Commit ·
4c9eba5
1
Parent(s): bf586e1
updated README.
Browse files
README.md
CHANGED
|
@@ -41,12 +41,12 @@ It powers the [**LilyScript**](https://huggingface.co/spaces/k-l-lambda/LilyScri
|
|
| 41 |
```
|
| 42 |
model_*.chkpt torch training checkpoint (full precision)
|
| 43 |
.state.yaml training / architecture config
|
|
|
|
| 44 |
onnx/
|
| 45 |
patch_kv_int8.onnx patch decoder, int8, with KV-cache (incremental)
|
| 46 |
token_kv_int8.onnx token decoder, int8, with KV-cache (incremental)
|
| 47 |
wte.npy token-embedding table [vocab, hidden]
|
| 48 |
geometry.json patch size, special ids, per-level KV geometry
|
| 49 |
-
lilylet-tokenizer.json tokenizer
|
| 50 |
```
|
| 51 |
|
| 52 |
The `onnx/` bundle is **torch-free**: a generator needs only `onnxruntime` + `numpy`
|
|
|
|
| 41 |
```
|
| 42 |
model_*.chkpt torch training checkpoint (full precision)
|
| 43 |
.state.yaml training / architecture config
|
| 44 |
+
tokenizer.json tokenizer
|
| 45 |
onnx/
|
| 46 |
patch_kv_int8.onnx patch decoder, int8, with KV-cache (incremental)
|
| 47 |
token_kv_int8.onnx token decoder, int8, with KV-cache (incremental)
|
| 48 |
wte.npy token-embedding table [vocab, hidden]
|
| 49 |
geometry.json patch size, special ids, per-level KV geometry
|
|
|
|
| 50 |
```
|
| 51 |
|
| 52 |
The `onnx/` bundle is **torch-free**: a generator needs only `onnxruntime` + `numpy`
|