macleginn

import random
import gzip
import json
import os
import requests
import pickle
from typing import List, Dict, Any, Tuple
from datetime import datetime

macleginn

import argparse
import os

cache_root = os.path.abspath("./hf_cache")
os.environ["HF_HOME"] = cache_root
os.environ["HF_HUB_CACHE"] = os.path.join(cache_root, "hub")
os.environ["VLLM_CACHE_ROOT"] = os.path.join(cache_root, "vllm")

import torch
import pandas as pd

macleginn

import pandas as pd
from vllm import LLM, SamplingParams
from transformers import AutoTokenizer
from tqdm import tqdm

def llama3_call(user_prompt, temperature=0.0):
    user_prompt = "### DOCUMENT:\n" + user_prompt + "\n### ANSWER" 
    messages = [{"role": "system", "content": system_prompt}, {"role": "user", "content": user_prompt}]
    messages = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt = True)
    sampling_params = SamplingParams(temperature=temperature, max_tokens=10)

macleginn

from math import pi
import numpy as np

x = np.array([1, 0, 0])
y = np.array([1, 0, 0.001])  # very similar to 
z = np.array([0, 1, 0])      # orthogonal to x
q = np.array([-1, 0, 0])     # looks in the opposite direction to x

def angular_distance(v1, v2):
    cos_of_angle = np.dot(v1, v2) / np.linalg.norm(v1) / np.linalg.norm(v2)

macleginn

"""
Conceptual pseudocode explaining how sliding-window attention works
using loops. This is NOT meant to be efficient or runnable in a real
model. It simply illustrates what PyTorch operations like:

    unfold
    unsqueeze
    squeeze
    batched matrix multiplication

macleginn

a
abandon
ability
able
abortion
about
above
abroad
absence
absolute

macleginn

class Counter:
    def __init__(self):
        self.count = 0
        
    def get_value(self):
        return self.count
    
    def increment(self):
        self.count += 1

macleginn

class Counter:
    def __init__(self):
        self.count = 0
        
    def get_value(self):
        return self.count
    
    def increment(self):
        self.count += 1

macleginn

class Counter:
    def __init__(self):
        self.count = 0
        
    def get_value(self):
        return self.count
    
    def increment(self):
        self.count += 1

macleginn

import json
import random
from math import ceil
import numpy as np
import torch
import torch.nn.functional as F
from transformers import AutoModelForCausalLM, AutoTokenizer
from tqdm.auto import tqdm