mBERT-B-offensive

README.md

@@ -1,6 +1,6 @@
 ---
-license: mit
-base_model: neuralmind/bert-base-portuguese-cased
+license: apache-2.0
+base_model: google-bert/bert-base-multilingual-cased
 tags:
 - generated_from_trainer
 metrics:
@@ -17,13 +17,13 @@ should probably proofread and complete it, then remove this comment. -->
 
 # content
 
-This model is a fine-tuned version of [neuralmind/bert-base-portuguese-cased](https://huggingface.co/neuralmind/bert-base-portuguese-cased) on an unknown dataset.
+This model is a fine-tuned version of [google-bert/bert-base-multilingual-cased](https://huggingface.co/google-bert/bert-base-multilingual-cased) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7314
-- Accuracy: 0.7625
-- F1-score: 0.7462
-- Recall: 0.8237
-- Precision: 0.6821
+- Loss: 0.4643
+- Accuracy: 0.7959
+- F1-score: 0.7686
+- Recall: 0.8062
+- Precision: 0.7343
 
 ## Model description
 
@@ -54,19 +54,19 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch  | Step | Validation Loss | Accuracy | F1-score | Recall | Precision |
 |:-------------:|:------:|:----:|:---------------:|:--------:|:--------:|:------:|:---------:|
-| 0.5117        | 0.3814 | 500  | 0.4886          | 0.7659   | 0.7709   | 0.8595 | 0.6988    |
-| 0.4755        | 0.7628 | 1000 | 0.4602          | 0.7584   | 0.7561   | 0.8170 | 0.7036    |
-| 0.4107        | 1.1442 | 1500 | 0.5348          | 0.7730   | 0.7774   | 0.8651 | 0.7059    |
-| 0.3685        | 1.5256 | 2000 | 0.4585          | 0.7728   | 0.7755   | 0.8563 | 0.7085    |
-| 0.3652        | 1.9069 | 2500 | 0.4497          | 0.7802   | 0.7733   | 0.8182 | 0.7331    |
-| 0.2919        | 2.2883 | 3000 | 0.5390          | 0.7659   | 0.7561   | 0.7920 | 0.7233    |
-| 0.2614        | 2.6697 | 3500 | 0.5387          | 0.7636   | 0.7647   | 0.8382 | 0.7030    |
-| 0.2518        | 3.0511 | 4000 | 0.6425          | 0.7679   | 0.7411   | 0.7252 | 0.7578    |
-| 0.1791        | 3.4325 | 4500 | 0.6974          | 0.7682   | 0.7478   | 0.7502 | 0.7455    |
-| 0.1803        | 3.8139 | 5000 | 0.6828          | 0.7831   | 0.7744   | 0.8126 | 0.7396    |
-| 0.1531        | 4.1953 | 5500 | 0.8737          | 0.7690   | 0.7439   | 0.7320 | 0.7561    |
-| 0.1267        | 4.5767 | 6000 | 0.9225          | 0.7730   | 0.7555   | 0.7651 | 0.7460    |
-| 0.1344        | 4.9580 | 6500 | 0.9057          | 0.7753   | 0.7573   | 0.7651 | 0.7497    |
+| 0.5842        | 0.3814 | 500  | 0.5475          | 0.7275   | 0.7439   | 0.8704 | 0.6496    |
+| 0.5066        | 0.7628 | 1000 | 0.5066          | 0.7527   | 0.7544   | 0.8351 | 0.6879    |
+| 0.4702        | 1.1442 | 1500 | 0.5164          | 0.7524   | 0.7611   | 0.8672 | 0.6781    |
+| 0.4287        | 1.5256 | 2000 | 0.4908          | 0.7902   | 0.7760   | 0.7992 | 0.7542    |
+| 0.428         | 1.9069 | 2500 | 0.5179          | 0.7553   | 0.7643   | 0.8722 | 0.6801    |
+| 0.368         | 2.2883 | 3000 | 0.5774          | 0.7476   | 0.7377   | 0.7804 | 0.6994    |
+| 0.3507        | 2.6697 | 3500 | 0.5190          | 0.7770   | 0.7784   | 0.8609 | 0.7103    |
+| 0.3285        | 3.0511 | 4000 | 0.6028          | 0.7745   | 0.7684   | 0.8225 | 0.7209    |
+| 0.2697        | 3.4325 | 4500 | 0.5910          | 0.7725   | 0.7745   | 0.8590 | 0.7051    |
+| 0.2697        | 3.8139 | 5000 | 0.5870          | 0.7679   | 0.7554   | 0.7879 | 0.7254    |
+| 0.2274        | 4.1953 | 5500 | 0.7693          | 0.7690   | 0.7558   | 0.7860 | 0.7279    |
+| 0.2076        | 4.5767 | 6000 | 0.7267          | 0.7676   | 0.7535   | 0.7810 | 0.7279    |
+| 0.2057        | 4.9580 | 6500 | 0.7228          | 0.7653   | 0.7494   | 0.7716 | 0.7285    |
 
 
 ### Framework versions

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "neuralmind/bert-base-portuguese-cased",
+  "_name_or_path": "google-bert/bert-base-multilingual-cased",
   "architectures": [
     "BertForSequenceClassification"
   ],
@@ -16,7 +16,6 @@
   "model_type": "bert",
   "num_attention_heads": 12,
   "num_hidden_layers": 12,
-  "output_past": true,
   "pad_token_id": 0,
   "pooler_fc_size": 768,
   "pooler_num_attention_heads": 12,
@@ -29,5 +28,5 @@
   "transformers_version": "4.42.4",
   "type_vocab_size": 2,
   "use_cache": true,
-  "vocab_size": 29794
+  "vocab_size": 119547
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b80012ee25e3913fd5fa531e5edfe7466dd0d0118b661476fb24b668c0d11980
-size 435722224
+oid sha256:c33091dc8209b6f5f8c5ec6e98fd00024b740f822eee53117c73768684cfd1d7
+size 711443456

runs/Jul21_12-17-59_c71e73600a1c/events.out.tfevents.1721564280.c71e73600a1c.2763.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:124a12445c0f59b89d9a591cd5cfea4c344c1d4fdf8e91556246e0f6598a9c22
+size 14348

runs/Jul21_12-17-59_c71e73600a1c/events.out.tfevents.1721568656.c71e73600a1c.2763.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a7822185f1096037d6cfdb6fa7cb7d7d164d90b9162cee40a58cbec80b14f05
+size 1044

tokenizer_config.json

@@ -46,7 +46,7 @@
   "do_basic_tokenize": true,
   "do_lower_case": false,
   "mask_token": "[MASK]",
-  "model_max_length": 1000000000000000019884624838656,
+  "model_max_length": 512,
   "never_split": null,
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79f538af0e34deb2ccfd88dba0d2e783b7ddd8e198dfa37a03c03480a41b2d83
+oid sha256:5dfb155b50fb3b974ed6cb6eef52c91e5ec03b9cd631dada2baa392b95fc4e3c
 size 5112