1. Tokenize the input text into individual words. Vietnamese doesn't have spaces for words like English, so word segmentation is tricky. Wait, but the user provided the input text as a pre-tokenized version, maybe? The example input is "từ đồng nghĩa", which are three words. But in reality, Vietnamese words are separated by spaces. So assuming the input text is properly tokenized with spaces, I can split by spaces.
But the user provided an example where the output is hình đồng nghĩa. The example might not be complete, but it shows the structure. So my task is to take each word in the input text, check if it's a proper noun, and if not, replace it with three synonyms in the specified format.
2. For each word, check if it's a proper noun. Since Vietnamese doesn't use capitalization, this might be hard. Proper nouns can be names (e.g., "Anh", "Minh"), locations ("Hà Nội", "TP.HCM"), companies ("Vingroup"), etc. Maybe the user expects me to leave any word that is a known proper noun untouched. But how to detect that? Without prior knowledge, it's tricky. Maybe the user will include proper nouns in the text, and I just need to leave them as is. So perhaps the instructions are to replace all words unless they are proper nouns, but how to determine that. Wait, but the user provided the input text
Now, the challenge is identifying proper nouns in Vietnamese. Since Vietnamese doesn't capitalize words, proper nouns might be context-based. For example, names of people, locations, brands. If the text mentions "Hà Nội", that's a proper noun. But how to distinguish it from other words? Maybe the user expects me to leave any word that is likely a proper noun as is. So perhaps the user is using proper nouns in the text, and I need to identify and not replace them.
Additionally, the same word might have different sets of synonyms depending on its part of speech. The example uses "đồng nghĩa", a noun, but if it were a verb or adjective, different synonyms might apply. So assuming the input text is properly tokenized
Wait, the user's message here is the instruction, and the assistant is supposed to generate a response when given a text. So in the current scenario, the user has provided the instruction, and the assistant is to process the input text when it's given. However, in this case, the user hasn't included a "Text:" section after "Text:" in the query. The user wrote "Text:" and then "only." So perhaps the user is indicating that after providing the instruction with the example, they will now provide the actual text.
Given these challenges, the assistant should proceed by attempting to find valid synonyms where possible, noting that some replacements might be incorrect, and handling proper nouns by leaving them as is, perhaps making an educated guess based on context. find three synonyms.
Wait, looking back at the user's message: they wrote "modify all words with 3 synonyms formatted v2. Don't touch proper nouns. Text only.
So the user has not included the actual text here. So the assistant's task is to respond with the instructions for when the user will provide the text.
Okay, let's see what the user wants here. They provided a text and want me to modify all the words by replacing them with three synonyms, formatted as v2. Proper nouns should stay the same. First, I need to process each word in the text, check if it's a proper noun. If not, find three synonyms.