Add safetensors weights and config
Browse files- README.md +2 -0
- config.json +8 -0
- model.safetensors +3 -0
README.md
CHANGED
|
@@ -46,6 +46,8 @@ Each block uses pre-layer-norm, causal self-attention, and an MLP
|
|
| 46 |
| --- | --- |
|
| 47 |
| `model.pt` | Trained weights: a dict with `state_dict`, `chars` (the vocabulary) and `config` |
|
| 48 |
| `model.py` | The model code in PyTorch, about 60 lines |
|
|
|
|
|
|
|
| 49 |
|
| 50 |
## How to use
|
| 51 |
|
|
|
|
| 46 |
| --- | --- |
|
| 47 |
| `model.pt` | Trained weights: a dict with `state_dict`, `chars` (the vocabulary) and `config` |
|
| 48 |
| `model.py` | The model code in PyTorch, about 60 lines |
|
| 49 |
+
| `model.safetensors` | The same weights in safetensors format |
|
| 50 |
+
| `config.json` | The vocabulary (`chars`) and model settings, for use with `model.safetensors` |
|
| 51 |
|
| 52 |
## How to use
|
| 53 |
|
config.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chars": "\n !',.:?ABCDEFGHIJKLMNOPRSTUVWXYabcdefghijklmnoprstuvwxy",
|
| 3 |
+
"block_size": 128,
|
| 4 |
+
"n_layer": 4,
|
| 5 |
+
"n_head": 4,
|
| 6 |
+
"n_embd": 128,
|
| 7 |
+
"dropout": 0.1
|
| 8 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:435e2724ef0d0a8ffe4df843da947f80978a3b1c7903ef1b1282fe05bdebeb0e
|
| 3 |
+
size 3301072
|