Import HTML embeds
Browse files
app/src/content/article.mdx
CHANGED
|
@@ -48,6 +48,8 @@ tableOfContentsAutoCollapse: true
|
|
| 48 |
pdfProOnly: false
|
| 49 |
---
|
| 50 |
|
|
|
|
|
|
|
| 51 |
|
| 52 |
On-policy distillation is a highly effective strategy for compressing LLMs, as recently highlighted by [Thinking Machines' excellent blog post.](https://thinkingmachines.ai/blog/on-policy-distillation/) The technique trains a small "student" model by transferring knowledge from a high-performing "teacher" model's probability distribution. This allows the student to emulate the teacher's task performance, while significantly reducing size and latency.
|
| 53 |
|
|
|
|
| 48 |
pdfProOnly: false
|
| 49 |
---
|
| 50 |
|
| 51 |
+
import HtmlEmbed from '../components/HtmlEmbed.astro'
|
| 52 |
+
|
| 53 |
|
| 54 |
On-policy distillation is a highly effective strategy for compressing LLMs, as recently highlighted by [Thinking Machines' excellent blog post.](https://thinkingmachines.ai/blog/on-policy-distillation/) The technique trains a small "student" model by transferring knowledge from a high-performing "teacher" model's probability distribution. This allows the student to emulate the teacher's task performance, while significantly reducing size and latency.
|
| 55 |
|