feat(model): publie le modèle v2 (20 labels, 96% test acc)

Artefacts entraînés sur FlowRank/labeled_emails@enrichment.
Mise à jour README (versioning v1/v2).

Co-authored-by: Hilarion Lefuneste <hilarionlefuneste@tutamail.com>
Co-authored-by: Cursor <cursoragent@cursor.com>

README.md

@@ -12,6 +12,9 @@ language:
 
 Repo minimal (Python) pour **entraîner, évaluer et publier** un modèle de classification multi-classes d’e-mails à partir du dataset Hugging Face [`FlowRank/labeled_emails`](https://huggingface.co/datasets/FlowRank/labeled_emails).
 
+**v2** (branche `main`, tag `v2.0`) : **20 classes** (catégorie + urgence), entraîné sur `FlowRank/labeled_emails@main`.  
+**v1** (tag `v1.0`) : 10 classes — `revision="v1.0"` pour le dataset et le modèle.
+
 Le script principal est `mailsort.train` (Transformers `Trainer`).
 
 ### Prérequis
@@ -42,6 +45,15 @@ uv run python -m mailsort.train \
   --num-train-epochs 2
 ```
 
+Pour réentraîner sur une révision précise du dataset (ex. avant merge) :
+
+```bash
+uv run python -m mailsort.train \
+  --dataset-id FlowRank/labeled_emails \
+  --dataset-revision enrichment \
+  --num-train-epochs 2
+```
+
 ### Tester / évaluer uniquement
 
 Le script n’a pas (encore) de mode “eval-only”.

model/config.json

@@ -11,29 +11,49 @@
   "eos_token_id": null,
   "hidden_dim": 3072,
   "id2label": {
-    "0": "family",
-    "1": "finance",
-    "2": "games",
-    "3": "human resources",
-    "4": "medical",
-    "5": "pets",
-    "6": "school",
-    "7": "software engineering",
-    "8": "sport",
-    "9": "work/airbus"
+    "0": "family_not_urgent",
+    "1": "family_urgent",
+    "2": "finance_not_urgent",
+    "3": "finance_urgent",
+    "4": "games_not_urgent",
+    "5": "games_urgent",
+    "6": "human resources_not_urgent",
+    "7": "human resources_urgent",
+    "8": "medical_not_urgent",
+    "9": "medical_urgent",
+    "10": "pets_not_urgent",
+    "11": "pets_urgent",
+    "12": "school_not_urgent",
+    "13": "school_urgent",
+    "14": "software engineering_not_urgent",
+    "15": "software engineering_urgent",
+    "16": "sport_not_urgent",
+    "17": "sport_urgent",
+    "18": "work/airbus_not_urgent",
+    "19": "work/airbus_urgent"
   },
   "initializer_range": 0.02,
   "label2id": {
-    "family": 0,
-    "finance": 1,
-    "games": 2,
-    "human resources": 3,
-    "medical": 4,
-    "pets": 5,
-    "school": 6,
-    "software engineering": 7,
-    "sport": 8,
-    "work/airbus": 9
+    "family_not_urgent": 0,
+    "family_urgent": 1,
+    "finance_not_urgent": 2,
+    "finance_urgent": 3,
+    "games_not_urgent": 4,
+    "games_urgent": 5,
+    "human resources_not_urgent": 6,
+    "human resources_urgent": 7,
+    "medical_not_urgent": 8,
+    "medical_urgent": 9,
+    "pets_not_urgent": 10,
+    "pets_urgent": 11,
+    "school_not_urgent": 12,
+    "school_urgent": 13,
+    "software engineering_not_urgent": 14,
+    "software engineering_urgent": 15,
+    "sport_not_urgent": 16,
+    "sport_urgent": 17,
+    "work/airbus_not_urgent": 18,
+    "work/airbus_urgent": 19
   },
   "max_position_embeddings": 512,
   "model_type": "distilbert",

model/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e88532e8b0d0583f7a96aa686f19f0ab2256ee741f6b55773cbb8f36f520c276
-size 267857176
+oid sha256:8bf81d5d4f90a3cf9ebda4fcafe53bf6bfe6bcdb331a34d67044e935d67a7455
+size 267887936

model/tokenizer.json

@@ -1,6 +1,11 @@
 {
   "version": "1.0",
-  "truncation": null,
+  "truncation": {
+    "direction": "Right",
+    "max_length": 256,
+    "strategy": "LongestFirst",
+    "stride": 0
+  },
   "padding": null,
   "added_tokens": [
     {

model/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e5f07cdfe1546cecc7b90ed8a2b923d1de3ed2925d0c1de004f011b09e643764
+oid sha256:e1d931f0b60fe6bac81527b10c5c2ae84c436c292869ee27f3c24890e52d52ad
 size 5265