Upload ToxicLord v1 model

README.md

@@ -1,3 +1,120 @@
----
-license: cc-by-nc-nd-4.0
----
+---
+license: cc-by-nc-nd-4.0
+language:
+  - ru
+library_name: transformers
+pipeline_tag: text-classification
+tags:
+  - toxicity-classification
+  - russian
+  - telegram
+  - moderation
+  - deberta
+base_model: deepvk/deberta-v1-base
+model-index:
+  - name: ToxicLord v1
+    results:
+      - task:
+          type: text-classification
+          name: Text Classification
+        dataset:
+          name: Internal held-out toxicity test split
+          type: private
+        metrics:
+          - type: accuracy
+            value: 0.968937125748503
+            name: Accuracy
+          - type: precision
+            value: 0.9309514251304697
+            name: Toxic precision
+          - type: recall
+            value: 0.905152224824356
+            name: Toxic recall
+          - type: f1
+            value: 0.9178705719374629
+            name: Toxic F1
+          - type: f1
+            value: 0.9493585102268198
+            name: Macro F1
+---
+
+# ToxicLord v1
+
+ToxicLord v1 is a Russian text classification model for chat moderation. It classifies messages as `clean` or `toxic` and is tuned for short Russian Telegram-style messages.
+
+The model is intended for assistive moderation workflows. It can make mistakes and should be used with logging, review, and project-specific thresholds.
+
+## Labels
+
+```text
+0: clean
+1: toxic
+```
+
+## Recommended Threshold
+
+For conservative Telegram moderation, use the toxic probability instead of only argmax:
+
+```text
+toxic if P(toxic) >= 0.90
+```
+
+## Evaluation
+
+Internal held-out test split:
+
+```text
+accuracy:        0.9689
+precision_toxic: 0.9310
+recall_toxic:    0.9052
+f1_toxic:        0.9179
+macro_f1:        0.9494
+```
+
+External fixed benchmark samples at threshold `0.90`:
+
+```text
+Telegram clean chat sample: 2/500 triggered, 0.4% trigger rate
+Toxic sample:               364/500 triggered, 72.8% trigger rate
+```
+
+## Usage
+
+```python
+import torch
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+model_id = "Egor-3926/ToxicLord"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForSequenceClassification.from_pretrained(model_id)
+model.eval()
+
+text = "ты еблан"
+inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=192)
+
+with torch.inference_mode():
+    probs = torch.softmax(model(**inputs).logits, dim=-1)[0]
+
+clean_score = float(probs[0])
+toxic_score = float(probs[1])
+label = "toxic" if toxic_score >= 0.90 else "clean"
+
+print(label, toxic_score)
+```
+
+## Training Data
+
+The model was fine-tuned on a mixture of public Russian toxicity datasets and private moderation annotations/corrections. Raw training data, Telegram logs, user identifiers, and private annotations are not redistributed with this model.
+
+## Limitations
+
+- The model is optimized for Russian Telegram-style moderation and may not transfer well to formal text, long documents, or other languages.
+- Short insults and slurs may be classified as toxic even without broader context.
+- Sarcasm, quotes, jokes, reclaimed language, and moderation discussions can be misclassified.
+- The model should not be used as the only source of truth for irreversible moderation actions.
+
+## License
+
+This model is released under `cc-by-nc-nd-4.0`.
+
+Non-commercial use is allowed with attribution. Commercial use and derivative redistribution are not allowed under this license.

config.json

@@ -0,0 +1,37 @@
+{
+  "architectures": [
+    "DebertaForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "clean",
+    "1": "toxic"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "clean": 0,
+    "toxic": 1
+  },
+  "layer_norm_eps": 1e-07,
+  "legacy": true,
+  "max_position_embeddings": 512,
+  "max_relative_positions": -1,
+  "model_type": "deberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_dropout": 0,
+  "pooler_hidden_act": "gelu",
+  "pooler_hidden_size": 768,
+  "pos_att_type": null,
+  "position_biased_input": true,
+  "relative_attention": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.55.0",
+  "type_vocab_size": 0,
+  "vocab_size": 50265
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90c59a2d41d3a1934dd483fddca5944d04aa5e756c82f0de5d39f2af019705c9
+size 498562144

special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "cls_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "mask_token": {
+    "content": "<mask>",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,62 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "cls_token": "<s>",
+  "eos_token": "</s>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "mask_token": "<mask>",
+  "max_length": 192,
+  "model_max_length": 512,
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "stride": 0,
+  "tokenizer_class": "RobertaTokenizer",
+  "trim_offsets": true,
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "<unk>"
+}