Mini Parameter Golf

MiniMind is an open-source tiny large-language-model project and end-to-end training tutorial developed by Jingyao Gong. Its main dense model is approximately 64M parameters, and the repository provides the model architecture, tokenizer, datasets, inference code, and training pipeline. The pipeline covers pretraining, supervised fine-tuning, hand-written LoRA, DPO, PPO, GRPO, CISPO, model distillation, tool calling, adaptive thinking, and agentic reinforcement learning. Core algorithms are implemented directly with native PyTorch rather than relying on high-level abstractions from third-party training libraries. Its agentic reinforcement-learning path performs multi-turn rollouts, executes generated tool calls, appends tool observations to the context, calculates trajectory-level rewards, and updates the policy; rollout can use local PyTorch generation or an SGLang server. MiniMind supports dense and mixture-of-experts variants, single- and multi-GPU training, YaRN-based RoPE length extrapolation, Transformers-format models, and inference through llama.cpp, vLLM, or Ollama. The repository also includes a Streamlit chat interface and a lightweight OpenAI-compatible API server with tool-call and reasoning fields. It is released under the Apache License 2.0.

mini-AGI is an experimental continual-learning, byte-level language model that trains from scratch on a CUDA-capable GPU with at least 8 GB of VRAM. It reads 256 byte values directly rather than using a conventional tokenizer, processes data one chunk at a time, and uses the same forward path for training and generation. Its architecture combines dense prelude blocks with a recurrent block applied up to 24 times, adaptive PonderNet-style halting, and per-application top-8 routing through a dynamically growing and pruning expert pool. Expert weights and Adam moments are stored as files on disk; a working set is paged through RAM and VRAM as needed, allowing the pool size to exceed available VRAM. The project includes Python commands for building corpora, reading files, streaming training, serving a local web interface, and inspecting or replicating continual-learning experiments. The repository describes the current model as toy-level rather than frontier-capable, and its weights are generated in the local weights directory rather than distributed with the source.

Weco AI is an autoresearch agent platform for running and inspecting optimization experiments. It repeatedly proposes code changes, evaluates them against a fixed command and measurable goal, keeps changes that improve the score, and discards weaker attempts. Its dashboard presents the search as an experiment tree with scores and diffs, allowing users to branch from promising runs and steer an ongoing search toward a different idea. The workshop materials demonstrate the approach with Mini Parameter Golf and a Flappy Bird controller evaluated across multiple conditions, including hidden validation data and limits on editable code.

Mini-SGLang is a compact implementation of the SGLang large-language-model inference and serving architecture. Its roughly 5,000-line, type-annotated Python codebase is intended as both a usable inference engine and a readable reference for understanding LLM serving systems. The framework serves models through an OpenAI-compatible API or an interactive shell. Its implementation includes radix caching for reusing KV cache across shared prefixes, chunked prefill, overlap scheduling to run CPU scheduling alongside GPU computation, tensor parallelism across multiple GPUs, and optimized kernels using FlashAttention and FlashInfer. Mini-SGLang runs on Linux with NVIDIA CUDA dependencies and supports installation from source or via Docker. The repository documents deployment on one or more GPUs, including models such as Qwen and Llama, and provides offline and online benchmark configurations.

AirLLM is an open-source Python library for running large language model inference on GPUs with limited memory. It reduces GPU memory use by loading and processing one model layer at a time instead of holding the entire model in GPU memory; for sparse mixture-of-experts models, it streams individual experts as needed. The project is designed to run large models without requiring quantization, distillation, or pruning, and provides a pip-installable package with support for multiple model families, CPU inference, macOS, and optional model compression and quantization features.