| ---
|
| language: [en]
|
| license: mit
|
| tags: [conceptual-metaphor, annotation, lakoff, parallel-corpus, lora, qwen2.5]
|
| base_model: Qwen/Qwen2.5-1.5B-Instruct
|
| ---
|
|
|
| > **⚠️ Terms of use — research only, non-commercial, no redistribution.**
|
| > These models were trained on UM-Corpus sentences (Tian et al., LREC 2014; CC BY-NC-ND 4.0)
|
| > **with written permission from NLP2CT, University of Macau** (on file, 2026-11-11).
|
| > By downloading you agree: (1) academic, non-commercial use only; (2) no redistribution of
|
| > the weights; (3) attribution in any resulting publication to UM-Corpus (NLP2CT, University
|
| > of Macau) and to CM-Annotator. Full provenance: see docs/DATA_CARD.md in the code repo.
|
|
|
| # CM-Annotator (English) — Conceptual Metaphor Annotator
|
|
|
| LoRA adapter that annotates English sentences with conceptual metaphor spans
|
| (Lakoff & Johnson tradition) over a curated closed-set inventory (215 entries, based on the
|
| Master Metaphor List, Lakoff et al. 1991). Outputs strict JSON:
|
| `{"has_metaphor": bool, "metaphor_spans": [{text, cm_id, cm_name, source_domain, target_domain,
|
| conventionality, confidence}]}`. Sentences without conceptual metaphors yield empty spans
|
| (Null-CM negatives are ~30% of training).
|
|
|
| ## Training (distillation)
|
|
|
| - Silver labels distilled from a local LM Studio teacher (qwen3.6-35b-a3b), k=2 self-consistency;
|
| only agreeing labels kept. Details: `docs/TRAINING.md`. Silver labels are NOT gold-human labels.
|
| - Base: `Qwen/Qwen2.5-1.5B-Instruct` · LoRA r=16 α=32 ·
|
| completion-only loss · seed=42 · trained 2026-11-07T18:00:10.229852+00:00
|
|
|
| ## Evaluation (dev vs silver, greedy)
|
|
|
| | metric | value |
|
| |---|---|
|
| | n_eval | 120 |
|
| | JSON valid rate | 1.0 |
|
| | span precision / recall / F1 | 0.0962 / 0.1111 / 0.1031 |
|
|
|
| ## Usage
|
|
|
| Load base + adapter with peft; build prompts exactly as in `service/inference.py`
|
| (system rules + INVENTORY block + SENTENCE). Trained-on data derives from UM-Corpus (CC BY-NC-ND):
|
| training data is not redistributed; reproduce via the pipeline in the repository.
|
|
|
| ## Limitations
|
|
|
| Precision bounded by teacher quality (see audit table in docs/TRAINING.md); recall conservative
|
| on conventional metaphors; confidence not calibrated; Chinese side corpus is mostly simplified script.
|
|
|
