first commit

README.md

@@ -1,15 +1,19 @@
 ---
-title: Lit2vec Subfield Classifier Space
-emoji: 💬
-colorFrom: yellow
-colorTo: purple
+title: Lit2Vec TL;DR Summarizer
+emoji: 🧪
+colorFrom: blue
+colorTo: green
 sdk: gradio
 sdk_version: 5.42.0
 app_file: app.py
 pinned: false
 hf_oauth: true
 hf_oauth_scopes:
-- inference-api
+  - inference-api
+short_description: TL;DR summarizer for chemistry research abstracts
 ---
 
-An example chatbot using [Gradio](https://gradio.app), [`huggingface_hub`](https://huggingface.co/docs/huggingface_hub/v0.22.2/en/index), and the [Hugging Face Inference API](https://huggingface.co/docs/api-inference/index).
+
+Lit2Vec TL;DR is an abstractive summarization tool for chemistry research abstracts, built using Gradio and a fine-tuned DistilBART model. It generates concise, structured summaries capturing the **methods**, **results**, and **significance** of scientific papers.
+🔬 Model: [`Bocklitz-Lab/lit2vec-tldr-bart-model`](https://huggingface.co/Bocklitz-Lab/lit2vec-tldr-bart-model)  
+

app.py

@@ -1,70 +1,117 @@
+import os
+import json
+import numpy as np
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-
-def respond(
-    message,
-    history: list[dict[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    hf_token: gr.OAuthToken,
-):
-    """
-    For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-    """
-    client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
-
-    messages = [{"role": "system", "content": system_message}]
-
-    messages.extend(history)
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        choices = message.choices
-        token = ""
-        if len(choices) and choices[0].delta.content:
-            token = choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-chatbot = gr.ChatInterface(
-    respond,
-    type="messages",
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
-
-with gr.Blocks() as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
-    chatbot.render()
+from huggingface_hub import hf_hub_download
+from tensorflow import keras
+from sentence_transformers import SentenceTransformer
 
+# -----------------------
+# Config
+# -----------------------
+REPO_ID = "Bocklitz-Lab/lit2vec-subfield-classifier-model"
+EMBED_MODEL = "intfloat/e5-large-v2"
+TEXT_PREFIX = {"abstract": "abstract: ", "summary": "summary: "}
+DEFAULT_THRESHOLD = 0.5
+TOPK_DEFAULT = 5
+
+# -----------------------
+# Load model + labels at startup
+# -----------------------
+# Keras model (saved as .h5 on the Hub)
+MODEL_PATH = hf_hub_download(REPO_ID, filename="mlp_model.h5")
+LABEL_MAP_PATH = hf_hub_download(REPO_ID, filename="label_mapping.json")
+
+with open(LABEL_MAP_PATH, "r", encoding="utf-8") as f:
+    mapping = json.load(f)
+INDEX_TO_LABEL = {int(k): v for k, v in mapping["index_to_label"].items()}
+
+# load Keras model for inference
+MODEL = keras.models.load_model(MODEL_PATH, compile=False)
+
+# SentenceTransformer encoder (CPU-only for portability)
+ENCODER = SentenceTransformer(EMBED_MODEL, device="cpu")
+
+def encode_text(text: str, text_type: str = "abstract") -> np.ndarray:
+    """Encode text into normalized embedding compatible with the classifier."""
+    prefix = TEXT_PREFIX.get(text_type, "")
+    emb = ENCODER.encode([prefix + text], normalize_embeddings=True)  # shape: (1, D)
+    return emb.astype("float32")
+
+def predict(text: str, text_type: str, threshold: float, topk: int):
+    """Return selected labels (by threshold), top-k labels, and a scores table."""
+    text = (text or "").strip()
+    if not text:
+        return ("", "", [])
+
+    X = encode_text(text, text_type=text_type)  # (1, D)
+    probs = MODEL.predict(X, verbose=0)[0]      # (18,)
+
+    # Thresholded predictions
+    pred_ids = [i for i, p in enumerate(probs) if p >= threshold]
+    pred_labels = [INDEX_TO_LABEL[i] for i in pred_ids]
+    pred_display = ", ".join(pred_labels) if pred_labels else "—"
+
+    # Top-k predictions (by score)
+    topk = max(1, int(topk))
+    order = np.argsort(-probs)[:topk]
+    topk_items = [f"{INDEX_TO_LABEL[i]}: {probs[i]:.3f}" for i in order]
+    topk_display = "\n".join(topk_items)
+
+    # Build a table of all scores (sorted desc)
+    sorted_ids = np.argsort(-probs)
+    table = [[INDEX_TO_LABEL[i], float(probs[i])] for i in sorted_ids]
+
+    return pred_display, topk_display, table
+
+# -----------------------
+# Gradio UI
+# -----------------------
+with gr.Blocks(fill_height=True) as demo:
+    gr.Markdown(
+        """
+        # 🔬 Lit2Vec Subfield Classifier
+        Enter a **chemistry abstract or summary**. The app encodes it with `e5-large-v2` and predicts one or more **subfields** using the MLP model.
+
+        **Model:** `Bocklitz-Lab/lit2vec-subfield-classifier-model`  
+        **Encoder:** `intfloat/e5-large-v2`
+        """
+    )
+
+    with gr.Row():
+        text_type = gr.Radio(
+            choices=["abstract", "summary"], value="abstract", label="Text type (prefix used for encoding)"
+        )
+        threshold = gr.Slider(0.0, 1.0, value=DEFAULT_THRESHOLD, step=0.01, label="Decision threshold")
+        topk = gr.Slider(1, 10, value=TOPK_DEFAULT, step=1, label="Top-K to display")
+
+    input_box = gr.Textbox(
+        label="Paste abstract / summary",
+        placeholder="Paste your chemistry abstract here…",
+        lines=12,
+        value="We investigate Pt-based nanoparticle catalysts for hydrogen evolution, "
+              "demonstrating improved activity via ligand tuning and support engineering."
+    )
+
+    run_btn = gr.Button("Predict subfield(s)")
+
+    with gr.Row():
+        selected_labels = gr.Textbox(label="Predicted fields (thresholded)", lines=2)
+        topk_labels = gr.Textbox(label="Top-K (scores)", lines=6)
+
+    scores_table = gr.Dataframe(
+        headers=["Subfield", "Score"],
+        datatype=["str", "number"],
+        label="All scores (sorted)",
+        interactive=False
+    )
+
+    run_btn.click(
+        fn=predict,
+        inputs=[input_box, text_type, threshold, topk],
+        outputs=[selected_labels, topk_labels, scores_table]
+    )
 
 if __name__ == "__main__":
+    # On Spaces, Gradio sets host/port; keep defaults.
     demo.launch()

requirements.txt

@@ -0,0 +1,7 @@
+gradio==5.42.0
+transformers==4.55.3
+torch==2.8.0
+sentence-transformers==2.7.0
+huggingface_hub==0.23.0
+tensorflow-cpu==2.15.0
+numpy==1.26