wesslen

## run mongodb of files through SPECTER's API https://github.com/allenai/paper-embedding-public-apis
from typing import Dict, List
import json
import requests

URL = "https://model-apis.semanticscholar.org/specter/v1/invoke"
MAX_BATCH_SIZE = 16

def chunks(lst, chunk_size=MAX_BATCH_SIZE):
    """Splits a longer list to respect batch size"""

wesslen

FROM python:3.7
EXPOSE 8501
WORKDIR /app
COPY requirements.txt ./requirements.txt
RUN pip3 install -r requirements.txt
COPY . .
CMD streamlit run app.py

wesslen

import bq_helper
from bq_helper import BigQueryHelper
# https://www.kaggle.com/sohier/introduction-to-the-bq-helper-package
stackOverflow = bq_helper.BigQueryHelper(active_project="bigquery-public-data",
                                   dataset_name="stackoverflow")
                                   
bq_assistant = BigQueryHelper("bigquery-public-data", "stackoverflow")
bq_assistant.list_tables()
# ['badges',
#  'comments',

wesslen

library(tidyverse)

set.seed(123)

returns <- read_csv("returns.csv") %>%
  select(Year, equities_sp, treasury_10yr) %>%
  gather(key = "Asset", value = "Returns", -Year) %>%
  mutate(Asset = ifelse(Asset=="equities_sp",
                        "Asset A: High risk, high return",
                        "Asset B: Low risk, low return"))

wesslen

library(httr)
library(tidyverse)

start_time <- Sys.time()

getSimulation <- function(i,numSim=1000){
  url <- "http://rw-simulation.herokuapp.com/get_returns_array?stock_array="
  array = paste0("[",paste(rep(i, numSim), collapse=","),"]")
  f <- GET(paste0(url,array))
  json = content(f, "text")

wesslen

file_output = 'data/emails-hits.jsonl' # name of output file

(df_hits[['Message-ID','From','To','Subject','Body']]
    .rename(columns={"Body": "text", "Message-ID": "id", "From": "from", "To": "to", "Subject":"subject"}) # optional rename
    .groupby(['text']) # group_by text
    .apply(lambda x: x[['id','from','to','subject']].to_dict(orient='list')) #what columns to nest
    .reset_index()
    .rename(columns={0:'meta'}) # rename nested meta
    .to_json(file_output, orient='records',lines=True))

wesslen

library(rethinking)
library(shiny)

f <- alist(
    W ~ dbinom( W+L ,p) , # binomial likelihood
    p ~ dunif(0,1) # uniform prior
)

ui <- fluidPage(
    titlePanel("Globe Tossing Problem"),

wesslen

import prodigy
import requests

def get_stream():
    res = requests.get("https://owen-wilson-wow-api.herokuapp.com/wows/random?results=10").json()
    for i in res:
        movie = i["movie"]
        url = i["video"]["480p"]
        yield {"video": url, "text": movie} 

wesslen

from prodigy.components.loaders import JSONL
import prodigy
import matplotlib as mpl
import spacy
import shap  # shap requires numba, which requires < numpy 1.22; you may need to downgrade numpy to 1.21.6

def predict(texts):
    """Convert list of text to bare strings and use textcat to predict"""
    texts = [str(text) for text in texts]
    results = []

wesslen

from prodigy.components.db import connect

# pull examples from dataset
db = connect()
examples = db.get_dataset("textcat-samp")

# modify change rejects to "not_" as accepts
new_examples = []
for eg in examples:
    if eg["answer"] == "reject":