delete_me

gistfile1.txt

import base64
import vertexai
from vertexai.generative_models import GenerativeModel, SafetySetting
import vertexai.preview.generative_models as generative_models
import pandas as pd
import os
import time

gcp_project_id =

def generate_and_save():
    vertexai.init(project=gcp_project_id, location="us-central1")
    model = GenerativeModel(
        "gemini-1.5-flash-001",
        system_instruction=[textsi_1]
    )

    responses = model.generate_content(
        '''
        input: kobe crash
        input: ja sex tape
        input: wizards at lakers time
        ''',
        generation_config=generation_config,
        safety_settings=safety_settings,
        stream=True,
    )

    # Open a file to append the JSON responses
    with open("responses_test.json", "a") as file:
        for response in responses:
            print(response.text)
            response_text = response.text
            file.write(response_text + "\n")  # Write each response to the file with a newline separator

# Set up the instruction
textsi_1 = """
You are a professional sports analyst/expert in entity extraction. Your task is to parse search queries to identify and specific entities from diverse sources into a highly efficient & structured format, following detailed guidelines for clarity & completeness.
Given a list of inputs, extract the entities in this format:

If the content is sports-related:

{
 \"o\": kobe and lebron played for the lakers and tom brady for the patriots
  \"s\": true,
  \"p\": [\"lebron james\", \"tom brady\"],
  \"t\": [\"LA Lakers\", \"New England Patriots\"],
  \"l\": [\"NBA\", \"NFL\"],
  \"c\": [\"Phil Jackson\", \"Bill Belichick\"],
  \"r\": [\"three-pointer\", \"touchdown\"]
 \"sp\": [\"basketball\", \"football\"]
}

If the content is not sports-related:

{
  \"s\": false
}

Legend:
o: original input
s: Is the content sports-related? (boolean)
p: Professional players (array of strings)
t: Professional teams (array of strings)
l: Sports leagues (array of strings)
c: Professional coaches (array of strings)
r: Sports terminology (array of strings)
sp: Sport name (array of strings)

Note: Each input will have "input:" and requires its own json output. Output the JSON for each input in the prompt. You must use the full name of the player, team or coach

Extra examples:

input: African Blazers For Ladies
output: {"o": "African Blazers For Ladies", "s": false}

input: ho you fat
output: { "o": ho you fat", "s": true, "p": ["Steeve Ho you Fat], "t": [], "l": [""], "c": [], "r": [""],sp:["basketball" }

"""

# Configuration for LLM generation
generation_config = {
    "max_output_tokens": 8192,
    "temperature": 0.5,
    "top_p": 0.95,
    "top_k": 40
}

safety_settings = [
    SafetySetting(
        category=SafetySetting.HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT,
        threshold=SafetySetting.HarmBlockThreshold.BLOCK_NONE
    ),
    SafetySetting(
        category=SafetySetting.HarmCategory.HARM_CATEGORY_HARASSMENT,
        threshold=SafetySetting.HarmBlockThreshold.BLOCK_NONE  
    ),
    SafetySetting(
        category=SafetySetting.HarmCategory.HARM_CATEGORY_HATE_SPEECH,
        threshold=SafetySetting.HarmBlockThreshold.BLOCK_NONE
    ),
    SafetySetting(
        category=SafetySetting.HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT,
        threshold=SafetySetting.HarmBlockThreshold.BLOCK_NONE
    ),
    SafetySetting(
        category=SafetySetting.HarmCategory.HARM_CATEGORY_UNSPECIFIED,
        threshold=SafetySetting.HarmBlockThreshold.BLOCK_NONE
    )
]

generate_and_save()