liuyisen-bin / transformer-sru-lys
Transformer-SRU LiDAR navigation for Unitree G1: Isaac Lab training, MuJoCo validation, and ROS 2 deployment.
liuyisen-bin / transformer-sru-lys
znxzsy / AlignLedger
Auditable human-feedback annotation, review, provenance, and frozen training-data export.
cneuralnetwork / smol-kimi-k3
A 49M-parameter Kimi K3-inspired language model trainable on one 8GB GPU
YuehHanChen / automated_alignment_researcher
[Anthropic research] We built automated alignment researchers to autonomously post-train other AIs to improve alignment
dahan6 / whetstones-lab
Whetstones Lab: measuring coevolution between adaptive malware and behavioral defense. De-weaponized research release — genomes, evolution rounds, telemetry, training harness (stubbed primitives), and lab range tooling. Payloads, real primitives, and C2 infrastructure intentionally withheld. Research and defensive use only.
Thinking-Space / One-Shot-OPD
Rethinking On-Policy Distillation of Large Language Models II: One Training Example
hokindeng / object-permanence
Training Object Permanence in World Models — the codebase
SalesforceAIResearch / Beagle
:dog: Beagle is designed to be the unified platform for the broader Recursive Self-Improvement (RSI) ecosystem. Think of it as Hugging Face Trainer for agent evolution. Its efficient rollout engine enables evaluation and evolution at scale across different agents, datasets, benchmarks, and evolution methods.
nftechie / flm
Train and chat with a frozen language model coupled to the full retained MaleCNS fly connectome.
sohu-mptc / FlashRec
FlashRec is a CUDA-graph engine for generative recommendation: wide beam search (3–5 SID steps, n=50–512+) over a trie-constrained catalog, in-process FP8 serving, and ranked beams on /v1/chat/completions.
MaxwellCCC / hangzhou-mahjong
Hangzhou Mahjong training and evaluation framework with public-information Transformers, paired branch reinforcement learning, and the r2600 checkpoint.
jerryjliu / fly_ocr
Printed-character recognition through a fixed fruit-fly connectome, with a small trained decoder, compound-eye replays and a full research report.
smallfly / Carveout
Carveout finds and labels objects in trained 3D Gaussian Splatting scenes.
limingrui679-design / CascadeLens
Python-first systemic-risk research software for evidence-graded, bitemporal dependency graphs. Import JSON, CSV, GraphML, or NetworkX; run bounded shocks, compare constrained interventions and sensitivity surfaces, and export recomputation-verifiable RiskPacks. Includes 16 scenario-only cases and 11 connector contracts.
avbiswas / bev-train
Training Jev-like decision models using Qwen3 with choice-order invariance
datologyai / zephon
Fast, flexible, elastically deterministic data loading for foundation model training
siqiwang0712-commits / financial-risk-agent
Evidence-grounded financial risk analysis from corporate filings, combining constrained LLM interpretation, deterministic analytics, versioned rules, and auditable evidence trails.
Zhengsw03 / RL-Token-Pi05-open
RL Token pipeline for the SO-101 arm: π0.5 fine-tuning, RL Token training, critical-phase classifier, online RL and deployment.
bnsd55 / jevmlx
Jev-style parallel constrained decisions for any MLX model on Apple Silicon. Typed, schema-valid JSON in one forward pass.
Starnodes2024 / ComfyUI-YuE2-Trainer
LoRa Trainer For YuE in ComfyUI
NVIDIA / sports-intelligence-playbooks
This repository provides a starter kit for sports intelligence built on NVIDIA AI stack: playbooks, training recipes, and inference for Multimodal Language Models.
l1934332574-maker / Kohya-LoRA-Tool
Windows desktop workspace for local LoRA training. Integrates kohya, musubi-tuner, AI Toolkit and Fizgig with guided setup, dataset tools, sample comparison and local MCP access.
kajeesan / Open-Health-Atlas
Explore sleep, training, mood and nutrition together using local records, traceable calculations and optional MCP tools for your preferred AI client.
lok-i / vibe
Post-training humanoid whole-body trackers for perceptive control tasks
RenaGao / jev-dataops
An open-source JEV-powered workbench for streaming data selection, quality evaluation, automatic LoRA training and held-out model evaluation.
mallahyari / nemotron-asr-streaming-farsi
Streaming Persian (Farsi) speech recognition: fine-tuning NVIDIA Nemotron 3.5 ASR streaming. Data prep, training, evaluation, inference.
Kai26-Han / miniLLM
从 0 训练大模型，单卡训练一个 200M 参数模型的全流程（train a 200M‑parameter model from scratch）
haohuiliang / RobustOps
An Agentic RL project training 4B enterprise tool agents to complete business tasks reliably and recover from tool faults.
nyc-esq / hey-donut
A custom "Hey Donut" wake word for Home Assistant — 62 KB, runs on-device, zero false accepts per hour. Includes the full training pipeline so you can teach your assistant any name. CC0.
SAGAR-TAMANG / sarvam-jev
Generation-free typed decisions on Indic LLMs. An open Jev-style inference engine on sarvam-1: constrained logit readout instead of autoregressive JSON. Runs client-side in the browser.
3587jjh / HuRo
HuRo: Robotizing Human Videos for Scalable VLA Pretraining (CoRL 2026)
adeelarshad414 / devops-interns-proj
A deliberately breakable three-tier platform plus a full DevOps / SRE / DevSecOps training curriculum (handbook, capstone assignment, solutions) for an intern rotation.
Roboparty / I-BFM-Code
I-BFM training code and interactive Carry Box, Push Box, and Kick Box demos
Vxtzq / CrowdGPT
CrowdGPT is an opensource AI that aims to be trained without any datacenters
nourya-aliz / self_play_pretraining
—
YOMXXX / McOptima
麦门最优解引擎 · 用运筹学穷举麦当劳菜单，算出你这张嘴的全局最优解 | Multi-constraint 0/1 knapsack solver on McDonald's official MCP | 2026 麦当劳程序员节参赛作品
leeruibin / avatarforever
Real-time, long-horizon audio-driven avatar generation via decoupled parallel training.
trainingsites / campus-ai-os
Campus AI OS — an AI staff that remembers your business. One folder, one chief of staff, usable from Claude or Codex.
veda-sparse / Miowtion
Sparse acceleration and LoRA fine-tuning of MiniMax-H3 on resource-constrained machines.
openfabric-systems / simllm
Network-faithful simulation of LLM serving and training deployments
r34vtraining / Longshot_Studio
User friendly Interface for ComfyUI MInimax H3
Picrew / awesome-rsi
A curated collection of papers, research blogs, and open-source projects on recursive self-improvement (RSI), self-modifying agents, and iterative self-training.
AhmadJeddi / Plant-Disease-Classification
A ResNet50-DCBAM deep learning framework for plant disease classification using the PlantVillage dataset with reproducible training and evaluation pipelines.
xixiaouab / ReaLVR-code
ReaLVR: Grounding latent visual reasoning in visual evidence. Training and evaluation code.
MahdiKordian / DataForge
Automated data preparation for machine learning. Load a dataset, clean, transform, encode, scale and split it, explore it with built-in analysis, and export model-ready data with a full report. Built with Python and Streamlit.
corneliahawthorne8 / structured-json-output-guide
Argolink-focused schema-constrained JSON responses for GPT-6 Sol.
Tcotl / DVLAA
DVLAA (Damn Vulnerable LLM and Agent Application) is a local LLM and Agent security range for OWASP LLM Top 10 and Agent security training.
intikhab49 / open-jev-typed-decision-engine
Open reproduction of TypeSafe Jev: a 150M typed decision engine (noul/choice/score in one non-autoregressive pass, calibrated confidence). 0.697 vs Jev's 0.727, 2.5x better calibrated, 4x faster, free. Trains on a Colab T4 in 30 min.
caiyuchen-ustc / On_Policy_Vector_Training
On Policy Vector Training
huyxdang / Tiny-GPT-Live
Can you train a GPT-Live with $50?
michaelc-yu / cs336-llm-systems
Large language model systems implementation and optimization: Transformers, BPE tokenization, FlashAttention, Triton, distributed training, GPU profiling, and GRPO reasoning RL
Qin-sor / agent-context-lab
A minimal coding-agent harness built to measure what context management actually costs: constraint survival after compaction (8%→83%), default-FAIL verification against false completions, and KV-cache prefix design (21%→90%). Stdlib-only, real deepseek-chat runs
llm-infra-atlas / llm-infra-atlas.github.io
A living guide of the systems that train, serve, and scale LLMs
IamBusy / OpenJev-Vision
Open visual probability research: encode an image once, answer multiple structured questions. Public data, trained weights, reproducible baselines and failure cases.
dev-luigi / neural-network-digits
Neural network from scratch in pure Python + NumPy that recognises handwritten digits (MNIST), with an educational GUI: watch it learn, tweak it while it trains, look inside every layer.
tardellirs / gemma3moe-builder
Sparse-upcycle a dense Gemma-3-270m into a Mixture-of-Experts (gemma3moe) and serve it on llama.cpp (CPU). Upcycle, train, inspect routing, GGUF.
franciscocarloserra / ttslibre
Small, fast, free: an open recipe for a sub-100M TTS. Code, data, weights. The PoC trained to a working model in a few hours on one RTX 3090, and inspired a wave of open-weights TTS releases.
DenisSergeevitch / game-sensitivity-coach
An agent skill for evidence-based mouse sensitivity tuning, cross-game conversion, and gameplay review. Works with Codex, Claude Code, and Agent Skills hosts.
lindayoungk / Valheim-Assistant
Ultimate assistant for Valheim. Features: God Mode, Immortality, Infinite Health/Stamina/Eitr, One-Hit Kills, Invisibility, Ghost Mode, Flight Mode, Infinite Weight, Free Crafting, Unlock All Blueprints, Infinite Items, No Durability Loss, and more. 19+ functions.
WB-WaM / WB-WAM-Official
WB-WAM: Heterogeneous Body–Hand Pre-training for Humanoid Loco-Manipulation.