Add sentiment analyzer CPU Space

README.md

@@ -1,12 +1,13 @@
 ---
-title: Sentiment Analyzer Cpu
-emoji: 😻
-colorFrom: indigo
+title: Sentiment Analyzer CPU
+colorFrom: yellow
 colorTo: gray
 sdk: gradio
-sdk_version: 6.10.0
 app_file: app.py
 pinned: false
+license: mit
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Sentiment Analyzer CPU
+
+Free CPU sentiment analyzer using `cardiffnlp/twitter-roberta-base-sentiment-latest`.

app.py

@@ -0,0 +1,39 @@
+import gradio as gr
+
+from sentiment_tool.service import SentimentService
+
+
+service = SentimentService()
+
+
+def analyze_sentiment(text):
+    return service.analyze(text)
+
+
+with gr.Blocks(
+    title="Sentiment Analyzer CPU",
+    theme=gr.themes.Soft(primary_hue="yellow", secondary_hue="gray"),
+) as demo:
+    gr.Markdown(
+        """
+        # Sentiment Analyzer CPU
+        Paste text and get positive, neutral, and negative sentiment scores.
+        """
+    )
+
+    text_input = gr.Textbox(label="Input Text", lines=8, placeholder="Write review, opinion, or message here")
+    run_button = gr.Button("Analyze", variant="primary")
+
+    label_output = gr.Textbox(label="Top Label", lines=1)
+    score_output = gr.Textbox(label="Score Breakdown", lines=4)
+    status_output = gr.Textbox(label="Status", lines=2)
+
+    run_button.click(
+        fn=analyze_sentiment,
+        inputs=[text_input],
+        outputs=[label_output, score_output, status_output],
+    )
+
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio>=5.23.0
+huggingface_hub>=0.34.0,<1.0
+safetensors>=0.5.3
+torch>=2.3.0
+transformers>=4.49.0

sentiment_tool/__init__.py

@@ -0,0 +1,3 @@
+from .service import SentimentService
+
+__all__ = ["SentimentService"]

sentiment_tool/service.py

@@ -0,0 +1,52 @@
+import os
+
+import torch
+
+
+MODEL_ID = "cardiffnlp/twitter-roberta-base-sentiment-latest"
+
+
+class SentimentService:
+    def __init__(self):
+        self.pipe = None
+        cpu_count = os.cpu_count() or 1
+        torch.set_num_threads(max(1, min(4, cpu_count)))
+
+    def analyze(self, text):
+        clean_text = " ".join((text or "").split())
+        if not clean_text:
+            return "", "", "Write text first."
+
+        try:
+            results = self._run_model(clean_text)
+            top = max(results, key=lambda item: item["score"])
+            top_label = top["label"].title()
+            breakdown = self._format_results(results)
+            return top_label, breakdown, f"Analyzed sentiment with {MODEL_ID}."
+        except Exception as exc:
+            return "", "", f"Sentiment analysis failed: {type(exc).__name__}: {exc}"
+
+    def _load_pipeline(self):
+        if self.pipe is not None:
+            return
+
+        from transformers import pipeline
+
+        self.pipe = pipeline(
+            "text-classification",
+            model=MODEL_ID,
+            top_k=None,
+            device=-1,
+        )
+
+    def _run_model(self, text):
+        self._load_pipeline()
+        return self.pipe(text)[0]
+
+    def _format_results(self, results):
+        lines = []
+        for item in sorted(results, key=lambda x: x["score"], reverse=True):
+            label = item["label"].title()
+            score = f"{item['score'] * 100:.1f}%"
+            lines.append(f"{label}: {score}")
+        return "\n".join(lines)