Anyonexinansai/deepseek-r1-distill-qwen-1.5b-base
The Anyonexinansai/deepseek-r1-distill-qwen-1.5b-base is a 1.5 billion parameter language model, derived from the DeepSeek-R1-Distill-Qwen architecture, with a 32768 token context length. This model is provided as an obfuscated reviewer package, indicating its primary use for evaluation and review purposes. It includes sharded safetensors model weights, tokenizer files, and configuration files, along with a client secret for input mapping and output recovery.
Loading preview...
Model Overview
This repository hosts the Anyonexinansai/deepseek-r1-distill-qwen-1.5b-base, a 1.5 billion parameter language model based on the DeepSeek-R1-Distill-Qwen architecture. It features a substantial context length of 32768 tokens. This specific release is presented as an obfuscated reviewer package, intended for anonymous evaluation and review processes.
Key Components and Structure
The package includes essential files for model operation and review:
- Sharded safetensors model weights: The core model parameters are provided in a sharded format.
- Tokenizer files: Necessary for processing text inputs and outputs.
- Model configuration files: Define the model's architecture and operational settings.
client/client_secret.pt: A critical component for reviewer-side input mapping and output recovery, ensuring the integrity of the review process.- Upstream license and README: Preserved for transparency regarding the model's origins and licensing.
Usage and Setup
To utilize this model, users need to download the repository contents and place them in a designated directory. The runtime configuration requires specifying the model_path to the repository root and the client_secret_path to the client/client_secret.pt file. A minimal dependency set including torch, transformers, accelerate, and safetensors is required for local loading and operation.