Go Bot: IndexError while training with intent classifier

Hi, I am trying to add intent classifier to Go Bot, but get the error below:

I trained everything on the same datasets: simple slot filter with slotfill_simple_dstc2_raw.json, intent classifier with intents_dstc2_big.json, I use fastText Russian embedder and final Go Bot config is gobot_dstc2.json. Everything worked ok before I added intent classifier. Here is its config

{

“dataset_reader”:
{“class_name”: “main:AssistantDatasetReader”,
“data_path”: “{DATA_PATH}”},
“dataset_iterator”:
{“class_name”: “dialog_iterator”},
“chainer”: {
“in”: [“x”],
“in_y”: [“y”],
“out”: [“y_predicted”],
“pipe”: [
{“class_name”: “deeppavlov.models.go_bot.wrapper:DialogComponentWrapper”,
“component”:
{“class_name”: “split_tokenizer”},
“in”: [“x”],
“out”: [“x_tokens”]},
{“id”: “word_vocab”,
“class_name”: “simple_vocab”,
“fit_on”: [“x_tokens”],
“save_path”: “{MODEL_PATH}/word.dict”,
“load_path”: “{MODEL_PATH}/word.dict”},
{“class_name”: “go_bot”,
“load_path”: “{MODEL_PATH}/model”,
“save_path”: “{MODEL_PATH}/model”,
“in”: [“x”],
“in_y”: [“y”],
“out”: [“y_predicted”],
“main”: true, “debug”: false,
“learning_rate”: 0.003,
“learning_rate_drop_patience”: 5,
“learning_rate_drop_div”: 10.0,
“momentum”: 0.95,
“optimizer”: “tensorflow.train:AdamOptimizer”,
“clip_norm”: 2.0,
“dropout_rate”: 0.4,
“l2_reg_coef”: 0.0003,
“hidden_size”: 128,
“dense_size”: 160,
“word_vocab”: “#word_vocab”,
“database”: null,
“nlg_manager”:
{“class_name”: “gobot_nlg_manager”,
“template_path”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/intent_clf/dstc2-templates.txt”,
“template_type”: “DualTemplate”, “api_call_action”: null},
“api_call_action”: “api_call”,
“use_action_mask”: false,
“slot_filler”:
{“config_path”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/intent_clf/slotfiller_w_intents.json”},
“intent_classifier”:
{“config_path”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/intent_clf/gobot_intent_clf.json”},
“embedder”: {“class_name”: “fasttext”, “load_path”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/wiki.ru.bin”},
“bow_embedder”:
{“class_name”: “bow”,
“depth”: “#word_vocab.len()”,
“with_counts”: true},
“tokenizer”:
{“class_name”: “stream_spacy_tokenizer”,
“lowercase”: false},
“tracker”:
{“class_name”: “featurized_tracker”,
“slot_names”: [“еда”, “цена”, “адресс”]}}]},
“train”: {“epochs”: 200, “batch_size”: 4,
“metrics”: [“per_item_dialog_accuracy”],
“validation_patience”: 10, “val_every_n_batches”: 30,
“log_every_n_batches”: 5, “show_examples”: false,
“evaluation_targets”: [“valid”, “test”],
“class_name”: “nn_trainer”, “max_batches”: 30},
“metadata”:
{“variables”:
{“ROOT_PATH”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/”,
“CONFIGS_PATH”: “{DEEPPAVLOV_PATH}/configs”,
“DOWNLOADS_PATH”: “{ROOT_PATH}/intent_clf/”,
“DATA_PATH”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/intent_clf/”,
“MODELS_PATH”: “{ROOT_PATH}/intent_clf/gobot”,
“MODEL_PATH”: “C:/Users/Shuhratjon.Khalilbek/Downloads/tmp/intent_clf/gobot/”,
“DSTC2_DATA_PATH”: “{DOWNLOADS_PATH}”},
“requirements”: ["{DEEPPAVLOV_PATH}/requirements/tf.txt",
“{DEEPPAVLOV_PATH}/requirements/gensim.txt”, “{DEEPPAVLOV_PATH}/requirements/spacy.txt”,
“{DEEPPAVLOV_PATH}/requirements/en_core_web_sm.txt”],
“download”: [{“url”: “http://files.deeppavlov.ai/deeppavlov_data/gobot_dstc2_v9.tar.gz”,
“subdir”: “{MODELS_PATH}”}, {“url”: “http://files.deeppavlov.ai/embeddings/glove.6B.100d.txt”,
“subdir”: “{DOWNLOADS_PATH}/embeddings”}, {“url”: “http://files.deeppavlov.ai/datasets/dstc2_v3.tar.gz”,
“subdir”: “{DATA_PATH}”}]}}

Should I add anything in config when I add intent classificator?

Thank you

Hi! As said in an answer to another of your inquiries, we’re in the process of upgrading our Go-Bot. DSTC-2 format will be replaced with RASA-compatible DSLs. Currently we partially support their v1 of DSLs (w/o slot filler and intent classifier), and v2 is in the process. W/ v2 RASA DSLs support we will also add integrated support for training both slot filler and intent classifier using RASA DSLs. If you want to ask more questions feel free to join our TG group )