update optimized ORT model
Browse files- .gitattributes +1 -0
- model.onnx → model_optimized.onnx +2 -2
- ort_config.json +33 -0
.gitattributes
CHANGED
@@ -25,3 +25,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
25 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
26 |
*.zstandard filter=lfs diff=lfs merge=lfs -text
|
27 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
25 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
26 |
*.zstandard filter=lfs diff=lfs merge=lfs -text
|
27 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
28 |
+
model_optimized.onnx filter=lfs diff=lfs merge=lfs -text
|
model.onnx → model_optimized.onnx
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1ea31f6093dff978e39e6e2fcb107707d86d9e1d057b463a0cf768602ae43338
|
3 |
+
size 122385803
|
ort_config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"one_external_file": true,
|
3 |
+
"opset": null,
|
4 |
+
"optimization": {
|
5 |
+
"disable_attention": null,
|
6 |
+
"disable_attention_fusion": false,
|
7 |
+
"disable_bias_gelu": null,
|
8 |
+
"disable_bias_gelu_fusion": false,
|
9 |
+
"disable_bias_skip_layer_norm": null,
|
10 |
+
"disable_bias_skip_layer_norm_fusion": false,
|
11 |
+
"disable_embed_layer_norm": true,
|
12 |
+
"disable_embed_layer_norm_fusion": true,
|
13 |
+
"disable_gelu": null,
|
14 |
+
"disable_gelu_fusion": false,
|
15 |
+
"disable_layer_norm": null,
|
16 |
+
"disable_layer_norm_fusion": false,
|
17 |
+
"disable_shape_inference": false,
|
18 |
+
"disable_skip_layer_norm": null,
|
19 |
+
"disable_skip_layer_norm_fusion": false,
|
20 |
+
"enable_gelu_approximation": false,
|
21 |
+
"enable_transformers_specific_optimizations": true,
|
22 |
+
"fp16": false,
|
23 |
+
"no_attention_mask": false,
|
24 |
+
"optimization_level": 2,
|
25 |
+
"optimize_for_gpu": false,
|
26 |
+
"optimize_with_onnxruntime_only": null,
|
27 |
+
"use_mask_index": false
|
28 |
+
},
|
29 |
+
"optimum_version": "1.6.3",
|
30 |
+
"quantization": {},
|
31 |
+
"transformers_version": "4.26.0",
|
32 |
+
"use_external_data_format": false
|
33 |
+
}
|