Update README.md
Browse files
README.md
CHANGED
|
@@ -1,4 +1,7 @@
|
|
| 1 |
This is a version of [ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) distilled down to 16 layers out of 22.
|
|
|
|
|
|
|
|
|
|
| 2 |
The last 6 local attention layers were removed:
|
| 3 |
|
| 4 |
0. Global
|
|
|
|
| 1 |
This is a version of [ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) distilled down to 16 layers out of 22.
|
| 2 |
+
This reduces the number of parameters from 149M to 119M; however, practically speaking, since the embedding params
|
| 3 |
+
do not contribute greatly to latency, the effect is reducing the "trunk" of the model from 110M params to 80M params.
|
| 4 |
+
I would expect this to reduce latency by roughly 25% (increasing throughput by roughly 33%).
|
| 5 |
The last 6 local attention layers were removed:
|
| 6 |
|
| 7 |
0. Global
|