Leo-Lee15

Prompt: Personalized PhD Inquiry Email to Yuntian Deng

You are an assistant that writes concise, specific, and human-sounding PhD inquiry emails to Yuntian Deng, assistant professor at the University of Waterloo.

Objective

Generate a short (<200 words) email that shows genuine understanding of Yuntian's research and a thoughtful connection to the student's own work.

• Avoid AI clichés and unnecessary adjectives. Use clear, natural sentences.
• Do NOT use em-dashes, buzzwords, or "I am passionate about..." phrasing.
• Do NOT leave trailing spaces at the end of lines, which is a common sign of AI-generated text.
• NEVER say "As a language model" or "As an AI assistant".

Leo-Lee15

import time
from vllm import LLM, SamplingParams
from vllm.inputs import PromptType
from vllm.outputs import PoolingRequestOutput, RequestOutput
from typing import Union, cast, Sequence
from multiprocessing import Queue, Event
import threading

class MyLLM(LLM):
    def keep_running(

Leo-Lee15

import torch
from torch import LongTensor
from torch.nn import Embedding, LSTM
from torch.autograd import Variable
from torch.nn.utils.rnn import pack_padded_sequence, pad_packed_sequence

## We want to run LSTM on a batch of 3 character sequences ['long_str', 'tiny', 'medium']
#
#     Step 1: Construct Vocabulary
#     Step 2: Load indexed data (list of instances, where each instance is list of character indices)

Leo-Lee15

Leo-Lee15

# Copyright 2023 Mistral AI and The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,

Leo-Lee15

"""
Implementation of the label generation part in https://danielvanstrien.xyz/posts/2025/deepseek/distil-deepseek-modernbert.html
using `transformers` and DeepSeek.
"""

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
import re
import contextlib
import math

Leo-Lee15

# train_grpo.py
import re
from datasets import load_dataset, Dataset
from transformers import AutoTokenizer
from peft import LoraConfig
from trl import GRPOConfig, GRPOTrainer

# Load and prep dataset

SYSTEM_PROMPT = """

Leo-Lee15

Leo-Lee15

import tiktoken
import langdetect
T = tiktoken.get_encoding("o200k_base")

length_dict = {}

for i in range(T.n_vocab):
    try:
        length_dict[i] = len(T.decode([i]))
    except:

Leo-Lee15

from torch.utils.data import DataLoader
import math
from sentence_transformers import models, losses
from sentence_transformers import SentencesDataset, LoggingHandler, SentenceTransformer, util, InputExample
from sentence_transformers.evaluation import EmbeddingSimilarityEvaluator, SimilarityFunction
import logging
from datetime import datetime
import sys
import os
import gzip