Модели в образе: docker-models/models в Git LFS, CI берёт их из репозитория

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

.gitattributes

@@ -3,3 +3,5 @@
 *.sh text eol=lf
 Dockerfile text eol=lf
 *.py text eol=lf
+docker-models/models/**/*.pt filter=lfs diff=lfs merge=lfs -text
+docker-models/models/**/*.safetensors filter=lfs diff=lfs merge=lfs -text

.gitea/workflows/docker.yml

@@ -27,6 +27,8 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
+        with:
+          lfs: true          # модели в docker-models/models лежат в Git LFS
 
       - name: Проверка синтаксиса кода сервиса
         run: |
@@ -38,7 +40,10 @@ jobs:
           MODELS_URL: ${{ secrets.MODELS_URL }}
           MODELS_TOKEN: ${{ secrets.ACTION_PAT || vars.ACTION_PAT }}
         run: |
-          if [ -n "$MODELS_URL" ]; then
+          if [ -s docker-models/models/best/model.pt ] && [ "$(stat -c %s docker-models/models/best/model.pt)" -gt 1000000 ]; then
+            echo "модели взяты из репозитория (Git LFS)"
+            ls -la docker-models/models/best docker-models/models/segmenter
+          elif [ -n "$MODELS_URL" ]; then
             echo "скачиваю модели"
             if [ -n "$MODELS_TOKEN" ]; then
               curl -fsSL -H "Authorization: token $MODELS_TOKEN" "$MODELS_URL" -o models-bundle.tar.gz

.gitignore

@@ -26,8 +26,6 @@ target/
 data/raw/
 
 # Docker: модели для образа и архив для CI (большие бинарники)
-docker-models/models/*
-!docker-models/models/.gitkeep
 models-bundle.tar.gz
 .env
 

docker-models/models/best/config.json

@@ -0,0 +1,33 @@
+{
+  "add_cross_attention": false,
+  "architectures": [
+    "BertForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": null,
+  "classifier_dropout": null,
+  "directionality": "bidi",
+  "eos_token_id": null,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "is_decoder": false,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_fc_size": 768,
+  "pooler_num_attention_heads": 12,
+  "pooler_num_fc_layers": 3,
+  "pooler_size_per_head": 128,
+  "pooler_type": "first_token_transform",
+  "tie_word_embeddings": true,
+  "transformers_version": "5.21.0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 120138
+}

docker-models/models/best/dub_config.json

@@ -0,0 +1,10 @@
+{
+ "backbone": "ai-forever/ruBert-base",
+ "n_types": 14,
+ "n_bio": 3,
+ "dropout": 0.1,
+ "span_weight": 1.0,
+ "age_weight": 0.5,
+ "n_langs": 3,
+ "lang_weight": 0.3
+}

docker-models/models/best/label_maps.json

@@ -0,0 +1,29 @@
+{
+ "types": [
+  "авт. двухголосый",
+  "авт. многоголосый",
+  "авт. одноголосый",
+  "двухголосый",
+  "люб. двухголосый",
+  "люб. многоголосый",
+  "люб. одноголосый",
+  "многоголосый",
+  "неизвестно",
+  "одноголосый",
+  "проф. двухголосый",
+  "проф. дубляж",
+  "проф. многоголосый",
+  "проф. одноголосый"
+ ],
+ "bio": [
+  "O",
+  "B",
+  "I"
+ ],
+ "langs": [
+  "ENG",
+  "RUS",
+  "UKR"
+ ],
+ "audio_context": true
+}

docker-models/models/best/tokenizer_config.json

@@ -0,0 +1,15 @@
+{
+  "backend": "tokenizers",
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "[MASK]",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "unk_token": "[UNK]"
+}

docker-models/models/segmenter/config.json

@@ -0,0 +1,32 @@
+{
+  "add_cross_attention": false,
+  "architectures": [
+    "BertForSequenceClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": null,
+  "classifier_dropout": null,
+  "dtype": "float32",
+  "emb_size": 312,
+  "eos_token_id": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 312,
+  "initializer_range": 0.02,
+  "intermediate_size": 600,
+  "is_decoder": false,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 2048,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 3,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "problem_type": "single_label_classification",
+  "tie_word_embeddings": true,
+  "transformers_version": "5.21.0",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 83828
+}

docker-models/models/segmenter/tokenizer_config.json

@@ -0,0 +1,24 @@
+{
+  "backend": "tokenizers",
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": false,
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "[MASK]",
+  "max_length": 512,
+  "model_max_length": 2048,
+  "never_split": null,
+  "pad_to_multiple_of": null,
+  "pad_token": "[PAD]",
+  "pad_token_type_id": 0,
+  "padding_side": "right",
+  "sep_token": "[SEP]",
+  "stride": 0,
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "[UNK]"
+}