fine-tuning

This repository collects awesome representative papers and resources for "From Pre-training to Post-training: A Survey on Time Series Foundation Models" (TechRxiv-26).

ModelSoups for Tensorflow2 and Torch

Code for our paper "Generalized Orderless Pooling Performs Implicit Salient Matching" published at ICCV 2017.

Awesome List of Vision Language Prompt Papers

Code repository for AI Builders Bootcamp #2

A zero-dependency Python module for inspecting and converting coding-agent session files (.jsonl) — Claude Code, Codex, and Pi — into the messages format expected by Hugging Face Transformers.

Tuning the Finetuning: An exploration of achieving success with QLoRA

Language-interfaced fine-tuning for chemistry

FTPipe and related pipeline model parallelism research.

A Continual Learning Framework for Production LLM Agents

LLM-powered assistant for first-time dads, fine-tuned on real parenting discussions using Mistral 7B

Language Model Fine-tuning for Moby Dick

Pure RL to post-train base models for social reasoning capabilities. Lightweight replication of DeepSeek-R1-Zero with Social IQa dataset.

[AAAI 2026] D²PPO: Diffusion Policy Policy Optimization with Dispersive Loss.

This is the official code for the paper "Booster: Tackling Harmful Fine-tuning for Large Language Models via Attenuating Harmful Perturbation" (ICLR2025 Oral).

从零开始学大模型Transformer、GPT2、BERT pre-training and fine-tuning from scratch

A python implementation of “Self-Supervised Learning of Spatial Acoustic Representation with Cross-Channel Signal Reconstruction and Multi-Channel Conformer” [TASLP 2024]

🚀 Engram-PEFT: An unofficial implementation of DeepSeek Engram. Inject high-capacity conditional memory into LLMs via sparse retrieval PEFT without increasing inference FLOPs / DeepSeek Engram 架构的非官方实现。通过参数高效微调 (PEFT) 为大语言模型注入超大规模条件记忆，支持稀疏更新且不增加推理开销。

A Streamlit app for generating high-quality Q&A training datasets from text and PDFs, leveraging Gemini, Claude, and OpenAI for LLM fine-tuning.

Use Large Language Models like OpenAI's GPT-3.5 for data annotation and model enhancement. This framework combines human expertise with LLMs, employs Iterative Active Learning for continuous improvement, and integrates CleanLab (Confident Learning) to ensure high-quality datasets and better model performance