End of training

README.md

@@ -21,11 +21,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [FacebookAI/xlm-roberta-large](https://huggingface.co/FacebookAI/xlm-roberta-large) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2867
-- Precision: 0.6163
-- Recall: 0.7379
-- F1: 0.6716
-- Accuracy: 0.8883
+- Loss: 0.1297
+- Precision: 0.9233
+- Recall: 0.9608
+- F1: 0.9416
+- Accuracy: 0.9764
 
 ## Model description
 
@@ -44,21 +44,28 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 2e-05
-- train_batch_size: 2
-- eval_batch_size: 2
+- learning_rate: 0.0002
+- train_batch_size: 8
+- eval_batch_size: 8
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 3
+- num_epochs: 10
 
 ### Training results
 
 | Training Loss | Epoch | Step  | Validation Loss | Precision | Recall | F1     | Accuracy |
 |:-------------:|:-----:|:-----:|:---------------:|:---------:|:------:|:------:|:--------:|
-| 0.4493        | 1.0   | 4667  | 0.3927          | 0.5157    | 0.6138 | 0.5605 | 0.8482   |
-| 0.3428        | 2.0   | 9334  | 0.3049          | 0.5957    | 0.7215 | 0.6526 | 0.8799   |
-| 0.3306        | 3.0   | 14001 | 0.2867          | 0.6163    | 0.7379 | 0.6716 | 0.8883   |
+| 0.2578        | 1.0   | 1167  | 0.1724          | 0.8145    | 0.8758 | 0.8440 | 0.9436   |
+| 0.1792        | 2.0   | 2334  | 0.1305          | 0.8636    | 0.9089 | 0.8857 | 0.9603   |
+| 0.1224        | 3.0   | 3501  | 0.1390          | 0.8714    | 0.9398 | 0.9043 | 0.9617   |
+| 0.0894        | 4.0   | 4668  | 0.1201          | 0.8856    | 0.9444 | 0.9140 | 0.9646   |
+| 0.0731        | 5.0   | 5835  | 0.1264          | 0.8848    | 0.9497 | 0.9161 | 0.9696   |
+| 0.0597        | 6.0   | 7002  | 0.1247          | 0.9231    | 0.9516 | 0.9371 | 0.9751   |
+| 0.044         | 7.0   | 8169  | 0.1141          | 0.9177    | 0.9468 | 0.9320 | 0.9749   |
+| 0.0373        | 8.0   | 9336  | 0.1235          | 0.9126    | 0.9597 | 0.9355 | 0.9739   |
+| 0.024         | 9.0   | 10503 | 0.1286          | 0.9226    | 0.9578 | 0.9399 | 0.9762   |
+| 0.0209        | 10.0  | 11670 | 0.1297          | 0.9233    | 0.9608 | 0.9416 | 0.9764   |
 
 
 ### Framework versions

adapter_config.json

@@ -2,7 +2,7 @@
   "alpha_pattern": {},
   "auto_mapping": null,
   "base_model_name_or_path": "FacebookAI/xlm-roberta-large",
-  "bias": "none",
+  "bias": "all",
   "fan_in_fan_out": false,
   "inference_mode": true,
   "init_lora_weights": true,
@@ -10,7 +10,7 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 16,
+  "lora_alpha": 128,
   "lora_dropout": 0.1,
   "megatron_config": null,
   "megatron_core": "megatron.core",
@@ -19,7 +19,7 @@
     "score"
   ],
   "peft_type": "LORA",
-  "r": 8,
+  "r": 64,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:526bff66eb2679a559b8a51b46b0081101899b535bf9ee714d9b183e2a7f74c9
-size 3245876
+oid sha256:1ad0cde9018a9f0fa9431684cd13d379075425548f6ac72332cedf95cab66822
+size 26378372

tokenizer_config.json

@@ -1,4 +1,5 @@
 {
+  "add_prefix_space": true,
   "added_tokens_decoder": {
     "0": {
       "content": "<s>",

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cf7ea53ac963056c404e9f7f98c9127d14267bdb2ef072d868736667034c0093
+oid sha256:902e66f1890b1420baee10b95c917fc78534b08f79fbd9a12a6b1693a427ed45
 size 5176