End of training

README.md

@@ -18,12 +18,12 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [google/rembert](https://huggingface.co/google/rembert) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6768
-- Accuracy: 0.7781
-- 1-f1: 0.1758
-- 1-recall: 0.32
-- 1-precision: 0.1212
-- Balanced Acc: 0.5673
+- Loss: 0.7012
+- Accuracy: 0.4083
+- 1-f1: 0.1667
+- 1-recall: 0.8
+- 1-precision: 0.0930
+- Balanced Acc: 0.5885
 
 ## Model description
 
@@ -48,6 +48,7 @@ The following hyperparameters were used during training:
 - seed: 42
 - optimizer: Use 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_ratio: 0.06
 - num_epochs: 20
 - mixed_precision_training: Native AMP
 
@@ -55,10 +56,9 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1   | 1-recall | 1-precision | Balanced Acc |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|:--------:|:-----------:|:------------:|
-| 0.8084        | 1.0   | 85   | 0.6709          | 0.9260   | 0.0    | 0.0      | 0.0         | 0.5          |
-| 0.6112        | 2.0   | 170  | 0.6704          | 0.9260   | 0.0    | 0.0      | 0.0         | 0.5          |
-| 0.5892        | 3.0   | 255  | 1.0901          | 0.8077   | 0.1558 | 0.24     | 0.1154      | 0.5465       |
-| 0.4492        | 4.0   | 340  | 0.6768          | 0.7781   | 0.1758 | 0.32     | 0.1212      | 0.5673       |
+| 0.7524        | 1.0   | 85   | 0.6789          | 0.9260   | 0.0    | 0.0      | 0.0         | 0.5          |
+| 0.5657        | 2.0   | 170  | 0.8155          | 0.1864   | 0.1486 | 0.96     | 0.0805      | 0.5423       |
+| 0.8312        | 3.0   | 255  | 0.7012          | 0.4083   | 0.1667 | 0.8      | 0.0930      | 0.5885       |
 
 
 ### Framework versions

config.json

@@ -3,14 +3,14 @@
   "architectures": [
     "RemBertForSequenceClassification"
   ],
-  "attention_probs_dropout_prob": 0,
+  "attention_probs_dropout_prob": 0.2,
   "bos_token_id": 312,
   "classifier_dropout_prob": 0.1,
   "embedding_dropout_prob": 0,
   "embedding_size": 256,
   "eos_token_id": 313,
   "hidden_act": "gelu",
-  "hidden_dropout_prob": 0,
+  "hidden_dropout_prob": 0.2,
   "hidden_size": 1152,
   "initializer_range": 0.02,
   "input_embedding_size": 256,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ec3b84d711c43d28b1691c0af3a9cbc8a388c0a2d008fa52ff4af5e4bfd3dee9
+oid sha256:7ce81a1e774f26ff7b7ae0f87271336e9a9606940236ad3498501a35cb672de9
 size 2303755328

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3698bd017fb7aa1392dde8423e7253b50a98ec480c12b3114671766c49303188
+oid sha256:b734c62f3df6549d73ed7f708ec6c4ff00548b7bafc574e400fd30324edb3c18
 size 5432