Raibows

import os
import json
from dataclasses import dataclass
from multiprocessing import Pool
from collections import defaultdict
from datetime import datetime

from math_verify import math_metric, LatexExtractionConfig, ExprExtractionConfig, StringExtractionConfig, parse, verify
from tqdm import tqdm
from simpleArgParser import parse_args

Raibows

import torch
import os
import json
from safetensors.torch import load_file, save_file

def replicate_lora_a(name: str, weight: "torch.Tensor") -> dict[str, "torch.Tensor"]:
    prefix, suffix = name.split('qkv_proj')
    res = {}
    for t in ['q_proj', 'k_proj', 'v_proj']:
        name = f"{prefix}{t}{suffix}"

Raibows

import os
import time
import requests
from requests.adapters import HTTPAdapter, Retry
from huggingface_hub import configure_http_backend
import argparse


parser = argparse.ArgumentParser()
parser.add_argument('--model', type=str, help='name of huggingface model to download', default='sshleifer/tiny-gpt2')

Raibows

import openai
import asyncio
from typing import Any

async def dispatch_openai_requests(
    messages_list: list[list[dict[str,Any]]],
    model: str,
    temperature: float,
    max_tokens: int,
    top_p: float,

Raibows

channels:
- anaconda
- pytorch
- nvidia
- conda-forge
dependencies:
- python=3.8
- pip
- pytorch=1.13.1
- torchvision

Raibows

# note this is a copy from https://paste.ubuntu.com/p/Nx5CcSmhHn/ for convenience
import torch
import torch.nn.functional as F
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

def SimCSE_loss(pred, tau=0.05):
    ids = torch.arange(0, pred.shape[0], device=device)
    y_true = ids + 1 - ids % 2 * 2
    similarities = F.cosine_similarity(pred.unsqueeze(1), pred.unsqueeze(0), dim=2)