AI Train2025
Forecasts financial candlesticks (OHLCV K-lines) with a decoder-only transformer pre-trained on 12B+ records from 45 exchanges. A tokenizer turns market data into discrete tokens, enabling price/volatility forecasting and synthetic K-line generation.
AI Image2025
Self-supervised vision foundation model producing dense, patch-level features that transfer to classification, segmentation, depth, and detection with a frozen backbone. Spans ViT-S (21M) to ViT-7B (6.7B params), plus ConvNeXt and satellite variants.
AI Train2025
Worked examples and reusable abstractions for fine-tuning open LLMs via the Tinker training API: you write the training loop while distributed execution runs remotely. Covers SFT, math/code RL, DPO, three-stage RLHF, distillation, and tool use.
Embodied AI2026
Enables research-grade character animation with neural networks in a single NumPy/PyTorch environment — train models, run inference, and visualize results without leaving Python. Includes ECS-style architecture, mocap import (GLB/FBX/BVH), built-in renderer, and headless/standalone modes for rapid prototyping.
Embodied AI2026
Train robot reinforcement-learning agents with a heterogeneous runtime that streams CPU-parallel physics simulations (MuJoCo / Motrix) via shared memory into GPU/accelerator policy learners; provides a unified CLI, cross-platform backend support and demo checkpoints.
AI Train2026
A challenge repository for training the best language model that fits inside a 16,000,000‑byte (16MB) submission artifact; provides baseline training code, FineWeb bpb evaluation, a public leaderboard, and compute-grant instructions for short 8×H100 runs.
AI Train2026
Provides a one-command CLI to fine-tune and post-train LLMs, with layer streaming that lets an 8B model be fine-tuned on a 4 GB laptop GPU. Auto-configures quantization, LoRA adapters, batching and evaluation gates, and supports export and serving workflows.
AI Train2026
Runs transformer forward+backward training directly on Apple's Neural Engine by reverse-engineering private ANE APIs; includes per-layer ANE kernels, INT8 optimizations and benchmarks. Proof-of-concept only—relies on undocumented APIs, has low utilization and CPU fallbacks.
AI Agent2026
Lets an LLM autonomously propose, edit, run, and evaluate short single‑GPU LLM training experiments — fixed 5‑minute runs (~12 experiments/hour). Agent edits a single train.py; humans supply goals via program.md. Single‑GPU, val_bpb metric.
AI Agent2026
Runs an autonomous self-improvement loop where a meta agent crafts a task-specific agent, a target agent executes trials, and a feedback agent updates both harness (code) and model weights—provider-agnostic profiles with reproducible runs and a live dashboard.
AI Agent2026
Trains reusable natural-language 'skills' for frozen LLM agents by optimizing the skill document in text-space — using trajectory-driven edits, validation-gated updates, and deployable best_skill.md artifacts. Multi-backend, zero inference-time cost at deployment, designed for iterative, validation-led skill improvement.
AI Train2026
End-to-end Python framework for training and serving NVIDIA's Cosmos world models (Cosmos3), integrating distributed training (FSDP/TP/CP/PP), DCP/safetensors checkpoints, dataset adapters, multiple inference backends, online serving, and agent skills.