Final final final updates to app.py

app.py

@@ -1,8 +1,8 @@
 try:
-    import spaces          #only on hugging face zerogpu, has to come before torch
+    import spaces          
     GPU = spaces.GPU
 except ImportError:
-    GPU = lambda f: f       #running locally --> no decorator needed
+    GPU = lambda f: f       
 
 import gradio as gr
 import torch
@@ -32,9 +32,9 @@ NAMES = {"WORKLOAD": "Sleep and workload", "TEAM": "Team dynamics",
          "ADJUSTMENT": "Adjustment and coping", "OUTSIDE_COMMS": "Outside communication"}
 
 
-# ---------------- the three models ----------------
+# the three models 
 
-#1. vader baseline: keyword lists + sentiment gate (thresholds picked on the validation split in layer 6)
+#1. vader baseline: keyword lists + sentiment gate 
 KEYWORDS = {
     "WORKLOAD": ["exhausted", "exhaustion", "tired", "fatigue", "overworked", "sleep", "schedule", "timeline",
                  "overtime", "demanding", "heavy", "restless", "shift", "tasks", "pressure"],
@@ -56,7 +56,7 @@ STOP = set(stopwords.words("english"))
 stemmer = SnowballStemmer("english")
 
 #3. bert-base-uncased + my lora adapter (the adapter files sit next to this app.py)
-#thresholds tuned on the validation split in layer 7
+
 BERT_THRESHOLDS = {"WORKLOAD": 0.65, "TEAM": 0.90, "ADJUSTMENT": 0.85, "OUTSIDE_COMMS": 0.90}
 tok = AutoTokenizer.from_pretrained(HERE)
 base = AutoModelForSequenceClassification.from_pretrained(
@@ -78,7 +78,7 @@ def vader_flags(seg):
 
 
 def clean_tfidf(text):
-    #lowercase, strip punctuation, drop stopwords, stem (same five steps as layer 6)
+    #lowercase, strip punctuation, drop stopwords, stem
     text = re.sub(r"[^a-z\s]", " ", str(text).lower())
     return " ".join(stemmer.stem(w) for w in text.split() if w not in STOP and len(w) > 1)
 
@@ -97,7 +97,7 @@ def bert_probs(segs):
     return np.vstack(out)
 
 
-# ---------------- the text pipeline (layers 2 and 3) ----------------
+#the text pipeline 
 
 #five cleaning rules in a fixed order
 WRAPPED_RE = re.compile(r"\n([A-Z])\n([A-Z][A-Za-z\-]{1,30}:)")
@@ -194,7 +194,7 @@ def segment(text):
     return [c for c in chunks if len(c.split()) >= MIN_WORDS]
 
 
-# ---------------- running everything ----------------
+#running everything 
 
 def read_input(file, text):
     if file is None:
@@ -278,13 +278,11 @@ def analyse(file, text):
     rows = ""
     for i, c in enumerate(CATEGORIES):
         counts = [int(v[:, i].sum()), int((t[:, i] >= 0.5).sum()), int((b[:, i] >= BERT_THRESHOLDS[c]).sum())]
-        cells = "".join(bar(k, n, col) for k, col in zip(counts, ["#7c86a8", "#38bdf8", "#818cf8"]))
+        cells = "".join(bar(k, n, col) for k, col in zip(counts, ["#f472b6", "#a78bfa", "#34d399"]))
         rows += f"<tr><td>{NAMES[c]}</td>{cells}<td>{meter(b[:, i].max(), BERT_THRESHOLDS[c])}</td></tr>"
     s5 = (f"<h4>5. Classification: all three models</h4>"
           f"<table><tr><th>Category</th><th>VADER</th><th>TF-IDF</th><th>BERT + LoRA</th><th>BERT top score vs threshold</th></tr>{rows}</table>"
-          f"<p class='small'>Segments flagged out of {n}. White line = BERT threshold, the bar fills to the top segment score "
-          f"(indigo if it passes). BERT is my final model. A research prototype, it flags passages for a person to read "
-          f"and does not diagnose anyone.</p>")
+          f"<p class='small'>Segments flagged out of {n}. White line = BERT threshold, the bar fills to the top segment score </p>")
 
     order = np.argsort(-b.max(axis=1))   #strongest segments first for shap
     choices = [f"#{k + 1}  (BERT max {b[k].max():.2f})  {segs[k][:60]}..." for k in order]
@@ -313,7 +311,7 @@ def explain(state, choice, category):
             f"<div class='seg' style='line-height:1.9'>{spans}</div>")
 
 
-# ---------------- the page ----------------
+#  the page 
 
 CSS = """
 .gradio-container {background: radial-gradient(ellipse at top, #141b33 0%, #05070f 70%) !important;}
@@ -345,7 +343,7 @@ with gr.Blocks(title="SPACEMIND") as demo:
         #right side: the stages
         with gr.Column():
             out1 = gr.Textbox(label="1. Input (word count)")
-            out2 = gr.Textbox(label="2. Cleaning (five rules, removed counts)")
+            out2 = gr.Textbox(label="2. Cleaning")
             out3 = gr.Textbox(label="3. Removing the interviewer", lines=3)
             out4 = gr.HTML(label="4. Segmenting")
             out5 = gr.HTML(label="5. Classification")