synapti
/

nci-technique-classifier-v2

@@ -1,168 +1,90 @@
 ---
-license: apache-2.0
-language:
-- en
-tags:
-- text-classification
-- propaganda-detection
-- multi-label
-- modernbert
-datasets:
-- synapti/nci-propaganda-production
-- synapti/nci-synthetic-articles
-metrics:
-- f1
-- precision
-- recall
-pipeline_tag: text-classification
 library_name: transformers
 base_model: answerdotai/ModernBERT-base
 ---
-# NCI Technique Classifier v2
-**Multi-label propaganda technique classifier** trained on the NCI (Neural Counter-Intelligence) Protocol dataset.
-## Model Description
-This model detects **18 propaganda techniques** in text using a multi-label classification approach. It is designed to work as Stage 2 in a two-stage pipeline:
-1. **Stage 1**: Binary detection (is there propaganda?) using `synapti/nci-binary-detector`
-2. **Stage 2**: Technique classification (what techniques are used?) using this model
-### Supported Techniques
-| Technique | Description |
-|-----------|-------------|
-| `Loaded_Language` | Words/phrases with strong emotional implications |
-| `Appeal_to_fear-prejudice` | Building support by exploiting fear |
-| `Exaggeration,Minimisation` | Making something more/less important than it is |
-| `Repetition` | Repeating the same message over and over |
-| `Flag-Waving` | Playing on national/group identity |
-| `Name_Calling,Labeling` | Attacking through labels rather than arguments |
-| `Reductio_ad_hitlerum` | Persuading by comparing to disliked groups |
-| `Black-and-White_Fallacy` | Presenting only two choices |
-| `Causal_Oversimplification` | Assuming single cause for complex issue |
-| `Whataboutism,Straw_Men,Red_Herring` | Deflection and misdirection |
-| `Straw_Man` | Misrepresenting someone's argument |
-| `Red_Herring` | Introducing irrelevant topics |
-| `Doubt` | Questioning credibility without evidence |
-| `Appeal_to_Authority` | Relying on authority rather than evidence |
-| `Thought-terminating_Cliches` | Phrases that discourage critical thought |
-| `Bandwagon` | Appeals to popularity |
-| `Slogans` | Brief, memorable phrases |
-| `Obfuscation,Intentional_Vagueness,Confusion` | Deliberately unclear language |
-## Usage
-```python
-from transformers import AutoModelForSequenceClassification, AutoTokenizer
-import torch
-# Load model and tokenizer
-model = AutoModelForSequenceClassification.from_pretrained(
-    "synapti/nci-technique-classifier-v2",
-    trust_remote_code=True
-)
-tokenizer = AutoTokenizer.from_pretrained(
-    "synapti/nci-technique-classifier-v2",
-    trust_remote_code=True
-)
-# Prepare input
-text = "Wake up, patriots! The radical elites are destroying our country!"
-inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
-# Get predictions
-with torch.no_grad():
-    outputs = model(**inputs)
-    probs = torch.sigmoid(outputs.logits)[0]
-# Get technique labels
-id2label = model.config.id2label
-threshold = 0.3
-# Print detected techniques
-for idx, prob in enumerate(probs):
-    if prob.item() >= threshold:
-        technique = id2label[str(idx)]
-        print(f"{technique}: {prob.item():.1%}")
-```
-### Two-Stage Pipeline Usage
-```python
-from nci.transformers.two_stage_pipeline import TwoStagePipeline
-# Load two-stage pipeline
-pipeline = TwoStagePipeline.from_pretrained(
-    binary_model="synapti/nci-binary-detector",
-    technique_model="synapti/nci-technique-classifier-v2",
-)
-# Analyze text
-result = pipeline.analyze("Some text to analyze...")
-print(f"Has propaganda: {result.has_propaganda}")
-print(f"Confidence: {result.propaganda_confidence:.1%}")
-print(f"Detected techniques: {result.detected_techniques}")
-```
-## Training Details
-### Training Data
-- **Primary**: [synapti/nci-propaganda-production](https://huggingface.co/datasets/synapti/nci-propaganda-production) (11,573 samples)
-- **Augmentation**: [synapti/nci-synthetic-articles](https://huggingface.co/datasets/synapti/nci-synthetic-articles) (~5,485 synthetic article-length samples)
-- **Total**: ~17,000 training samples
-### Training Procedure
-- **Base model**: [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base)
-- **Fine-tuning**: HuggingFace AutoTrain on A100 GPU
-- **Epochs**: 3
-- **Batch size**: 16
-- **Learning rate**: 2e-5
-- **Loss function**: Focal Loss (gamma=2) for class imbalance handling
-### Performance Metrics
-**Test Set Performance:**
-| Metric | Score |
-|--------|-------|
-| **Micro F1** | 80.1% |
-| **Macro F1** | 51.2% |
-**Top Performing Techniques:**
-| Technique | F1 Score |
-|-----------|----------|
-| Loaded_Language | 97.0% |
-| Appeal_to_fear-prejudice | 89.7% |
-| Name_Calling,Labeling | 81.8% |
-| Exaggeration,Minimisation | 75.4% |
-## Limitations
-- Trained primarily on English text
-- Performance varies by technique (common techniques perform better)
-- Best used as Stage 2 after binary detection for efficient inference
-- Requires `trust_remote_code=True` for ModernBERT architecture
-## Citation
-If you use this model, please cite:
-```bibtex
-@misc{nci-technique-classifier-v2,
-  title={NCI Technique Classifier v2: Multi-label Propaganda Detection},
-  author={Synapti},
-  year={2024},
-  publisher={Hugging Face},
-  url={https://huggingface.co/synapti/nci-technique-classifier-v2}
-}
-```
-## License
-Apache 2.0

 ---
 library_name: transformers
+license: apache-2.0
 base_model: answerdotai/ModernBERT-base
+tags:
+- generated_from_trainer
+model-index:
+- name: nci-technique-classifier-v2
+  results: []
 ---
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+# nci-technique-classifier-v2
+This model is a fine-tuned version of [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) on an unknown dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.0233
+- Micro F1: 0.8017
+- Macro F1: 0.6272
+- Micro Precision: 0.8311
+- Micro Recall: 0.7743
+## Model description
+More information needed
+## Intended uses & limitations
+More information needed
+## Training and evaluation data
+More information needed
+## Training procedure
+### Training hyperparameters
+The following hyperparameters were used during training:
+- learning_rate: 2e-05
+- train_batch_size: 16
+- eval_batch_size: 32
+- seed: 42
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 5
+- mixed_precision_training: Native AMP
+### Training results
+| Training Loss | Epoch  | Step | Validation Loss | Micro F1 | Macro F1 | Micro Precision | Micro Recall |
+|:-------------:|:------:|:----:|:---------------:|:--------:|:--------:|:---------------:|:------------:|
+| No log        | 0.1634 | 200  | 0.0350          | 0.6311   | 0.1526   | 0.7644          | 0.5373       |
+| No log        | 0.3268 | 400  | 0.0305          | 0.6658   | 0.1814   | 0.8020          | 0.5692       |
+| 0.0552        | 0.4902 | 600  | 0.0282          | 0.7023   | 0.2044   | 0.8244          | 0.6117       |
+| 0.0552        | 0.6536 | 800  | 0.0263          | 0.7268   | 0.2181   | 0.8509          | 0.6343       |
+| 0.0273        | 0.8170 | 1000 | 0.0256          | 0.7497   | 0.2610   | 0.8305          | 0.6832       |
+| 0.0273        | 0.9804 | 1200 | 0.0249          | 0.7462   | 0.2371   | 0.8740          | 0.6510       |
+| 0.0273        | 1.1438 | 1400 | 0.0245          | 0.7626   | 0.2862   | 0.8450          | 0.6949       |
+| 0.0231        | 1.3072 | 1600 | 0.0242          | 0.7583   | 0.2371   | 0.8582          | 0.6793       |
+| 0.0231        | 1.4706 | 1800 | 0.0238          | 0.7650   | 0.3155   | 0.8457          | 0.6984       |
+| 0.0226        | 1.6340 | 2000 | 0.0238          | 0.7624   | 0.3074   | 0.8542          | 0.6885       |
+| 0.0226        | 1.7974 | 2200 | 0.0230          | 0.7626   | 0.3634   | 0.8681          | 0.68         |
+| 0.0226        | 1.9608 | 2400 | 0.0223          | 0.7747   | 0.4246   | 0.8675          | 0.6998       |
+| 0.0214        | 2.1242 | 2600 | 0.0225          | 0.7731   | 0.4412   | 0.8752          | 0.6924       |
+| 0.0214        | 2.2876 | 2800 | 0.0221          | 0.7775   | 0.4101   | 0.8733          | 0.7005       |
+| 0.0189        | 2.4510 | 3000 | 0.0219          | 0.7819   | 0.4757   | 0.8414          | 0.7303       |
+| 0.0189        | 2.6144 | 3200 | 0.0224          | 0.7796   | 0.4224   | 0.8606          | 0.7126       |
+| 0.0189        | 2.7778 | 3400 | 0.0217          | 0.7922   | 0.5512   | 0.8389          | 0.7504       |
+| 0.0187        | 2.9412 | 3600 | 0.0217          | 0.7813   | 0.4680   | 0.8610          | 0.7150       |
+| 0.0187        | 3.1046 | 3800 | 0.0224          | 0.7912   | 0.5458   | 0.8341          | 0.7526       |
+| 0.0155        | 3.2680 | 4000 | 0.0231          | 0.7922   | 0.5455   | 0.8475          | 0.7437       |
+| 0.0155        | 3.4314 | 4200 | 0.0231          | 0.7996   | 0.5843   | 0.8295          | 0.7717       |
+| 0.0155        | 3.5948 | 4400 | 0.0223          | 0.8004   | 0.5706   | 0.8398          | 0.7646       |
+| 0.0148        | 3.7582 | 4600 | 0.0228          | 0.8096   | 0.6067   | 0.8527          | 0.7706       |
+| 0.0148        | 3.9216 | 4800 | 0.0229          | 0.8135   | 0.6228   | 0.8457          | 0.7837       |
+| 0.0126        | 4.0850 | 5000 | 0.0255          | 0.8095   | 0.6251   | 0.8379          | 0.7830       |
+| 0.0126        | 4.2484 | 5200 | 0.0267          | 0.8061   | 0.6223   | 0.8325          | 0.7812       |
+| 0.0126        | 4.4118 | 5400 | 0.0261          | 0.8081   | 0.6338   | 0.8372          | 0.7809       |
+### Framework versions
+- Transformers 4.57.3
+- Pytorch 2.9.1+cu128
+- Datasets 4.4.1
+- Tokenizers 0.22.1

calibration_config.json ADDED Viewed

	@@ -0,0 +1,43 @@

+{
+  "temperature": 1.0,
+  "thresholds": {
+    "Loaded_Language": 0.5,
+    "Appeal_to_fear-prejudice": 0.5,
+    "Exaggeration,Minimisation": 0.5,
+    "Repetition": 0.5,
+    "Flag-Waving": 0.5,
+    "Name_Calling,Labeling": 0.5,
+    "Reductio_ad_hitlerum": 0.5,
+    "Black-and-White_Fallacy": 0.5,
+    "Causal_Oversimplification": 0.5,
+    "Whataboutism,Straw_Men,Red_Herring": 0.5,
+    "Straw_Man": 0.5,
+    "Red_Herring": 0.5,
+    "Doubt": 0.5,
+    "Appeal_to_Authority": 0.5,
+    "Thought-terminating_Cliches": 0.5,
+    "Bandwagon": 0.5,
+    "Slogans": 0.5,
+    "Obfuscation,Intentional_Vagueness,Confusion": 0.5
+  },
+  "technique_labels": [
+    "Loaded_Language",
+    "Appeal_to_fear-prejudice",
+    "Exaggeration,Minimisation",
+    "Repetition",
+    "Flag-Waving",
+    "Name_Calling,Labeling",
+    "Reductio_ad_hitlerum",
+    "Black-and-White_Fallacy",
+    "Causal_Oversimplification",
+    "Whataboutism,Straw_Men,Red_Herring",
+    "Straw_Man",
+    "Red_Herring",
+    "Doubt",
+    "Appeal_to_Authority",
+    "Thought-terminating_Cliches",
+    "Bandwagon",
+    "Slogans",
+    "Obfuscation,Intentional_Vagueness,Confusion"
+  ]
+}

model.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b352010ce6054ba31b79158ecba01272ef2c8b0cf8f35c7ee9ad0bfa4774724c
 size 598489008

 version https://git-lfs.github.com/spec/v1
+oid sha256:3f8a1177759ab1d9f67ee8d2de26fdb6bda5eaa6e382125e17f7a494388d838d
 size 598489008