Upload tokenizer

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer-model/tokenizer.json filter=lfs diff=lfs merge=lfs -text

special_tokens_map.json

@@ -0,0 +1 @@
+{}

tokenizer-model/special_tokens_map.json

@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer-model/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929

tokenizer-model/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003

tokenizer-model/tokenizer_config.json

@@ -0,0 +1,70 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "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
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

tokenizer-retriever/special_tokens_map.json

@@ -0,0 +1,37 @@
+{
+  "cls_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "mask_token": {
+    "content": "[MASK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "[PAD]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": {
+    "content": "[SEP]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer-retriever/tokenizer_config.json

@@ -0,0 +1,57 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[MASK]",
+      "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
+    },
+    "3": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "5": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "never_split": null,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": false,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,12 @@
+{
+  "added_tokens_decoder": {},
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenizer_keeper.KeeperTokenizer",
+      null
+    ]
+  },
+  "clean_up_tokenization_spaces": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "tokenizer_class": "KeeperTokenizer"
+}

tokenizer_keeper.py

@@ -0,0 +1,131 @@
+import os
+import json
+
+import torch
+from transformers import (
+    AutoTokenizer,
+    PreTrainedTokenizer
+)
+
+from configuration_keeper import KeeperConfig
+
+from typing import Optional, List, Union
+
+
+class KeeperTokenizer(PreTrainedTokenizer):
+
+    config_class = KeeperConfig
+
+    def __init__(self, cfg=None):
+
+
+        self.tokenizer_retriever = None
+        self.tokenizer_model = None
+
+        if cfg:
+            print("Initializing KeeperTokenizer with cfg")
+            # Inicialización con configuración
+            self.tokenizer_retriever = AutoTokenizer.from_pretrained(cfg.retriever_config['_name_or_path'])
+            self.tokenizer_model = AutoTokenizer.from_pretrained(cfg.model_config['_name_or_path'])
+
+            # Almacena kwargs para la serialización y carga futura
+            self.init_kwargs = {'cfg': cfg}
+
+            super().__init__()  # Inicializa la clase base al principio
+            print("Initialization complete")
+        else:
+            # Si cfg no se proporciona, esto se manejará en el método from_pretrained
+            print("Initializing KeeperTokenizer without cfg")
+
+
+
+    @classmethod
+    def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
+        # Crea una nueva instancia de KeeperTokenizer sin cfg
+        instance = cls()
+
+        print("Loading tokenizer_retriever from", pretrained_model_name_or_path)
+        instance.tokenizer_retriever = AutoTokenizer.from_pretrained(
+            pretrained_model_name_or_path, subfolder='tokenizer-retriever'
+        )
+
+        print("Loading tokenizer_model from", pretrained_model_name_or_path)
+        instance.tokenizer_model = AutoTokenizer.from_pretrained(
+            pretrained_model_name_or_path, subfolder='tokenizer-model'
+        )
+
+        return instance
+
+    @property
+    def vocab_size(self):
+        # Obtiene los vocabularios de ambos tokenizadores
+        vocab_retriever = self.tokenizer_retriever.get_vocab()
+        vocab_model = self.tokenizer_model.get_vocab()
+        
+        # Combina los vocabularios
+        combined_vocab = {**vocab_retriever, **vocab_model}
+        
+        # Devuelve el tamaño del vocabulario combinado
+        return len(combined_vocab)
+
+
+    def get_vocab(self):
+        # Obtiene los vocabularios de ambos tokenizadores
+        vocab_retriever = self.tokenizer_retriever.get_vocab()
+        vocab_model = self.tokenizer_model.get_vocab()
+
+        # Organiza los vocabularios en un diccionario con claves separadas
+        separated_vocabularies = {
+            'vocab_retriever': vocab_retriever,
+            'vocab_model': vocab_model
+        }
+
+        return separated_vocabularies
+
+    def _tokenize(self, text, **kwargs):
+        # You must implement this method for your tokenization logic
+        pass
+
+    def encode(self, text, **kwargs):
+        tokens_retriever = self.tokenizer_retriever(text, return_tensors='pt', **kwargs)
+        tokens_model = self.tokenizer_model(text, return_tensors='pt', **kwargs)
+
+        return {
+            'tokens_retriever': tokens_retriever,
+            'tokens_model': tokens_model
+        }
+
+    def decode(
+        self,
+        token_ids: Union[int, List[int], "torch.Tensor"],
+        skip_special_tokens: bool = False,
+        **kwargs,
+    ) -> str:
+      return self.tokenizer_model.decode(token_ids, skip_special_tokens, **kwargs)
+
+    def save_vocabulary(self, save_directory, filename_prefix=None):
+        # Asegúrate de que el directorio de salida existe
+        os.makedirs(save_directory, exist_ok=True)
+
+        # Guarda el tokenizador retriever
+        retriever_save_directory = os.path.join(save_directory, "tokenizer-retriever")
+        os.makedirs(retriever_save_directory, exist_ok=True)
+        self.tokenizer_retriever.save_pretrained(retriever_save_directory)
+
+        # Guarda el tokenizador model
+        model_save_directory = os.path.join(save_directory, "tokenizer-model")
+        os.makedirs(model_save_directory, exist_ok=True)
+        self.tokenizer_model.save_pretrained(model_save_directory)
+      
+        # Devuelve los nombres de los archivos guardados (opcional)
+        saved_files = [
+            "tokenizer-retriver/tokenizer_config.json",
+            "tokenizer-retriver/special_tokens_map.json",
+            "tokenizer-retriver/vocab.json",
+            "tokenizer-retriver/added_tokens.json",
+            "tokenizer-model/tokenizer_config.json",
+            "tokenizer-model/special_tokens_map.json",
+            "tokenizer-model/vocab.json",
+            "tokenizer-model/added_tokens.json"
+        ]
+        return tuple(os.path.join(save_directory, file) for file in saved_files)