suyashnpande/cinematic-music-descriptor-v2-module1

Cinematic Music Descriptor — Module 1 – Local Scene Encoder

RoBERTa-base finetuned to encode individual movie script scenes into 768-dim vectors, with multi-task heads for scene-level cinematic attributes.

Label Schema

Classification

• —emotional_valence: 4 classes
• —conflict_nature: 6 classes
• —acoustic_space: 6 classes
• —reality_layer: 5 classes

Regression

• —pacing_intensity: 1–10
• —scene_arousal: 0.0–1.0

Training Details

• —Base model: roberta-base
• —Dataset: ~11,000 scenes from 60–80 movies
• —Framework: PyTorch + HuggingFace Transformers
• —Logging: Weights & Biases

Usage

import torch
from huggingface_hub import hf_hub_download
import config as C

# Download weights  (repo ID is built from HF_REPO_ID in config.py)
# e.g. "suyashnpande/cinematic-music-descriptor-v2-module3"
repo_id = f"{C.HF_REPO_ID}-module3"
path = hf_hub_download(repo_id=repo_id,
                       filename="module3_final.pt",
                       token=C.HF_READ_TOKEN)

Citation

If you use this model, please cite the project.