Pinned Loading

1. Low_rank_Monte_Carlo_Smoluchowski Low_rank_Monte_Carlo_Smoluchowski Public

   Low-rank Monte-Carlo simulation for aggregation-fragmentation kinetics

   7

2. FDMSk-Smoluchowski FDMSk-Smoluchowski Public

   This repository contains the code for the paper titled "Mosaic-skeleton approximation is all you need for Smoluchowski equations"

   C++ 5

3. llm-efficient-training llm-efficient-training Public

   Building a distributed LLM training pipeline from scratch in plain PyTorch — no HuggingFace Trainer. Four progressive tasks: single-GPU baseline (GPT-2), mixed precision (bf16 + gradient checkpoint…

   Python 10

4. gsm8k-post-training gsm8k-post-training Public

   Post-training pipeline for Qwen2.5-1.5B on GSM8K math benchmark. Two-stage training: LoRA SFT on 67K OpenMathInstruct-2 examples + GRPO reinforcement learning (DAPO / Dr. GRPO). Achieves 80% accura…

   Python 8