Update README.md

README.md

@@ -1,36 +1,107 @@
 ---
-dataset_info:
-  features:
-  - name: text
-    dtype: string
-  - name: level
-    dtype: string
-  - name: domain
-    dtype: string
-  - name: label
-    dtype:
-      class_label:
-        names:
-          '0': A1
-          '1': A2
-          '2': B1
-          '3': B2
-          '4': C1
-          '5': C2
-  splits:
-  - name: train
-    num_bytes: 417880
-    num_examples: 1605
-  - name: test
-    num_bytes: 46604
-    num_examples: 179
-  download_size: 271419
-  dataset_size: 464484
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
-  - split: test
-    path: data/test-*
+language:
+  - en
+license: apache-2.0
+tags:
+  - cefr
+  - english
+  - text-classification
+  - synthetic
+  - language-level
+  - education
+size_categories:
+  - 1K<n<10K
+task_categories:
+  - text-classification
+task_ids:
+  - multi-class-classification
+pretty_name: CEFR English Level Dataset
 ---
+
+# CEFR English Level Dataset
+
+A synthetic dataset of **1,785 English texts** labeled with CEFR proficiency levels (A1 → C2), generated using Groq API (Llama-3.3-70b) with detailed per-level linguistic profiles.
+
+## Dataset Summary
+
+| Split | Samples |
+|-------|---------|
+| Train | 1,605 (90%) |
+| Test  | 180 (10%) |
+| **Total** | **1,785** |
+
+Balanced across 6 CEFR levels — ~298 samples per level.
+
+## CEFR Levels
+
+| Label | Level | Description |
+|-------|-------|-------------|
+| 0 | A1 | Beginner — simple words, short sentences |
+| 1 | A2 | Elementary — basic phrases, familiar topics |
+| 2 | B1 | Intermediate — clear standard language |
+| 3 | B2 | Upper-Intermediate — complex text, abstract topics |
+| 4 | C1 | Advanced — fluent, flexible, precise |
+| 5 | C2 | Mastery — sophisticated, nuanced, idiomatic |
+
+## Domains (10)
+
+`formal email` · `short personal essay` · `informal chat message` · `product review` · `social media post` · `travel diary entry` · `academic paragraph` · `job application paragraph` · `news commentary` · `casual forum reply`
+
+## Dataset Structure
+
+```python
+{
+  "text":   "Despite the challenging circumstances, she managed to articulate...",
+  "level":  "C1",
+  "domain": "academic paragraph",
+  "label":  4   # int 0-5
+}
+```
+
+## Usage
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("yanou16/cefr-dataset")
+
+# Access train split
+for example in ds["train"]:
+    print(example["text"][:80], "→", example["level"])
+```
+
+## Generation Method
+
+Texts were generated via **Groq API** using `llama-3.3-70b-versatile` with detailed per-level linguistic profiles:
+
+- Each level has a specific vocabulary range, grammatical complexity, and discourse features
+- 10 different domain prompts to ensure variety
+- Post-generation validation: word count (5–250 words), deduplication
+- Stratified 90/10 train/test split
+
+## Model trained on this dataset
+
+👉 [yanou16/cefr-english-classifier](https://huggingface.co/yanou16/cefr-english-classifier) — 84.9% accuracy
+
+## Limitations
+
+- **Synthetic data** — generated by LLM, not written by real language learners
+- May not capture authentic learner errors or L1 interference patterns
+- C1/C2 boundary texts are very similar (by design — mirrors human annotator difficulty)
+
+## Citation
+
+```bibtex
+@dataset{louzazna2025cefr,
+  author    = {Louzazna, Rayane},
+  title     = {CEFR English Level Dataset},
+  year      = {2025},
+  publisher = {HuggingFace},
+  url       = {https://huggingface.co/datasets/yanou16/cefr-dataset}
+}
+```
+
+## Author
+
+**Rayane Louzazna** — AI Engineering 
+[HuggingFace](https://huggingface.co/yanou16) · [GitHub](https://github.com/yanou16)