HaleES/sensei-sous-v2

Sensei Sous v2

A fine-tuned Qwen2.5-1.5B-Instruct model optimized as a fast local fallback for the Sensei OS orchestration layer at HaleES.

Quantization: Q8_0 (1.53 GB)
Context window: 32,768 tokens
Roles: chat, tools
Provider: local (node-llama-cpp / llama.cpp)

Usage (Sensei OS)

Loaded automatically under the alias sensei.sous. Manifest entry:

{
  "modelId": "sensei.sous",
  "repo": "SenseiOS/sensei-sous-v2",
  "filename": "sensei-sous-v2-q8_0.gguf",
  "quant": "Q8_0",
  "roles": ["chat", "tools"],
  "contextWindow": 32768,
  "provider": "local"
}

Downloads

Download the GGUF file and use it with any llama.cpp-compatible inference engine.