Prime IntellectSYNTHETIC-1
SYNTHETIC-1-SFT-7B
SYNTHETIC-1-SFT-7B adapts Qwen2.5-7B-Instruct with the supervised subset of SYNTHETIC-1. The dataset contains reasoning traces generated with DeepSeek-R1.
Research release
What the release provides
The card identifies this as an initial supervised model from the SYNTHETIC-1 data project. The dataset collection is published separately, so researchers can inspect the training material.SYNTHETIC-1-SFT-7B official model card
How to use it well
Practical guidance. Use this release when comparing reasoning datasets or reproducing a supervised training stage. Keep dataset versions fixed when comparing it with another model.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | SYNTHETIC-1-7B-SFT |
| Context window | 32,768 tokens Configured position limit. Serving limits depend on the runtime and settings. SYNTHETIC-1-SFT-7B model configuration |
| Repository license | Apache 2.0SYNTHETIC-1-SFT-7B official model card |
| Base model | Qwen2.5-7B-InstructSYNTHETIC-1-SFT-7B official model card |