add vae encoder component

README.md

@@ -3,8 +3,9 @@ license: apache-2.0
 ---
 
 ```python
-from optimum.intel.openvino.modeling_diffusion import OVStableDiffusionPipeline
+from optimum.intel.openvino import OVStableDiffusionPipeline
 
-pipe = OVStableDiffusionPipeline.from_pretrained("hf-internal-testing/tiny-stable-diffusion-openvino")
+model_id = "hf-internal-testing/tiny-stable-diffusion-openvino"
+pipe = OVStableDiffusionPipeline.from_pretrained(model_id)
 ```
 

model_index.json

@@ -1,6 +1,6 @@
 {
   "_class_name": "OVStableDiffusionPipeline",
-  "_diffusers_version": "0.13.1",
+  "_diffusers_version": "0.14.0",
   "feature_extractor": [
     "transformers",
     "CLIPFeatureExtractor"
@@ -28,5 +28,9 @@
   "vae_decoder": [
     "optimum",
     "OVModelVaeDecoder"
+  ],
+  "vae_encoder": [
+    "optimum",
+    "OVModelVaeEncoder"
   ]
 }

scheduler/scheduler_config.json

@@ -1,6 +1,6 @@
 {
   "_class_name": "PNDMScheduler",
-  "_diffusers_version": "0.13.1",
+  "_diffusers_version": "0.14.0",
   "beta_end": 0.012,
   "beta_schedule": "scaled_linear",
   "beta_start": 0.00085,

tokenizer/tokenizer_config.json

@@ -19,7 +19,7 @@
   },
   "errors": "replace",
   "model_max_length": 77,
-  "name_or_path": "/tmp/tmp3si_ovlc/tokenizer",
+  "name_or_path": "/tmp/tmp6i7rva3d/tokenizer",
   "pad_token": "<|endoftext|>",
   "special_tokens_map_file": "/home/lysandre/.cache/huggingface/transformers/18a566598f286c9139f88160c99f84eec492a26bd22738fa9cb44d5b7e0a5c76.cce1206abbad28826f000510f22f354e53e66a97f7c23745a7dfe27609cc07f5",
   "tokenizer_class": "CLIPTokenizer",

unet/openvino_model.xml

@@ -27,7 +27,7 @@
 			</output>
 		</layer>
 		<layer id="0" name="encoder_hidden_states" type="Parameter" version="opset1">
-			<data shape="?,77,?" element_type="f32" />
+			<data shape="?,77,32" element_type="f32" />
 			<rt_info>
 				<attribute name="fused_names" version="0" value="encoder_hidden_states" />
 			</rt_info>
@@ -35,7 +35,7 @@
 				<port id="0" precision="FP32" names="encoder_hidden_states">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 			</output>
 		</layer>
@@ -1436,7 +1436,7 @@
 				</port>
 			</output>
 		</layer>
-		<layer id="92" name="Constant_256350" type="Const" version="opset1">
+		<layer id="92" name="Constant_274101" type="Const" version="opset1">
 			<data element_type="i64" shape="1" offset="58936" size="8" />
 			<rt_info>
 				<attribute name="fused_names" version="0" value="/Reshape" />
@@ -12418,7 +12418,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -13418,7 +13418,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -22714,7 +22714,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -23714,7 +23714,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -33010,7 +33010,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -34010,7 +34010,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -45425,7 +45425,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -46425,7 +46425,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -55980,7 +55980,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -56980,7 +56980,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -66561,7 +66561,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>
@@ -67561,7 +67561,7 @@
 				<port id="0" precision="FP32">
 					<dim>-1</dim>
 					<dim>77</dim>
-					<dim>-1</dim>
+					<dim>32</dim>
 				</port>
 				<port id="1" precision="FP32">
 					<dim>32</dim>

vae_encoder/openvino_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14509b2db28c5e75866ae3795e5e636db33ec5d47bdefa836f1bf97b5bfceba7
+size 1021864