Awesome LLM Papers and repos on very comprehensive topics.

gpt-o1 like chain of thoughts with local LLMs in R

Arcmind Autonomous AI Agent powered by Dfinity's Internet Computer

A proof-of-concept async pipeline framework in Delphi to demonstrate modern AI orchestration with Promises, streaming, and chained execution. Designed by a math researcher to demonstrate that modern, elegant AI architectures can emerge from non-mainstream tools.

The official fork of THoR Chain-of-Thought framework, enhanced and adapted for Emotion Cause Analysis (ECAC-2024)

RelateLang (former RelateScript) is a declarative meta-language designed to streamline the creation of structured and consistent prompts for large language models (LLMs).

The official code for CoT / ZSL reasoning framework 🧠, utilized in paper: "Large Language Models in Targeted Sentiment Analysis in Russian"

CoT Reasoning in Autoregressive Image Generation

GPT natural-language user interface that fakes emotions and plans actions

Fine-tuned Qwen 3-1.7 B on the open-source PHYBench dataset using memory-light 4-bit LoRA adapters and Group Relative Policy Optimization with Chain-of-Thought prompts. A custom reward (boxed-answer format 0.2 | trace ≥ 6 lines 0.3 | SymPy-verified correctness 0.5) lifts mean reward from 0.19 to 0.54 in 550 Steps.

按照CoT原理设计的AI工作流，可配置、可中断编辑

This project explores GPT-2 and Llama models through pre-training, fine-tuning, and Chain-of-Thought (CoT) prompting. It includes memory-efficient optimizations (SGD, LoRA, BAdam) and evaluations on math datasets (GSM8K, NumGLUE, StimulEq, SVAMP).

A collection of prompts and techniques to enhance the reasoning, response development, and quality of Large Language Models (LLMs).

ragTAG is a conversational AI script that creates a roundtable dialogue between user assigned characters with their own different objectives and perspectives.

Distress Support Chatbot Using llama3-8b

LLM model that aims to accurately solve calculations from expressions and word problems

OmniLlama é minha tentativa de criar uma maneira eficiente, local de trabalhar com cadeias de raciocínio com modelos relativamente pequenos.

combating the llm fomo, feeding the shiny object syndrome, for folly and partially for curiousity

This repository is a proof of concept for replicating the reasoning capabilities of OpenAI's O1 model using alternative frameworks. It employs a sequential agent-based system powered by the Gemini API, facilitating iterative problem-solving for coding-related challenges through a Flask web application.

LLMs are commonly used to rewrite or make stylistic changes to text. The goal is to recover the LLM prompt that was used to transform a given text.