AI Model Training

FS-DFM Apple

The few-step discrete flow matching text generation model launched by Apple can achieve thousand-step quality with 8 steps of sampling.

Numerai Numerai

The world's first quantitative hedge fund platform driven by crowdsourced AI models, supporting NMR mortgage and meta-model aggregation

ACE-Ego ACE-Ego

One brain, multiple types of embodied operation VLA model research project

Gemini Diffusion Google DeepMind

Experimental text diffusion model launched by Google DeepMind, non-autoregressive generation, iterative text refinement

OmniManip OmniManip

Object center constraint-driven universal robot operation method

Gemma 4 12B Google

Google's open source coder-less unified architecture multi-modal large model, the local running performance of 12B parameters exceeds the previous generation 27B

Agent Lightning Microsoft

Microsoft's agent training framework uses reinforcement learning and other methods to allow Agents to continuously optimize from experience.

Aim Stack Aim

Open source ML experiment tracking and visualization platform for teams focused on data sovereignty

Groq Groq

Self-developed LPU hardware-driven ultra-high-speed AI inference platform, developer API can reach up to 1000 tokens/second at the fastest

Arize AI Arize

AI Agent observability, evaluation and continuous learning platform

HoneyHive HoneyHive

Observation, evaluation, monitoring and continuous improvement platform for production-level AI Agents

JAX JAX

A differentiable programming framework from Google designed for high-performance ML research