diff --git a/README.md b/README.md
index 1eea802..3cc8194 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,5 @@
 ## News Board
+>_2022/05/03_ New features experimented, details introduced in [`notebooks`](https://github.com/qiaochen/VeloAE/tree/main/notebooks)
 
 >_2022/04/29_ VeloAE updated to 0.2.0. To simplify the project, the folder `notebooks` is reorganized and only notebooks involving veloAE experiments are kept, scvelo dynamical mode models are additionally included for comparison. The previous data are backuped in the branch [`paper-version-backup`](https://github.com/qiaochen/VeloAE/tree/paper-version-backup)
 
diff --git a/notebooks/readme.md b/notebooks/readme.md
index 531131c..8c8382e 100644
--- a/notebooks/readme.md
+++ b/notebooks/readme.md
@@ -1,22 +1,18 @@
-## New feartures experimented:
-New Feature: Transfer knowledge from a given velocity estimates in the high-dimentional space into latent space.
+## New feartures experimented
+New Feature: Transfer knowledge from a given velocity estimation in the raw space into latent space.
 
-In notebooks `retina` and `oligodendrocyte`, we experimented a new feature to enhance veloAE with signals from a given velocity estimation, e.g., a velocity matrix inferred with scVelo stochastic mode in the high-dimensional space.
+In notebooks `retina` and `oligodendrocyte`, we experimented a new feature that enhances veloAE with signals from a given velocity estimation, e.g., a velocity matrix inferred by scVelo stochastic mode in the raw space.
 
 Briefly, now we can have two versions of velocities in the low-dimensional space of veloAE:
 
-1. a latent velocity of steady-state estimation using projected Spliced and Unspliced reads in the low-dimensional space;
-2. a projected velocity by passing the given velocity from the raw space through the encoder of veloAE;
+1. a latent velocity of steady-state estimation using projected Spliced and Unspliced reads in the low-dimensional space
+2. a projected velocity by passing the given velocity from the raw space through the encoder of veloAE
 
-During training, we may choose to add an auxiliary loss/constraint to make the two versions of velocites to be close, hence transfering the knowledge from, e.g., scvelo to veloAE.
+During training, we may optionally choose to add an auxiliary loss/constraint to make the two versions of velocites to be close, hence transfering the knowledge from, e.g., scvelo to veloAE.
 
 In
-- `retina`, we apply this auxiliary loss to all cell types, making veloAE function as denoising the output of scvelo stochastic mode.
-- `Oligodendrocyte`, we apply this auxiliary loss to only the `NFOLs`  cell type, encouraging veloAE to retain the velocity for the designated cell type. This new feature provides an interface for injecting only part of the knowledge (e.g., velocites of cell clusters we are more confident) from previous estimates that we are more certain about.
-
-
-
-
+- `retina`, we applied this auxiliary loss to all cell types, making veloAE function as denoising the output of scvelo estimate.
+- `Oligodendrocyte`, we applied this auxiliary loss to only the `NFOLs`  cell type, encouraging veloAE to retain the velocity for the designated cell type. This new feature provides an interface for injecting only part of the knowledge (e.g., given velocites of cell clusters we are more confident about) from the given  estimation.
 
 
 ## Retrieving the Datasets