--- license: apache-2.0 tags: - mistral - bittensor - subnet5 - model-provider language: en --- # Supreme_V2 — Subnet 5 Model Provider **Model:** Supreme_V2 **Architecture:** Mistral 7B **Usage:** Bittensor Subnet 5 (HONE / Manifold Labs) **Status:** Production-ready **License:** Apache 2.0 --- ## 🧠 Overview Supreme_V2 is a customized Mistral 7B model optimized specifically for Bittensor **Subnet 5**. It has been configured for: - stable and consistent responses - fast inference on consumer GPUs - deterministic behavior required by Subnet 5 validators - full compatibility with the HONE Sandbox The model is provided in the `safetensors` format, split into **3 shards** to ensure maximum compatibility during streaming and loading. --- ## 📁 Included Files ### Model Weights - `model-00001-of-00003.safetensors` - `model-00002-of-00003.safetensors` - `model-00003-of-00003.safetensors` - `model.safetensors.index.json` ### Configuration - `config.json` - `generation_config.json` - `special_tokens_map.json` - `tokenizer.model` - `tokenizer_config.json` ### Sandbox (Required for Subnet 5) - `sandbox/sandbox_runner.py` - `sandbox/requirements.txt` - `sandbox/__init__.py` These files allow Subnet 5 validators to load the model deterministically and evaluate it during inference tasks. --- ## 🔧 Sandbox Entrypoint Subnet 5 validators will load the model using the following interface: ```python from sandbox import SandboxRunner runner = SandboxRunner() response = runner.infer("Hello!") print(response)