azarnyx

# Deserialize the Invoke request body into an object we can perform prediction on
input_object = input_fn(request_body, request_content_type)

# Load the model
model = model_fn(model_dir)

# Perform prediction on the deserialized object, with the loaded model
prediction = predict_fn(input_object, model)

# Serialize the prediction result into the desired response content type

azarnyx

import os
import json
import numpy as np
from joblib import load
import argparse
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import f1_score
from joblib import dump, load

azarnyx

from transformers import AutoTokenizer, AutoModel, TFAutoModel

MODEL = "cardiffnlp/twitter-roberta-base"
TOKENIZER_EMB = AutoTokenizer.from_pretrained(MODEL)
MODEL_EMB = AutoModel.from_pretrained(MODEL)

def preprocess(text):
    new_text = []
    for t in text.split(" "):
        t = '@user' if t.startswith('@') and len(t) > 1 else t

azarnyx

sep_sol = [i*j for i,j in  itertools.product(sol1.x, sol2.x)]
mse = np.sqrt(np.mean((sol_common.x-sep_sol)**2))
print("Difference:", mse)

azarnyx

bnds = tuple([(0, None)]*(num_sides2*num_sides1))

# remember number of probabilities
lookup_map = {}
for ni, piqj in enumerate(itertools.product(sides1, sides2)):
    lookup_map[ni] = piqj

# common guesses
guess_common = np.ones(len(lookup_map))/len(lookup_map)

azarnyx

ini_guess1 = np.array([1/num_sides1]*num_sides1)
ini_guess2 = np.array([1/num_sides2]*num_sides2)

sol1 = minimize(sum_form, ini_guess1,
                bounds=bnds1, constraints=[cons01, cons21], options={'maxiter':1001})
sol2 = minimize(sum_form, ini_guess2,
                bounds=bnds2, constraints=[cons02, cons22], options={'maxiter':1001})

azarnyx

sides1 = np.arange(0, num_sides1)+1.
sides2 = np.arange(0, num_sides2)+1.
cons21 = ({'type': 'eq', 'fun': lambda p: sides1.dot(p) - mean1})
cons22 = ({'type': 'eq', 'fun': lambda q: sides2.dot(q) - mean2})

azarnyx

cons01 = ({'type': 'eq', 'fun': lambda p: np.sum(p) - 1.})
cons02 = ({'type': 'eq', 'fun': lambda p: np.sum(p) - 1.})

bnds1 = tuple([(0, None)]*(num_sides1))
bnds2 = tuple([(0, None)]*(num_sides2))

azarnyx

import numpy as np
from scipy.optimize import minimize
import pylab as plt
import itertools

num_sides1 = 6
num_sides2 = 10
mean1 = 4.5
mean2 = mean1

azarnyx