wsosnowski2 commited on
Commit
cc11df7
·
0 Parent(s):

Add narrative classifier (RoBERTa-large, multi-label) + model card

Browse files
.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,206 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ license: cc-by-nc-4.0
5
+ library_name: transformers
6
+ pipeline_tag: text-classification
7
+ base_model: FacebookAI/roberta-large
8
+ tags:
9
+ - roberta
10
+ - text-classification
11
+ - multi-label-classification
12
+ - disinformation
13
+ - narrative-detection
14
+ - propaganda
15
+ - media-analysis
16
+ metrics:
17
+ - f1
18
+ - precision
19
+ - recall
20
+ ---
21
+
22
+ # Narrative Classifier (RoBERTa-large, multi-label)
23
+
24
+ A multi-label text classifier that detects **disinformation / propaganda narratives** in news and
25
+ social-media text. Given a piece of text, the model predicts which of **41 predefined narratives**
26
+ (spanning topics such as the war in Ukraine, migration, climate change, COVID-19 / vaccines,
27
+ gender & LGBT+, anti-establishment / anti-EU / anti-NATO framings, etc.) are present.
28
+
29
+ The model was developed at the **Polish-Japanese Academy of Information Technology (PJAIT / PJATK)**.
30
+
31
+ - **Architecture:** `RobertaNarrativeModel` — a `roberta-large` encoder + a single linear
32
+ classification head (`narrative_head`, 1024 → 41) applied to the `<s>` (CLS) token.
33
+ - **Task:** multi-label classification (one input can carry several narratives at once).
34
+ - **Base model:** [`FacebookAI/roberta-large`](https://huggingface.co/FacebookAI/roberta-large)
35
+ - **Parameters:** ~0.4B · **Precision:** FP32 · **Format:** safetensors
36
+ - **Language:** English
37
+
38
+ > **Note on the architecture.** This repository uses a *custom* model class
39
+ > (`RobertaNarrativeModel`) whose weights are stored under the `transformer.*` and
40
+ > `narrative_head.*` prefixes. It therefore does **not** load directly with
41
+ > `AutoModelForSequenceClassification`. Use the self-contained loading code in the
42
+ > [How to use](#how-to-use) section below.
43
+
44
+ ## Labels
45
+
46
+ The model outputs 41 labels. The full mapping is in
47
+ [`narrative_labels.json`](./narrative_labels.json) / [`label_config.json`](./label_config.json).
48
+
49
+ <details>
50
+ <summary>Show all 41 narratives</summary>
51
+
52
+ | ID | Narrative |
53
+ |----|-----------|
54
+ | 0 | Abortion is evil/immoral/dangerous |
55
+ | 1 | Alternative treatments are more effective than conventional ones |
56
+ | 2 | Climate change is a hoax |
57
+ | 3 | Collapse of Western civilization is imminent |
58
+ | 4 | Conflict is a staged event prepared by outside forces |
59
+ | 5 | Contraception is against nature/dangerous/immoral |
60
+ | 6 | Conventional medicine is ineffective and corrupt |
61
+ | 7 | Conventional medicine is wrong about the causes of diseases |
62
+ | 8 | Elites manipulate elections |
63
+ | 9 | Elites want to take over the world |
64
+ | 10 | European Union is authoritarian |
65
+ | 11 | Feminism is a tool to destroy the natural order and traditional values |
66
+ | 12 | Global elites deliberately cause pandemics and diseases |
67
+ | 13 | Global warming does not exist/is not a serious threat |
68
+ | 14 | Governments fail to take proper action on migration crisis |
69
+ | 15 | Homosexuals are a threat |
70
+ | 16 | Humanity is not responsible for global warming |
71
+ | 17 | LGBT+ is a tool to destroy the natural order and traditional values |
72
+ | 18 | LGBT+ people are mentally ill |
73
+ | 19 | LGBT+ people are privileged |
74
+ | 20 | Media deliberately spreads lies |
75
+ | 21 | Migrants are a burden on the economy |
76
+ | 22 | Migrants are dangerous |
77
+ | 23 | Migrants are destroying local culture and breaking up local communities |
78
+ | 24 | Migration is a conspiracy of global elites |
79
+ | 25 | Most European countries are puppets of the West |
80
+ | 26 | NATO is authoritarian/warmongering |
81
+ | 27 | Official information is a tool to deceive citizens |
82
+ | 28 | Other |
83
+ | 29 | Russia is strong and winning the war |
84
+ | 30 | Sex education is a threat to children |
85
+ | 31 | Solutions to reduce human impact on environment and climate are a conspiracy |
86
+ | 32 | State and international institutions only serve to oppress citizens. |
87
+ | 33 | The West and their allies are immoral/hostile/ineffective |
88
+ | 34 | The energy crisis is artificially created |
89
+ | 35 | Transgender people are a threat |
90
+ | 36 | Ukraine is an evil, aggressive and dangerous country |
91
+ | 37 | Ukrainian refugees are a danger/burden |
92
+ | 38 | Vaccines are dangerous/ineffective/immoral |
93
+ | 39 | Western elites want to destroy the natural order and traditional values |
94
+ | 40 | other |
95
+
96
+ </details>
97
+
98
+ ## How to use
99
+
100
+ ```python
101
+ import json
102
+ import torch
103
+ from torch import nn
104
+ from transformers import AutoTokenizer, AutoConfig, RobertaModel
105
+ from huggingface_hub import hf_hub_download
106
+ from safetensors.torch import load_file
107
+
108
+ REPO_ID = "pjait/narrative_classifier"
109
+
110
+
111
+ class RobertaNarrativeModel(nn.Module):
112
+ """roberta-large encoder + a linear head over the <s> (CLS) token."""
113
+
114
+ def __init__(self, config, num_labels):
115
+ super().__init__()
116
+ self.transformer = RobertaModel(config, add_pooling_layer=False)
117
+ self.narrative_head = nn.Linear(config.hidden_size, num_labels)
118
+
119
+ def forward(self, input_ids, attention_mask=None):
120
+ out = self.transformer(input_ids=input_ids, attention_mask=attention_mask)
121
+ cls = out.last_hidden_state[:, 0] # <s> token representation
122
+ return self.narrative_head(cls) # raw logits (multi-label)
123
+
124
+
125
+ # --- load config, labels and weights ---------------------------------------
126
+ config = AutoConfig.from_pretrained(REPO_ID)
127
+ tokenizer = AutoTokenizer.from_pretrained(REPO_ID)
128
+
129
+ with open(hf_hub_download(REPO_ID, "narrative_labels.json")) as f:
130
+ labels = json.load(f)
131
+ id2narrative = {int(k): v for k, v in labels["id2narrative"].items()}
132
+ num_labels = labels["num_labels"]
133
+
134
+ model = RobertaNarrativeModel(config, num_labels)
135
+ state_dict = load_file(hf_hub_download(REPO_ID, "model.safetensors"))
136
+ model.load_state_dict(state_dict)
137
+ model.eval()
138
+
139
+ # --- inference --------------------------------------------------------------
140
+ text = "The vaccines were rushed and are far more dangerous than the virus itself."
141
+ inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512)
142
+
143
+ with torch.no_grad():
144
+ logits = model(**inputs)
145
+ probs = torch.sigmoid(logits)[0] # multi-label -> sigmoid
146
+
147
+ THRESHOLD = 0.5
148
+ predicted = [(id2narrative[i], float(p)) for i, p in enumerate(probs) if p >= THRESHOLD]
149
+ print(sorted(predicted, key=lambda x: -x[1]))
150
+ ```
151
+
152
+ `THRESHOLD` controls precision/recall trade-off; tune it on your own validation data.
153
+
154
+ ## Evaluation
155
+
156
+ Metrics from [`metrics.txt`](./metrics.txt) (evaluation split, epoch 3):
157
+
158
+ | Metric | Value |
159
+ |--------|-------|
160
+ | Micro F1 | 0.494 |
161
+ | Macro F1 | 0.185 |
162
+ | Precision | 0.700 |
163
+ | Recall | 0.382 |
164
+ | Subset accuracy | 0.787 |
165
+ | Eval loss | 0.023 |
166
+
167
+ The gap between micro and macro F1, together with high precision but lower recall, indicates the
168
+ model is conservative and performs unevenly across narratives — likely better on
169
+ well-represented narratives and weaker on rare ones. Treat predictions as a **decision-support
170
+ signal**, not ground truth, and calibrate the threshold for your use case.
171
+
172
+ ## Intended use & limitations
173
+
174
+ **Intended use.** Research and analysis of disinformation/propaganda narratives in English-language
175
+ media; content moderation triage; media-monitoring dashboards; academic studies of narrative spread.
176
+
177
+ **Out of scope / cautions.**
178
+ - The model identifies whether text *expresses or discusses* a narrative; it does **not** establish
179
+ truth, intent, or that the author endorses the narrative (quotation, debunking and reporting can
180
+ trigger labels).
181
+ - Trained on English; performance on other languages is not guaranteed.
182
+ - Macro F1 is low — rare narratives are unreliable. Do not use for automated, consequential
183
+ decisions about individuals without human review.
184
+ - Sensitive topics (health, politics, gender, migration). Outputs can reflect biases in the
185
+ training data. Human oversight is required for any deployment.
186
+
187
+ ## Training
188
+
189
+ - **Base model:** `FacebookAI/roberta-large` fine-tuned for multi-label narrative classification.
190
+ - **Epochs:** 3 (see `training_args.bin` for the full `TrainingArguments`).
191
+ - **Objective:** multi-label classification (sigmoid + binary cross-entropy over 41 narratives).
192
+
193
+ ## Citation
194
+
195
+ If you use this model, please cite the Polish-Japanese Academy of Information Technology (PJAIT)
196
+ and the author. *(Add the relevant paper / BibTeX here.)*
197
+
198
+ ```bibtex
199
+ @misc{narrative_classifier_pjait,
200
+ title = {Narrative Classifier (RoBERTa-large, multi-label)},
201
+ author = {Sosnowski, Witold},
202
+ howpublished = {\url{https://huggingface.co/pjait/narrative_classifier}},
203
+ note = {Polish-Japanese Academy of Information Technology (PJAIT)},
204
+ year = {2025}
205
+ }
206
+ ```
config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "roberta-large",
3
+ "architectures": [
4
+ "RobertaNarrativeModel"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 4096,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 514,
17
+ "model_type": "roberta",
18
+ "num_attention_heads": 16,
19
+ "num_hidden_layers": 24,
20
+ "pad_token_id": 1,
21
+ "position_embedding_type": "absolute",
22
+ "torch_dtype": "float32",
23
+ "transformers_version": "4.49.0",
24
+ "type_vocab_size": 1,
25
+ "use_cache": true,
26
+ "vocab_size": 50265
27
+ }
label_config.json ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "narrative2id": {
3
+ "Abortion is evil/immoral/dangerous": 0,
4
+ "Alternative treatments are more effective than conventional ones": 1,
5
+ "Climate change is a hoax": 2,
6
+ "Collapse of Western civilization is imminent": 3,
7
+ "Conflict is a staged event prepared by outside forces": 4,
8
+ "Contraception is against nature/dangerous/immoral": 5,
9
+ "Conventional medicine is ineffective and corrupt": 6,
10
+ "Conventional medicine is wrong about the causes of diseases": 7,
11
+ "Elites manipulate elections": 8,
12
+ "Elites want to take over the world": 9,
13
+ "European Union is authoritarian": 10,
14
+ "Feminism is a tool to destroy the natural order and traditional values": 11,
15
+ "Global elites deliberately cause pandemics and diseases": 12,
16
+ "Global warming does not exist/is not a serious threat": 13,
17
+ "Governments fail to take proper action on migration crisis": 14,
18
+ "Homosexuals are a threat": 15,
19
+ "Humanity is not responsible for global warming": 16,
20
+ "LGBT+ is a tool to destroy the natural order and traditional values": 17,
21
+ "LGBT+ people are mentally ill": 18,
22
+ "LGBT+ people are privileged": 19,
23
+ "Media deliberately spreads lies": 20,
24
+ "Migrants are a burden on the economy": 21,
25
+ "Migrants are dangerous": 22,
26
+ "Migrants are destroying local culture and breaking up local communities": 23,
27
+ "Migration is a conspiracy of global elites": 24,
28
+ "Most European countries are puppets of the West": 25,
29
+ "NATO is authoritarian/warmongering": 26,
30
+ "Official information is a tool to deceive citizens": 27,
31
+ "Other": 28,
32
+ "Russia is strong and winning the war": 29,
33
+ "Sex education is a threat to children": 30,
34
+ "Solutions to reduce human impact on environment and climate are a conspiracy": 31,
35
+ "State and international institutions only serve to oppress citizens.": 32,
36
+ "The West and their allies are immoral/hostile/ineffective": 33,
37
+ "The energy crisis is artificially created": 34,
38
+ "Transgender people are a threat": 35,
39
+ "Ukraine is an evil, aggressive and dangerous country": 36,
40
+ "Ukrainian refugees are a danger/burden": 37,
41
+ "Vaccines are dangerous/ineffective/immoral": 38,
42
+ "Western elites want to destroy the natural order and traditional values": 39,
43
+ "other": 40
44
+ },
45
+ "id2narrative": {
46
+ "0": "Abortion is evil/immoral/dangerous",
47
+ "1": "Alternative treatments are more effective than conventional ones",
48
+ "2": "Climate change is a hoax",
49
+ "3": "Collapse of Western civilization is imminent",
50
+ "4": "Conflict is a staged event prepared by outside forces",
51
+ "5": "Contraception is against nature/dangerous/immoral",
52
+ "6": "Conventional medicine is ineffective and corrupt",
53
+ "7": "Conventional medicine is wrong about the causes of diseases",
54
+ "8": "Elites manipulate elections",
55
+ "9": "Elites want to take over the world",
56
+ "10": "European Union is authoritarian",
57
+ "11": "Feminism is a tool to destroy the natural order and traditional values",
58
+ "12": "Global elites deliberately cause pandemics and diseases",
59
+ "13": "Global warming does not exist/is not a serious threat",
60
+ "14": "Governments fail to take proper action on migration crisis",
61
+ "15": "Homosexuals are a threat",
62
+ "16": "Humanity is not responsible for global warming",
63
+ "17": "LGBT+ is a tool to destroy the natural order and traditional values",
64
+ "18": "LGBT+ people are mentally ill",
65
+ "19": "LGBT+ people are privileged",
66
+ "20": "Media deliberately spreads lies",
67
+ "21": "Migrants are a burden on the economy",
68
+ "22": "Migrants are dangerous",
69
+ "23": "Migrants are destroying local culture and breaking up local communities",
70
+ "24": "Migration is a conspiracy of global elites",
71
+ "25": "Most European countries are puppets of the West",
72
+ "26": "NATO is authoritarian/warmongering",
73
+ "27": "Official information is a tool to deceive citizens",
74
+ "28": "Other",
75
+ "29": "Russia is strong and winning the war",
76
+ "30": "Sex education is a threat to children",
77
+ "31": "Solutions to reduce human impact on environment and climate are a conspiracy",
78
+ "32": "State and international institutions only serve to oppress citizens.",
79
+ "33": "The West and their allies are immoral/hostile/ineffective",
80
+ "34": "The energy crisis is artificially created",
81
+ "35": "Transgender people are a threat",
82
+ "36": "Ukraine is an evil, aggressive and dangerous country",
83
+ "37": "Ukrainian refugees are a danger/burden",
84
+ "38": "Vaccines are dangerous/ineffective/immoral",
85
+ "39": "Western elites want to destroy the natural order and traditional values",
86
+ "40": "other"
87
+ },
88
+ "num_labels": 41
89
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
metrics.txt ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ eval_loss: 0.02342209778726101
2
+ eval_narr_micro_f1: 0.49411764705882355
3
+ eval_narr_macro_f1: 0.18548155605115701
4
+ eval_narr_precision: 0.7
5
+ eval_narr_recall: 0.38181818181818183
6
+ eval_narr_subset_acc: 0.7867924528301887
7
+ eval_runtime: 1.7582
8
+ eval_samples_per_second: 301.446
9
+ eval_steps_per_second: 38.107
10
+ epoch: 3.0
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b961b4689633085e2e358a1f32d692677dac2247a477c2b77dacfc00e34a85c0
3
+ size 1421656876
narrative_labels.json ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "narrative2id": {
3
+ "Abortion is evil/immoral/dangerous": 0,
4
+ "Alternative treatments are more effective than conventional ones": 1,
5
+ "Climate change is a hoax": 2,
6
+ "Collapse of Western civilization is imminent": 3,
7
+ "Conflict is a staged event prepared by outside forces": 4,
8
+ "Contraception is against nature/dangerous/immoral": 5,
9
+ "Conventional medicine is ineffective and corrupt": 6,
10
+ "Conventional medicine is wrong about the causes of diseases": 7,
11
+ "Elites manipulate elections": 8,
12
+ "Elites want to take over the world": 9,
13
+ "European Union is authoritarian": 10,
14
+ "Feminism is a tool to destroy the natural order and traditional values": 11,
15
+ "Global elites deliberately cause pandemics and diseases": 12,
16
+ "Global warming does not exist/is not a serious threat": 13,
17
+ "Governments fail to take proper action on migration crisis": 14,
18
+ "Homosexuals are a threat": 15,
19
+ "Humanity is not responsible for global warming": 16,
20
+ "LGBT+ is a tool to destroy the natural order and traditional values": 17,
21
+ "LGBT+ people are mentally ill": 18,
22
+ "LGBT+ people are privileged": 19,
23
+ "Media deliberately spreads lies": 20,
24
+ "Migrants are a burden on the economy": 21,
25
+ "Migrants are dangerous": 22,
26
+ "Migrants are destroying local culture and breaking up local communities": 23,
27
+ "Migration is a conspiracy of global elites": 24,
28
+ "Most European countries are puppets of the West": 25,
29
+ "NATO is authoritarian/warmongering": 26,
30
+ "Official information is a tool to deceive citizens": 27,
31
+ "Other": 28,
32
+ "Russia is strong and winning the war": 29,
33
+ "Sex education is a threat to children": 30,
34
+ "Solutions to reduce human impact on environment and climate are a conspiracy": 31,
35
+ "State and international institutions only serve to oppress citizens.": 32,
36
+ "The West and their allies are immoral/hostile/ineffective": 33,
37
+ "The energy crisis is artificially created": 34,
38
+ "Transgender people are a threat": 35,
39
+ "Ukraine is an evil, aggressive and dangerous country": 36,
40
+ "Ukrainian refugees are a danger/burden": 37,
41
+ "Vaccines are dangerous/ineffective/immoral": 38,
42
+ "Western elites want to destroy the natural order and traditional values": 39,
43
+ "other": 40
44
+ },
45
+ "id2narrative": {
46
+ "0": "Abortion is evil/immoral/dangerous",
47
+ "1": "Alternative treatments are more effective than conventional ones",
48
+ "2": "Climate change is a hoax",
49
+ "3": "Collapse of Western civilization is imminent",
50
+ "4": "Conflict is a staged event prepared by outside forces",
51
+ "5": "Contraception is against nature/dangerous/immoral",
52
+ "6": "Conventional medicine is ineffective and corrupt",
53
+ "7": "Conventional medicine is wrong about the causes of diseases",
54
+ "8": "Elites manipulate elections",
55
+ "9": "Elites want to take over the world",
56
+ "10": "European Union is authoritarian",
57
+ "11": "Feminism is a tool to destroy the natural order and traditional values",
58
+ "12": "Global elites deliberately cause pandemics and diseases",
59
+ "13": "Global warming does not exist/is not a serious threat",
60
+ "14": "Governments fail to take proper action on migration crisis",
61
+ "15": "Homosexuals are a threat",
62
+ "16": "Humanity is not responsible for global warming",
63
+ "17": "LGBT+ is a tool to destroy the natural order and traditional values",
64
+ "18": "LGBT+ people are mentally ill",
65
+ "19": "LGBT+ people are privileged",
66
+ "20": "Media deliberately spreads lies",
67
+ "21": "Migrants are a burden on the economy",
68
+ "22": "Migrants are dangerous",
69
+ "23": "Migrants are destroying local culture and breaking up local communities",
70
+ "24": "Migration is a conspiracy of global elites",
71
+ "25": "Most European countries are puppets of the West",
72
+ "26": "NATO is authoritarian/warmongering",
73
+ "27": "Official information is a tool to deceive citizens",
74
+ "28": "Other",
75
+ "29": "Russia is strong and winning the war",
76
+ "30": "Sex education is a threat to children",
77
+ "31": "Solutions to reduce human impact on environment and climate are a conspiracy",
78
+ "32": "State and international institutions only serve to oppress citizens.",
79
+ "33": "The West and their allies are immoral/hostile/ineffective",
80
+ "34": "The energy crisis is artificially created",
81
+ "35": "Transgender people are a threat",
82
+ "36": "Ukraine is an evil, aggressive and dangerous country",
83
+ "37": "Ukrainian refugees are a danger/burden",
84
+ "38": "Vaccines are dangerous/ineffective/immoral",
85
+ "39": "Western elites want to destroy the natural order and traditional values",
86
+ "40": "other"
87
+ },
88
+ "num_labels": 41
89
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "cls_token": "<s>",
4
+ "eos_token": "</s>",
5
+ "mask_token": {
6
+ "content": "<mask>",
7
+ "lstrip": true,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "pad_token": "<pad>",
13
+ "sep_token": "</s>",
14
+ "unk_token": "<unk>"
15
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<s>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "<pad>",
14
+ "lstrip": false,
15
+ "normalized": true,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "</s>",
22
+ "lstrip": false,
23
+ "normalized": true,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "3": {
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": true,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "50264": {
37
+ "content": "<mask>",
38
+ "lstrip": true,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ }
44
+ },
45
+ "bos_token": "<s>",
46
+ "clean_up_tokenization_spaces": false,
47
+ "cls_token": "<s>",
48
+ "eos_token": "</s>",
49
+ "errors": "replace",
50
+ "extra_special_tokens": {},
51
+ "mask_token": "<mask>",
52
+ "model_max_length": 512,
53
+ "pad_token": "<pad>",
54
+ "sep_token": "</s>",
55
+ "tokenizer_class": "RobertaTokenizer",
56
+ "trim_offsets": true,
57
+ "unk_token": "<unk>"
58
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7053bb9d7b515628085d2f02e66e1a12fc9a1fcf26ca67ff374b8418d16ffa6f
3
+ size 5304
vocab.json ADDED
The diff for this file is too large to render. See raw diff