Update README.md

README.md

@@ -1,16 +1,125 @@
 ---
+library_name: transformers
 license: mit
+base_model: almanach/camembertav2-base
+tags:
+- generated_from_trainer
+- name
+- person
+- company
+metrics:
+- accuracy
+- precision
+- recall
+- f1
+model-index:
+- name: camembertav2-base-name-classifier-v2
+  results: []
 datasets:
 - ele-sage/person-company-names-classification
 language:
 - fr
 - en
-base_model:
-- almanach/camembertav2-base
-new_version: ele-sage/camembertav2-base-name-classifier-v3
-library_name: transformers
-tags:
-- name
-- person
-- company
----
+---
+
+# camembertav2-base-name-classifier-v2
+
+This model is a fine-tuned version of [almanach/camembertav2-base](https://huggingface.co/almanach/camembertav2-base) on [ele-sage/person-company-names-classification](https://huggingface.co/datasets/ele-sage/person-company-names-classification) dataset.
+
+
+It achieves the following results on the evaluation set:
+- Loss: 0.0260
+- Accuracy: 0.9931
+- Precision: 0.9967
+- Recall: 0.9911
+- F1: 0.9939
+
+## Model description
+
+This model is a high-performance binary text classifier, fine-tuned from `camembertav2-base`.
+Its purpose is to distinguish between a **person's name** and a **company/organization name** with high accuracy.
+
+### Direct Use
+
+This model is intended to be used for text classification. Given a string, it will return a label indicating whether the string is a `Person` or a `Company`.
+
+```python
+from transformers import pipeline
+
+classifier = pipeline("text-classification", model="ele-sage/camembertav2-base-name-classifier-v2")
+
+results = classifier([
+    "Satya Nadella",
+    "Global Innovations Inc.",
+    "Martinez, Alonso"
+])
+
+for result in results:
+    print(f"Text: '{result['text']}', Prediction: {result['label']}, Score: {result['score']:.4f}")
+```
+
+### Downstream Use
+
+This model is a key component of a two-stage name processing pipeline. It is designed to be used as a fast, efficient "gatekeeper" to first identify person names before passing them to a more complex parsing model, such as `ele-sage/distilbert-base-uncased-name-splitter`.
+
+### Out-of-Scope Use
+
+- This model is not a general-purpose classifier. It is highly specialized for distinguishing persons from companies and will not perform well on other classification tasks (e.g., sentiment analysis).
+
+## Bias, Risks, and Limitations
+
+- **Geographic & Cultural Bias:** The training data is heavily biased towards North American (Canadian) person names and Quebec-based company names. The model will be less accurate when classifying names from other cultural or geographic origins.
+- **Ambiguity:** Certain names can legitimately be both a person's name and a company's name (e.g., "Ford"). In these cases, the model makes a statistical guess based on its training data, which may not always align with the specific context.
+- **Data Source:** The person name data is derived from a Facebook data leak and contains noise. While a rigorous cleaning process was applied, the model may have learned from some spurious data.
+
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 1e-05
+- train_batch_size: 128
+- eval_batch_size: 128
+- seed: 42
+- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 1000
+- num_epochs: 1
+
+### Training results
+
+| Training Loss | Epoch  | Step  | Validation Loss | Accuracy | Precision | Recall | F1     |
+|:-------------:|:------:|:-----:|:---------------:|:--------:|:---------:|:------:|:------:|
+| 0.0419        | 0.0411 | 2000  | 0.0383          | 0.9902   | 0.9954    | 0.9873 | 0.9913 |
+| 0.0333        | 0.0821 | 4000  | 0.0351          | 0.9912   | 0.9950    | 0.9895 | 0.9922 |
+| 0.0354        | 0.1232 | 6000  | 0.0340          | 0.9911   | 0.9975    | 0.9869 | 0.9921 |
+| 0.0316        | 0.1642 | 8000  | 0.0321          | 0.9918   | 0.9957    | 0.9897 | 0.9927 |
+| 0.0325        | 0.2053 | 10000 | 0.0299          | 0.9918   | 0.9947    | 0.9908 | 0.9928 |
+| 0.0304        | 0.2464 | 12000 | 0.0301          | 0.9920   | 0.9951    | 0.9908 | 0.9929 |
+| 0.0288        | 0.2874 | 14000 | 0.0301          | 0.9921   | 0.9959    | 0.9902 | 0.9930 |
+| 0.0329        | 0.3285 | 16000 | 0.0283          | 0.9923   | 0.9957    | 0.9907 | 0.9932 |
+| 0.0314        | 0.3696 | 18000 | 0.0276          | 0.9925   | 0.9960    | 0.9907 | 0.9933 |
+| 0.0277        | 0.4106 | 20000 | 0.0277          | 0.9926   | 0.9964    | 0.9905 | 0.9935 |
+| 0.0318        | 0.4517 | 22000 | 0.0279          | 0.9926   | 0.9968    | 0.9902 | 0.9935 |
+| 0.0246        | 0.4927 | 24000 | 0.0284          | 0.9927   | 0.9963    | 0.9908 | 0.9936 |
+| 0.0294        | 0.5338 | 26000 | 0.0276          | 0.9927   | 0.9966    | 0.9904 | 0.9935 |
+| 0.0304        | 0.5749 | 28000 | 0.0275          | 0.9925   | 0.9950    | 0.9918 | 0.9934 |
+| 0.0283        | 0.6159 | 30000 | 0.0268          | 0.9928   | 0.9968    | 0.9904 | 0.9936 |
+| 0.0304        | 0.6570 | 32000 | 0.0276          | 0.9928   | 0.9969    | 0.9904 | 0.9936 |
+| 0.0295        | 0.6981 | 34000 | 0.0274          | 0.9929   | 0.9963    | 0.9912 | 0.9937 |
+| 0.0266        | 0.7391 | 36000 | 0.0271          | 0.9930   | 0.9964    | 0.9912 | 0.9938 |
+| 0.0271        | 0.7802 | 38000 | 0.0270          | 0.9929   | 0.9971    | 0.9904 | 0.9937 |
+| 0.0277        | 0.8212 | 40000 | 0.0266          | 0.9930   | 0.9968    | 0.9909 | 0.9938 |
+| 0.027         | 0.8623 | 42000 | 0.0265          | 0.9931   | 0.9969    | 0.9908 | 0.9939 |
+| 0.0286        | 0.9034 | 44000 | 0.0262          | 0.9931   | 0.9966    | 0.9912 | 0.9939 |
+| 0.0275        | 0.9444 | 46000 | 0.0262          | 0.9931   | 0.9967    | 0.9912 | 0.9939 |
+| 0.0287        | 0.9855 | 48000 | 0.0260          | 0.9931   | 0.9967    | 0.9911 | 0.9939 |
+
+
+### Framework versions
+
+- Transformers 4.57.1
+- Pytorch 2.9.0+cu128
+- Datasets 4.4.1
+- Tokenizers 0.22.1