EleutherAI
/
gpt-j-6B
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

This commit is contained in:
Eric Hallahan 2021-08-06 03:35:15 +00:00 committed by huggingface-web
parent d3d2956061
commit 8a44d5a1b4
1 changed files with 12 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
README.md
View File

@ -18,19 +18,20 @@ GPT-J 6B is a transformer model trained using Ben Wang's [Mesh Transformer JAX](
<figure>
| Hyperparameter | Value |
|----------------------|---------------|
| \\(n_{parameters}\\) | 6,053,381,344 |
| \\(n_{layers}\\) | 28&ast; |
| \\(d_{model}\\) | 4,096 |
| \\(d_{ff}\\) | 16,384 |
| \\(n_{heads}\\) | 16 |
| \\(d_{head}\\) | 256 |
| \\(n_{ctx}\\) | 2,048 |
| \\(n_{vocab}\\) | 50,257 (same tokenizer as GPT-2/3) |
| Hyperparameter | Value |
|----------------------|------------|
| \\(n_{parameters}\\) | 6053381344 |
| \\(n_{layers}\\) | 28&ast; |
| \\(d_{model}\\) | 4096 |
| \\(d_{ff}\\) | 16384 |
| \\(n_{heads}\\) | 16 |
| \\(d_{head}\\) | 256 |
| \\(n_{ctx}\\) | 2048 |
| \\(n_{vocab}\\) | 50257/50400&dagger; (same tokenizer as GPT-2/3) |
| Positional Encoding | [Rotary Position Embedding (RoPE)](https://arxiv.org/abs/2104.09864) |
| RoPE Dimensions | [64](https://github.com/kingoflolz/mesh-transformer-jax/blob/f2aa66e0925de6593dcbb70e72399b97b4130482/mesh_transformer/layers.py#L223) |
<figcaption><strong>&ast;</strong> Each layer consists of one feedforward block and one self attention block.</figcaption></figure>
<figcaption><p><strong>&ast;</strong> Each layer consists of one feedforward block and one self attention block.</p>
<p><strong>&dagger;</strong> Although the embedding matrix has a size of 50400, only 50257 entries are used by the GPT-2 tokenizer.</p></figcaption></figure>
The model consists of 28 layers with a model dimension of 4096, and a feedforward dimension of 16384. The model
dimension is split into 16 heads, each with a dimension of 256. Rotary Position Embedding (RoPE) is applied to 64