| <!doctype html> |
| <html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1"> |
| <title>Pratham Prateek Mohanty: research and open data</title> |
| <meta name="description" content="Research papers and open data on multi-turn emotion labeling by Pratham Prateek Mohanty, OpenNHE Technologies."> |
| <link rel="stylesheet" href="style.css"></head><body> |
| <nav><div class="in"><b>P. P. Mohanty</b><a href="#papers">Papers</a><a href="#data">Open data</a><a href="#explore">Explore</a><a href="#schema">Schema</a><a href="#cite">Cite</a></div></nav> |
| <header><div class="in"> |
| <span class="tag">Research 路 Open data 路 OpenNHE Technologies</span> |
| <h1>Emotion labeling research and an open dataset</h1> |
| <p class="authors"><b>Pratham Prateek Mohanty</b> 路 Masters' Union 路 OpenNHE Technologies</p> |
| <p>Three technical papers on small LoRA adapters that label emotions in text, and the public synthetic dataset behind the Elysium X 150 FR paper. Every limit is stated next to its number.</p> |
| <div class="btns"> |
| <a class="btn p" href="#papers">Read the papers</a> |
| <a class="btn" href="#explore">Browse the data</a> |
| <a class="btn" href="https://huggingface.co/open-nhe" target="_blank" rel="noopener">OpenNHE on Hugging Face</a> |
| </div> |
| <div class="stats"> |
| <div class="stat"><b data-n="3">3</b><span>technical papers with DOIs</span></div> |
| <div class="stat"><b data-n="4000">4,000</b><span>labeled target turns in the open dataset</span></div> |
| <div class="stat"><b data-n="150">150</b><span>emotion and appraisal coordinates</span></div> |
| <div class="stat"><b>CC BY 4.0</b><span>dataset license</span></div> |
| </div></div></header> |
| <main> |
| <section id="papers"><h2>Papers</h2><div class="two"> |
| <div class="card pcard"><span class="badge ok">Open: model, data, code</span><h3>Elysium X 150 FR</h3> |
| <p>A small LoRA adapter (73.9 MB on Qwen2.5-1.5B-Instruct) for sparse, per-speaker emotion and appraisal labeling over a 150-coordinate schema. On a frozen 78-row synthetic test it reaches micro-F1 0.7251 and 49 of 78 exact label sets, agreement with a reviewed synthetic teacher only.</p> |
| <p class="note">No external benchmark, 117 of 150 coordinates seen in training, English only. No state-of-the-art claim.</p> |
| <div class="btns"><a class="btn p" href="https://doi.org/10.5281/zenodo.23155240" target="_blank" rel="noopener">Paper 路 DOI 10.5281/zenodo.23155240</a> |
| <a class="btn" href="https://huggingface.co/spaces/open-nhe/Elysium-X-150-FR-Paper" target="_blank" rel="noopener">Paper page and demo</a> |
| <a class="btn" href="https://huggingface.co/open-nhe/Elysium-X-150-FR" target="_blank" rel="noopener">Model</a></div></div> |
| <div class="card pcard"><span class="badge ok">Open: adapter weights on Hugging Face</span><h3>Elysium X 20 FR</h3> |
| <p>A small LoRA adapter (18.5M trainable parameters) on Qwen2.5-1.5B-Instruct that reads one English message and returns structured JSON: GoEmotions labels with rater-agreement intensities, plus valence and arousal. On 1,500 GoEmotions test examples it reaches micro-F1 0.570 against 0.304 for always predicting neutral.</p> |
| <p class="note">One seed, one run, no confidence intervals, English Reddit text only. First version, no state-of-the-art claim.</p> |
| <div class="btns"><a class="btn p" href="https://doi.org/10.5281/zenodo.23161139" target="_blank" rel="noopener">Paper 路 DOI 10.5281/zenodo.23161139</a> |
| <a class="btn" href="https://huggingface.co/open-nhe/Elysium-X-20-FR" target="_blank" rel="noopener">Model</a></div></div> |
| <div class="card pcard"><span class="badge pr">Paper only: model and data are proprietary</span><h3>Elysium X 500 FR</h3> |
| <p>A first baseline for multilingual emotion labeling over a 500-coordinate schema in 12 languages, measured when the emotion word is hidden. The headline is the hard one: 19.4% exact match on 403 no-name rows. With the word present it is 95.1%, which mostly reflects recognising a label word.</p> |
| <p class="note">Template-generated synthetic data, possible translation leakage, one seed. The weights, data and code are not public.</p> |
| <div class="btns"><a class="btn p" href="https://doi.org/10.5281/zenodo.23159402" target="_blank" rel="noopener">Paper 路 DOI 10.5281/zenodo.23159402</a> |
| <a class="btn" href="https://huggingface.co/spaces/open-nhe/Elysium-X-500-FR-Paper" target="_blank" rel="noopener">Paper showcase</a></div></div> |
| </div></section> |
|
|
| <section id="data"><h2>Open data: Elysium X 150 FR Emotion Dataset</h2><div class="card"> |
| <p>4,000 labeled target turns from 1,000 four-turn English dialogues, annotated on the 150-coordinate schema. The dialogues are synthetic, written by an AI from scenario prompts. They are not real people's conversations.</p> |
| <ul> |
| <li>Labels came from the author's ChatGPT labeling workflow and were personally reviewed by him. That is owner-reported review, not independent multi-annotator gold.</li> |
| <li>1,401 of 4,000 rows have no label, meaning no expressed coordinate for that speaker at that turn.</li> |
| <li>English only. Not a clinical, diagnostic or safety dataset.</li> |
| <li>The model's 78 test rows are marked <code>test</code> in <code>model_split</code>. Do not train on them if you want to compare with the paper.</li></ul> |
| <div class="btns"><a class="btn p" href="https://huggingface.co/datasets/open-nhe/Elysium-X-150-FR-dataset" target="_blank" rel="noopener">Dataset on Hugging Face</a></div></div></section> |
|
|
| <section id="explore"><h2>Explore the data</h2> |
| <p class="note">A random sample of 250 rows from the open dataset (210 labeled, 40 unlabeled), shown with the label names. The target turn is outlined. Full data is in the dataset repo.</p> |
| <div class="ctl"><input id="q" placeholder="Search the dialogue text"><select id="fam"><option value="">All families</option></select><select id="coord"><option value="">All coordinates</option></select><select id="lab"><option value="all">Labeled and unlabeled</option><option value="yes">Labeled only</option><option value="no">Unlabeled only</option></select></div> |
| <p class="note" id="count"></p><div id="rows"></div><button class="btn more" id="more">Show more</button></section> |
|
|
| <section id="schema"><h2>The 150-coordinate schema</h2><p class="note">Ten families of fifteen coordinates. Each has an operational definition and a note on what is not enough on its own, in <code>taxonomy.json</code> in the dataset repo.</p><div class="fams" id="famgrid"></div></section> |
|
|
| <section id="cite"><h2>Cite</h2><pre>Mohanty, P. P. (2026). Elysium X 150 FR: A Small LoRA Adapter for Sparse, Per-Speaker Emotion and Appraisal |
| Labeling over a 150-Coordinate Schema. Zenodo. https://doi.org/10.5281/zenodo.23155240 |
|
|
| Mohanty, P. P. (2026). Elysium X 500 FR: A First Public Baseline for Multilingual Emotion Labeling over a |
| 500-Coordinate Schema When the Emotion Word Is Hidden. Zenodo. https://doi.org/10.5281/zenodo.23159402 |
|
|
| Mohanty, P. P. (2026). Elysium X 20 FR: A Small LoRA Adapter That Turns One Message into a Structured Emotion |
| Appraisal, Trained on GoEmotions. Zenodo. https://doi.org/10.5281/zenodo.23161139 |
|
|
| Mohanty, P. P. (2026). Elysium X 150 FR Emotion Dataset (CC BY 4.0). |
| https://huggingface.co/datasets/open-nhe/Elysium-X-150-FR-dataset</pre></section> |
| </main> |
| <footer><div class="in">Pratham Prateek Mohanty 路 OpenNHE Technologies 路 <a href="https://huggingface.co/open-nhe" target="_blank" rel="noopener">open-nhe</a></div></footer> |
| <script src="data.js"></script><script src="app.js"></script> |
| </body></html> |
|
|
