align post training loss at the center (#473)

## Summary
<!--- This is a required section; please describe the main purpose of
this proposed code change. --->

<!---
## Details
This is an optional section; is there anything specific that reviewers
should be aware of?
--->

## Testing Done
<!--- This is a required section; please describe how this change was
tested. --->

<!-- 
Replace BLANK with your device type. For example, A100-80G-PCIe

Complete the following tasks before sending your PR, and replace `[ ]`
with
`[x]` to indicate you have done them. 
-->

- Hardware Type: <BLANK>
- [ ] run `make test` to ensure correctness
- [ ] run `make checkstyle` to ensure code style
- [ ] run `make test-convergence` to ensure convergence

README.md

@@ -92,7 +92,9 @@ With one line of code, Liger Kernel can increase throughput by more than 20% and
 
 ## Optimize post training with Liger Kernel
 
-![Post Training](https://raw.githubusercontent.com/linkedin/Liger-Kernel/main/docs/images/post-training.png)
+<p align="center">
+    <img src="https://raw.githubusercontent.com/linkedin/Liger-Kernel/main/docs/images/post-training.png" width="50%" alt="Post Training">
+</p>
 
 We provide optimized post training kernels like DPO, ORPO, SimPO, and more which can reduce memory usage by up to 80%. You can easily use them as python modules.