--- base_model: HuggingFaceTB/SmolLM3-3B library_name: peft tags: [named-entity-recognition, chain-of-thought, lora] language: [en] pipeline_tag: text-generation datasets: [disi-unibo-nlp-students/pilener-teacher-cot-agreement-v1] --- # SmolLM3-3B ยท Structured Reasoning for NER Part of the code, data and models of the bachelor thesis *Structuring Universal Chain-of-thought for Named Entity Recognition* (Department of Computer Science and Engineering, University of Bologna, 2026). All the repositories cited by name belong to the same organization. **Second phase** of the project: a teacher reasons twice on the same queries, freely and in a structured format, and students that differ only in the format of their reasoning are trained on the traces. This is the **structured** student; the free one is `smollm3-3b-free-cot-lora`. Given a text and a list of entity types (some present, some absent), the model reasons inside its `` block with three one-line fields: `TYPES` (which requested types occur), `MENTIONS` (mentions to extract and candidates rejected) and `EDGE` (ambiguous boundaries or type assignments), then answers with a JSON object that maps every requested type to the mentions copied verbatim from the text (an empty list for an absent type). ## Training - **Base model**: `HuggingFaceTB/SmolLM3-3B`. - **Data**: the `structured_cot` traces of `disi-unibo-nlp-students/pilener-teacher-cot-agreement-v1` (train, 3,822 queries), each with its own prompt (`structured_prompt`); target: the teacher's answer (`answer`). The free and structured traces of a query lead to the same answer, so all the students see the same queries and the same targets. - **Method**: LoRA r 16, alpha 32, dropout 0.05 on the q, k, v, o, gate, up and down projections; lr 1e-4, cosine, warmup 3%; 2 epochs; batch 1 x 8 gradient accumulation; bf16; maximum sequence length 5,120; loss on the assistant turn only; 5% validation, best checkpoint by validation loss; seed 42. ## Evaluation Metric: micro F1 with exact, case-insensitive mention matching per type, over the requested types; reasoning length in tokens (generated before the end of the thinking block), mean per query. Queries: every gold type plus 1-3 distractors of the same domain, fixed seed (`free-vs-structured-cot-eval`). vLLM, T 0.6, top-p 0.95, top-k 20, seed 0, up to 12,288 new tokens. Base = the same model without fine-tuning, with the free prompt. **CrossNER** (full test sets) | Set | Base F1 | Free F1 | Structured F1 | Base tokens | Free tokens | Structured tokens | |---|---|---|---|---|---|---| | AI | 0.565 | 0.674 | 0.660 | 699 | 433 | 351 | | Literature | 0.597 | 0.707 | 0.694 | 919 | 495 | 377 | | Music | 0.677 | 0.756 | 0.735 | 849 | 517 | 363 | | Politics | 0.664 | 0.713 | 0.690 | 974 | 568 | 429 | | Science | 0.661 | 0.734 | 0.733 | 826 | 478 | 352 | | **Mean** | **0.633** | **0.717** | **0.703** | **853** | **498** | **374** | **Pile-NER test split** (in-domain, 2,000 queries) | Set | Base F1 | Free F1 | Structured F1 | Base tokens | Free tokens | Structured tokens | |---|---|---|---|---|---|---| | Pile-NER | 0.406 | 0.539 | 0.536 | 1,551 | 1,097 | 724 | **Biomedical benchmarks** (500 sentences each, 465 for JNLPBA-Rare) | Set | Base F1 | Free F1 | Structured F1 | Base tokens | Free tokens | Structured tokens | |---|---|---|---|---|---|---| | NCBI | 0.507 | 0.636 | 0.602 | 431 | 221 | 239 | | BC5CDR | 0.651 | 0.775 | 0.773 | 485 | 287 | 227 | | BC2GM | 0.381 | 0.481 | 0.482 | 534 | 238 | 232 | | BC4CHEMD | 0.448 | 0.576 | 0.572 | 494 | 284 | 307 | | AnatEM | 0.346 | 0.362 | 0.376 | 484 | 251 | 248 | | JNLPBA | 0.506 | 0.539 | 0.516 | 514 | 362 | 286 | | JNLPBA-Rare | 0.448 | 0.429 | 0.415 | 449 | 301 | 226 | | MedMentions-Rare | 0.466 | 0.564 | 0.544 | 587 | 452 | 288 | | BioRED | 0.531 | 0.670 | 0.640 | 656 | 469 | 332 | | **Mean** | **0.476** | **0.559** | **0.547** | **515** | **318** | **265** | All predictions and metrics: `eval-results` (`free-vs-structured-cot/` and `free-vs-structured-cot-bio/`).