Jev-style Structured Decision Models for Agent Workflows
AI & ML interests
All Models, One Mind.
Recent Activity
View all activity
Papers
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
Claw-SWE-Bench: A Benchmark for Evaluating OpenClaw-style Agent Harnesses on Coding Tasks
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
MLX format of NeoHorse-1 Models
-
TokenRhythm/NeoHorse-1-4B-MLX-4bit
Text Generation • 4B • Updated • 600 • 3 -
TokenRhythm/NeoHorse-1-9B-MLX-4bit
Text Generation • 9B • Updated • 630 • 4 -
TokenRhythm/NeoHorse-1-4B-MLX-8bit
Text Generation • 4B • Updated • 388 • 2 -
TokenRhythm/NeoHorse-1-9B-MLX-8bit
Text Generation • 9B • Updated • 772 • 6
Jev-style Structured Decision Models for Agent Workflows
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
GGUF format of NeoHorse-1 Models
MLX format of NeoHorse-1 Models
-
TokenRhythm/NeoHorse-1-4B-MLX-4bit
Text Generation • 4B • Updated • 600 • 3 -
TokenRhythm/NeoHorse-1-9B-MLX-4bit
Text Generation • 9B • Updated • 630 • 4 -
TokenRhythm/NeoHorse-1-4B-MLX-8bit
Text Generation • 4B • Updated • 388 • 2 -
TokenRhythm/NeoHorse-1-9B-MLX-8bit
Text Generation • 9B • Updated • 772 • 6