jgoodie

import torch
import numpy as np

import seaborn as sns
import matplotlib.pyplot as plt

from transformers import AutoTokenizer, AutoModel

# Load pre-trained tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased", clean_up_tokenization_spaces=False)

jgoodie

import scipy
import numpy as np
from sklearn.preprocessing import OneHotEncoder


sentence = "the otter swam across the river to the other bank"
d = dict.fromkeys(sentence.split())
vocab = list(d.keys())

tokens = sentence.lower().split()

jgoodie

import scipy
import numpy as np
from sklearn.preprocessing import OneHotEncoder


sentence = "the otter swam across the river to the other bank"
d = dict.fromkeys(sentence.split())
vocab = list(d.keys())

tokens = sentence.lower().split()

jgoodie

import numpy as np

def single_head_attention(X, beta_q, beta_k, beta_v, omega_q, omega_k, omega_v):
    query = beta_q + omega_q@X
    key = beta_k + omega_k@X
    value = beta_v + omega_v@X
    
    dp = np.dot(key.T, query) 
    scaled_dp = dp/np.sqrt(query.shape[0])
    attention_weights = scipy.special.softmax(scaled_dp, axis=0)

jgoodie

import numpy as np
from sklearn.preprocessing import OneHotEncoder


sentence = "the otter swam across the river to the other bank"
d = dict.fromkeys(sentence.split())
vocab = list(d.keys())

tokens = sentence.lower().split()
encoder = OneHotEncoder(categories=[vocab], sparse_output=False)

jgoodie

tavily_builder = StateGraph(TavilyState)
tavily_builder.add_node("TavilySearch", TavilySearch)
tavily_builder.add_node("TavilySummary", TavilySummary)

tavily_builder.add_edge(START, "TavilySearch")
tavily_builder.add_edge("TavilySearch", "TavilySummary")
tavily_builder.add_edge("TavilySummary", END)

graph = tavily_builder.compile()
display(Image(graph.get_graph().draw_mermaid_png()))

jgoodie

def TavilySummary(state):
    # Get state
    context = state["context"]
    question = state["question"]

    # Template
    answer_template = """
    You are a competitive research analytist helping a team of product managers conduct competitive market research. 
    Answer the research question:

jgoodie

def TavilySearch(state):
    
    """ Retrieve docs from web search """

    # Search
    tavily_search = TavilySearchResults(max_results=5)
    search_docs = tavily_search.invoke(state['question'])
    search_docs = [get_news_article_text(d['url']) for d in search_docs]

     # Format

jgoodie

llm = ChatOpenAI(model="gpt-4o", temperature=0)

class TavilyState(TypedDict):
    question: str
    answer: str
    context: Annotated[list, operator.add]

jgoodie

def get_news_article_text(url):
    try:
        article = newspaper.article(url)
        title = article.title
        text = article.text_cleaned
    except Exception as e:
        logger.debug(f"Error occurred while fetching article at {url}: {e}")
        return {"url": url, "title":"", "text":""}
    return {"url": url, "title":title, "text":text}