| --- |
| license: apache-2.0 |
| base_model: facebook/bart-large-cnn |
| tags: [summarization, scientific-papers, highlights, scihigh-2026, fire-2026] |
| --- |
| |
| # SciHigh-2026 Subtask 1 — bart-large-cnn titled (run 1) |
|
|
| Full fine-tune of bart-large-cnn. |
|
|
| Fine-tuned for the FIRE 2026 SciHigh shared task, Subtask 1: generating |
| research highlights from scientific paper abstracts (MixSub corpus). |
|
|
| - Input format: `<paper title> | <abstract>` (corpus punctuation-stripped |
| style; abstracts repaired via DOI-verified Semantic Scholar recovery) |
| - Output: 3-5 highlight sentences, ~55 words, greedy decoding |
| - Training: one epoch on 15,960 pairs, Kaggle dual T4 |
| - Validation ROUGE-L F1 (plain LCS, no stemming): **26.37** |
|
|
| ## Repository contents |
|
|
| | file | purpose | |
| | --- | --- | |
| | `train_final.py` | standalone training recipe that produced this model | |
| | `train_final.ipynb` | notebook version of the same recipe | |
| | `final/` | full data pipeline: abstract recovery, title fetch, dataset build, submission validation | |
|
|
| ## Data pipeline summary |
|
|
| ~40% of corpus abstracts are truncated mid-sentence. Each row was matched to |
| its ScienceDirect PII (100% verified join), resolved to a DOI via Elsevier's |
| keyless API, and its complete abstract recovered from Semantic Scholar under |
| a label-free >=90%-token-overlap acceptance filter (~95% of truncated rows |
| repaired). Paper titles were fetched for 100% of rows and prepended to |
| inputs. See `final/README.md` for the full pipeline and negative-results |
| summary. Trained 2026-08-08. |
|
|
