diff --git a/README.md b/README.md
index 8df8288..ce41770 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,7 @@
 
 Tools for inspecting sparse circuit models from [Gao et al. 2025](https://openai.com/index/understanding-neural-networks-through-sparse-circuits/). Provides code 
 for running inference as well as a Streamlit dashboard that allows you to interact
-with task-specific circuits found by pruning.
+with task-specific circuits found by pruning. Note: this README was AI-generated and lightly edited.
 
 ## Installation
 
@@ -23,6 +23,10 @@ visualizer loads you can choose a model, dataset, pruning sweep, and node budget
 using the controls in the left column. The plots are rendered with Plotly; most
 elements are interactive and support hover/click exploration.
 
+Example view of the Streamlit circuit visualizer (wte/wpe tab) with node ablation deltas and activation previews:
+
+![Streamlit circuit visualizer](annotated-circuit-sparsity-viz.png)
+
 ## Running Model Forward Passes
 
 Transformer definitions live in `circuit_sparsity.inference.gpt`. The module
diff --git a/annotated-circuit-sparsity-viz.png b/annotated-circuit-sparsity-viz.png
new file mode 100644
index 0000000..f2b2b54
Binary files /dev/null and b/annotated-circuit-sparsity-viz.png differ
diff --git a/circuit_sparsity/viz.py b/circuit_sparsity/viz.py
index d974b61..490200f 100644
--- a/circuit_sparsity/viz.py
+++ b/circuit_sparsity/viz.py
@@ -1485,13 +1485,8 @@ def _maybe_item(x):
         cols = st.columns([1, 1])
         model_config = viz_data["importances"]["beeg_model_config"]
         with cols[0]:
-            cols2 = st.columns([1, 2, 1])
+            cols2 = st.columns([2, 1])
             with cols2[0]:
-                use_pca = st.toggle(
-                    "PCA components",
-                    value=False,
-                )
-            with cols2[1]:
                 chidx = st.selectbox(
                     "res channel index",
                     options=[
@@ -1502,14 +1497,10 @@ def _maybe_item(x):
                 )
                 chidx = int(chidx.split(" ")[0])
 
-            with cols2[2]:
+            with cols2[1]:
                 st.text(f"encname: {model_config.tokenizer_name}")
 
-            if use_pca:
-                U, S, V = get_embed_weights_pca(model_path, q=100)
-                embsort = U[:, chidx].cpu().sort(descending=True)
-            else:
-                embsort = embed_weight[:, chidx].sort(descending=True)
+            embsort = embed_weight[:, chidx].sort(descending=True)
 
             def _filter_embsort(xs):
                 return [
@@ -1588,12 +1579,6 @@ def _filter_embsort(xs):
     status_placeholder.html("<pre>ready<br>&nbsp;</pre>")
 
 
-@cache("get_embed_weights_pca_v1")
-def get_embed_weights_pca(model_path, q):
-    wte = get_embed_weights(model_path).float()
-    return torch.pca_lowrank(wte, niter=10, q=q)
-
-
 @cache("get_embed_weights_v1")
 def get_embed_weights(model_path):
     return get_model_weights(model_path, lambda x: x["transformer.wte.weight"]).half()
@@ -1714,5 +1699,3 @@ def treemap(f, x):
     st.set_page_config(page_title="Circuit viz", layout="wide")
 
     main()
-
-