dropping**, while the conventional MoE training employs expert parallelism and deploys token dropping.
28
 
29
  ## Intended Uses
30