[WEB] Update models to 8dec and also default values (huggingface#620)

1. Update the models to 8 dec.
2. precision is default to `fp16` in CLI.
3. version is default to `v2.1base` in CLI as well as web.
4. The default scheduler is set to `EulerDiscrete` now.

Signed-Off-by: Gaurav Shukla <[email protected]>

Signed-off-by: Gaurav Shukla <[email protected]>

Author: Shukla-Gaurav

shark/examples/shark_inference/stable_diffusion/stable_args.py

@@ -23,7 +23,7 @@
 p.add_argument(
     "--version",
     type=str,
-    default="v1.4",
+    default="v2.1base",
     help="Specify version of stable diffusion model",
 )
 
@@ -48,7 +48,7 @@
 )
 
 p.add_argument(
-    "--precision", type=str, default="fp32", help="precision to run the model."
+    "--precision", type=str, default="fp16", help="precision to run the model."
 )
 
 p.add_argument(

web/index.py

@@ -101,13 +101,13 @@ def resource_path(relative_path):
                     )
                     version = gr.Radio(
                         label="Version",
-                        value="v1.4",
+                        value="v2.1base",
                         choices=["v1.4", "v2.1base"],
                     )
                 with gr.Row():
                     scheduler_key = gr.Dropdown(
                         label="Scheduler",
-                        value="DPMSolverMultistep",
+                        value="EulerDiscrete",
                         choices=[
                             "DDIM",
                             "PNDM",
@@ -174,9 +174,9 @@ def resource_path(relative_path):
             outputs=[generated_img, std_output],
         )
 
+shark_web.queue()
 shark_web.launch(
     share=False,
     server_name="0.0.0.0",
     server_port=8080,
-    enable_queue=True,
 )

web/models/stable_diffusion/model_wrappers.py

@@ -5,6 +5,7 @@
 
 model_config = {
     "v2": "stabilityai/stable-diffusion-2",
+    "v2.1base": "stabilityai/stable-diffusion-2-1-base",
     "v1.4": "CompVis/stable-diffusion-v1-4",
 }
 
@@ -19,6 +20,16 @@
             torch.tensor(1).to(torch.float32),  # guidance_scale
         ),
     },
+    "v2.1base": {
+        "clip": (torch.randint(1, 2, (1, 77)),),
+        "vae": (torch.randn(1, 4, 64, 64),),
+        "unet": (
+            torch.randn(1, 4, 64, 64),  # latents
+            torch.tensor([1]).to(torch.float32),  # timestep
+            torch.randn(2, 77, 1024),  # embedding
+            torch.tensor(1).to(torch.float32),  # guidance_scale
+        ),
+    },
     "v1.4": {
         "clip": (torch.randint(1, 2, (1, 77)),),
         "vae": (torch.randn(1, 4, 64, 64),),

web/models/stable_diffusion/opt_params.py

@@ -22,7 +22,7 @@ def get_unet(args):
             return get_shark_model(args, bucket, model_name, iree_flags)
         else:
             bucket = "gs://shark_tank/stable_diffusion"
-            model_name = "unet_1dec_fp16"
+            model_name = "unet_8dec_fp16"
             if args.version == "v2.1base":
                 model_name = "unet2base_8dec_fp16"
             iree_flags += [
@@ -56,7 +56,7 @@ def get_vae(args):
         )
     if args.precision == "fp16":
         bucket = "gs://shark_tank/stable_diffusion"
-        model_name = "vae_1dec_fp16"
+        model_name = "vae_8dec_fp16"
         if args.version == "v2.1base":
             model_name = "vae2base_8dec_fp16"
         iree_flags += [
@@ -119,7 +119,7 @@ def get_clip(args):
             f"-iree-vulkan-target-triple={args.iree_vulkan_target_triple}"
         )
     bucket = "gs://shark_tank/stable_diffusion"
-    model_name = "clip_1dec_fp32"
+    model_name = "clip_8dec_fp32"
     if args.version == "v2.1base":
         model_name = "clip2base_8dec_fp32"
     iree_flags += [

web/models/stable_diffusion/stable_args.py

@@ -25,7 +25,7 @@
 p.add_argument(
     "--version",
     type=str,
-    default="v1.4",
+    default="v2.1base",
     help="Specify version of stable diffusion model",
 )
 
@@ -60,8 +60,8 @@
 p.add_argument(
     "--scheduler",
     type=str,
-    default="DPMSolverMultistep",
-    help="can be [PNDM, LMSDiscrete, DDIM, DPMSolverMultistep]",
+    default="EulerDiscrete",
+    help="can be [PNDM, LMSDiscrete, DDIM, DPMSolverMultistep, EulerDiscrete]",
 )
 
 p.add_argument(