v3: all layers, lr=1e-5, class_weight=3.0, early_stop on loss, F1=0.505

README.md

@@ -0,0 +1,71 @@
+---
+library_name: transformers
+license: cc-by-nc-sa-4.0
+base_model: ufal/robeczech-base
+tags:
+- generated_from_trainer
+metrics:
+- precision
+- recall
+- f1
+model-index:
+- name: robeczech-propaganda
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# robeczech-propaganda
+
+This model is a fine-tuned version of [ufal/robeczech-base](https://huggingface.co/ufal/robeczech-base) on an unknown dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.5046
+- Precision: 0.4419
+- Recall: 0.5808
+- F1: 0.5019
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 1e-05
+- train_batch_size: 32
+- eval_batch_size: 64
+- seed: 42
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 100
+- num_epochs: 8
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     |
+|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|
+| 0.5125        | 1.0   | 299  | 0.4987          | 0.2992    | 0.7773 | 0.4320 |
+| 0.4588        | 2.0   | 598  | 0.4667          | 0.375     | 0.6943 | 0.4870 |
+| 0.4181        | 3.0   | 897  | 0.4481          | 0.4308    | 0.6114 | 0.5054 |
+| 0.3658        | 4.0   | 1196 | 0.4642          | 0.4222    | 0.6157 | 0.5009 |
+| 0.3050        | 5.0   | 1495 | 0.5046          | 0.4419    | 0.5808 | 0.5019 |
+
+
+### Framework versions
+
+- Transformers 5.0.0
+- Pytorch 2.10.0+cu128
+- Datasets 4.0.0
+- Tokenizers 0.22.2

config.json

@@ -0,0 +1,37 @@
+{
+  "add_cross_attention": false,
+  "architectures": [
+    "RobertaForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "dtype": "float32",
+  "eos_token_id": 2,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "not_propaganda",
+    "1": "propaganda"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "is_decoder": false,
+  "label2id": {
+    "not_propaganda": 0,
+    "propaganda": 1
+  },
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.0.0",
+  "type_vocab_size": 1,
+  "use_cache": false,
+  "vocab_size": 51997
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69af442bcb7b084dd9ad1d2d9e91742c9e7a2920686c86fb85693ee8c9c3d8e6
+size 503933504

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:401403987d06fb219f7540318a7def6a62c1525b90ae004a3da3094ca3efb2bb
+size 5201