This model is a SpanMarker model fine-tuned for fine-grained Named Entity Recognition (NER) in the climate change research domain, extracting 28 distinct entity types. It utilizes the domain-specific P0L3/clirebert_clirevocab_uncased as the underlying encoder.

📌 Model Details

Model Type: SpanMarker
Encoder: P0L3/clirebert_clirevocab_uncased
Maximum Sequence Length: 512 tokens
Maximum Entity Length: 14 words
Language: English
License: cc-by-sa-4.0

Model Labels

Label	Examples
Asset	"water resources", "raw material", "mental health"
Body Part	"deep tissue compartment", "leaves", "plant leaves"
Body of Water	"rivers", "peripheral rivers", "Dhaleshwari river"
Chemical	"domoic acid", "marine algal toxin", "cathode materials"
Disease	"chronic epileptic syndrome", "seizures", "acute neurologic signs"
Ecosystem	"polluted environment", "Tropical montane cloud forest", "cloud forests"
Energy Source	"battery cells", "12-cell series battery-pack prototype", "fossil fuels"
Field of Study	"reference laboratory", "veterinary medicine", "study"
Geographical Feature	"heterogenous topography", "mountainous regions", "low point"
Intellectual Artefact	"data", "Veterinary medical records", "Daily husbandry records"
Location	"Westbrook", "wild", "beaches"
Mathematical Expression	"Stepwise machine hour constraints", "gradient", "difference"
Measuring Device	"MRI scan", "station", "EEG"
Meteorological Phenomenon	"climate change", "climatic variability", "rainfall"
Method	"serum monitoring", "clinical efficacy", "dosing"
Natural Disaster	"seasonal air pollution", "environmental pollution", "heavy metal contamination"
Natural Phenomenon	"changing ocean conditions", "algal blooms", "biochemical changes"
Organism	"species", "Zalophus californianus", "California sea lions"
Organization	"long-term care facility", "reference laboratory", "NOAA National Marine Fisheries Service"
Other	"marine mammal health", "reports", "normal eating"
Person	"staff", "clinicians", "Clinicians"
Physical Artefact	"EVs", "paved east – west road", "electric vehicle"
Physical Phenomenon	"structural abnormalities", "normal food intake", "seasonal changes"
Policy	"energy security", "pollution", "safety"
Quantity	"200 mAhg − 1", ">", "energy density"
Satellite	"satellites", "Tropical Rainfall Measuring Mission", "TRMM"
System	"system structure", "climate", "global overturning circulation"
Time Period	"101 days", "several decades", "periods of prolonged anorexia"

🚀 Main Results (Selected Checkpoint)

This repository provides the best-performing checkpoint selected from 5 runs with different random seeds. While the internal training logs tracked performance on the validation split of CliReNER_silver, the final model selection and the metrics below are evaluated on the independent, expert-annotated CliReNER_gold dataset.

Metric	Score
Precision	XX.XX
Recall	XX.XX
F1	XX.XX

This checkpoint corresponds to the seed with the highest strict F1 on the gold evaluation set.

📊 Results Across Seeds

We fine-tuned the model using 5 different random seeds to assess the stability and robustness of the architecture on the domain-specific text.

Seed	Precision	Recall	Strict F1
1	XX.XX	XX.XX	XX.XX
2	XX.XX	XX.XX	XX.XX
3	XX.XX	XX.XX	XX.XX
4	XX.XX	XX.XX	XX.XX
5	XX.XX	XX.XX	XX.XX

Summary:

F1: mean = XX.XX, std = XX.XX
Precision: mean = XX.XX, std = XX.XX
Recall: mean = XX.XX, std = XX.XX

Model Selection Strategy: The uploaded checkpoint is the single best seed (highest strict F1 on the gold dataset), ensuring strong real-world performance and high-fidelity alignment with domain-expert consensus.

📂 Dataset & Evaluation

Training Dataset: CliReNER_silver
- Splits used: Stratified 80:10:10 ratio (Train/Validation/Test). The 80% split was used for training.
Evaluation Dataset: CliReNER_gold
- Splits used: Evaluated on the combined 192 sentences (expert-annotated via Weighted Expert Voting).
Preprocessing:
- Texts were tokenized using the WordPiece tokenizer corresponding to the CliREBERT encoder.
- The dataset utilizes a flat NER schema (nested entities are excluded, and overlapping entities are resolved to the most relevant span).
Metric Details:
- F1 type: Strict F1 (Entity-level exact match).
- Evaluation was performed ensuring entities match both the exact boundary span and the exact semantic label to be considered correct.

⚖️ Precision vs Recall Behavior

(Note to author: Describe the model’s tendency here based on your results. Example: "The model slightly favors precision over recall")

⚙️ Usage

Direct Use for Inference

Because this model was trained using the SpanMarker framework, it requires the span_marker library for inference.

pip install span_marker

python

from span_marker import SpanMarkerModel

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("P0L3/CliReNER-clirebert_clirevocab_uncased")

# Run inference
text = "The urgency of understanding complex, multi-scalar climate phenomena has led to rapid growth in research output."
entities = model.predict(text)

for entity in entities:
    print(f"Entity: {entity['span']} | Label: {entity['label']} | Score: {entity['score']:.4f}")

# Entity: multi-scalar | Label: Other | Score: 0.7188
# Entity: climate phenomena | Label: Meteorological Phenomenon | Score: 0.3017
# Entity: research output | Label: Other | Score: 0.5069

Downstream Use

You can easily continue fine-tuning this model on your own dataset.

Click to expand

python

from span_marker import SpanMarkerModel, Trainer
from datasets import load_dataset

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("your-huggingface-username/your-model-name")

# Specify a Dataset with "tokens" and "ner_tags" columns
dataset = load_dataset("your_custom_dataset")

# Initialize a Trainer using the pretrained model & dataset
trainer = Trainer(
    model=model,
    train_dataset=dataset["train"],
    eval_dataset=dataset["validation"],
)
trainer.train()
trainer.save_model("span_marker_model_id-finetuned")

📉 Training Details

Training Set Metrics

Training set	Min	Median	Max
Sentence length	3	31.4819	97
Entities per sentence	1	7.0100	22

Training Hyperparameters

learning_rate: 5e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42 (Note: This log corresponds to the specific seed run saved during training)
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: adamw_torch with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 20

Training Results (CliReNER_silver Validation Split)

Epoch	Step	Validation Loss	Validation Precision	Validation Recall	Validation F1	Validation Accuracy
1.0	62	0.1148	1.0	0.0014	0.0029	0.6081
2.0	124	0.0549	0.6834	0.3128	0.4291	0.7456
3.0	186	0.0435	0.6681	0.4534	0.5402	0.8038
4.0	248	0.0414	0.6655	0.5337	0.5924	0.8137
5.0	310	0.0486	0.6141	0.5638	0.5879	0.8213
6.0	372	0.0481	0.6406	0.5882	0.6133	0.8274
7.0	434	0.0485	0.6317	0.5882	0.6092	0.8328

Framework Versions

Python: 3.10.19
SpanMarker: 1.7.0
Transformers: 4.50.0
PyTorch: 2.9.1
Datasets: 3.0.0
Tokenizers: 0.21.4

📚 Citation

If you use this model or the CliReNER datasets in your research, please cite the project:

latex

@misc{poleksic2026named,
  author       = {Poleksić, Andrija and Martinčić-Ipšić, Sanda},
  title        = {Named Entity Recognition for Climate Change Research},
  year         = {2026},
  howpublished = {Research Square},
  note         = {Preprint}
}

Please also acknowledge the SpanMarker framework: