fix: use correct version of orbax (borisdayma#337)

Author: borisdayma

app/gradio/app.py

@@ -22,7 +22,6 @@ def infer(prompt):
     with gr.Group():
         with gr.Box():
             with gr.Row().style(mobile_collapse=False, equal_height=True):
-
                 text = gr.Textbox(
                     label="Enter your prompt", show_label=False, max_lines=1
                 ).style(

setup.cfg

@@ -31,6 +31,7 @@ install_requires =
     pillow
     jax==0.3.25
     flax==0.6.3
+    orbax==0.0.23
     wandb
 
 [options.extras_require]

src/dalle_mini/model/modeling.py

@@ -77,6 +77,7 @@ def _smelu(x: Any) -> Any:
 
 ACT2FN.update({"smelu": smelu()})
 
+
 # deepnet initialization
 def deepnet_init(init_std, gain=1):
     init = jax.nn.initializers.normal(init_std)
@@ -498,7 +499,6 @@ class GLU(nn.Module):
 
     @nn.compact
     def __call__(self, x: jnp.ndarray, deterministic: bool = True) -> jnp.ndarray:
-
         if self.config.use_deepnet_scaling:
             gain = deepnet_gain["encoder" if self.is_encoder else "decoder"]["beta"](
                 self.config
@@ -567,7 +567,6 @@ class FFN(nn.Module):
 
     @nn.compact
     def __call__(self, x: jnp.ndarray, deterministic: bool = True) -> jnp.ndarray:
-
         if self.config.use_deepnet_scaling:
             gain = deepnet_gain["encoder" if self.is_encoder else "decoder"]["beta"](
                 self.config
@@ -634,7 +633,6 @@ def __call__(
         output_attentions: bool = True,
         deterministic: bool = True,
     ) -> Tuple[jnp.ndarray]:
-
         if self.config.use_scan:
             hidden_states = hidden_states[0]
 
@@ -742,7 +740,6 @@ def __call__(
         output_attentions: bool = True,
         deterministic: bool = True,
     ) -> Tuple[jnp.ndarray]:
-
         if self.config.use_scan:
             hidden_states = hidden_states[0]