---
license: other
license_name: upstage-solar-license
license_link: https://huggingface.co/upstage/Solar-Open2-250B/blob/main/LICENSE
base_model: vessl/Solar-Open2-250B-W4A8
tags:
  - safetensors
  - vllm
  - solar-open2
  - speculative-decoding
  - dspark
  - draft-model
pipeline_tag: text-generation
---

# Solar-Open2-250B-DSpark

A [DSpark](https://huggingface.co/Inferact/Kimi-K3-DSpark) speculative-decoding draft model for
[Solar-Open2-250B-W4A8](https://huggingface.co/vessl/Solar-Open2-250B-W4A8).

## Model Detail

| Field | Value |
|---|---|
| Method | DSpark (block-diffusion draft, block rejection sampling) |
| Target model | Solar-Open2-250B-W4A8 |
| Block size / `num_speculative_tokens` | 4 (exported `ns4` variant) |
| Hidden layers | 5 |
| Attention geometry, RoPE, KV-head count, `rms_norm_eps` | inherited from target |
| Auxiliary hidden-state taps | layers 4, 12, 24, 36, 44 of the target — all full (softmax) attention layers; Solar-Open2-250B interleaves 12 full-attention layers among 36 linear-attention (KDA) layers, and only the full-attention layers are tapped |
| Training | on-policy (responses regenerated by the target itself); prompt mixture = public instruction/code/multilingual datasets + a private dataset (not published) |

## Usage

```bash
vllm serve vessl/Solar-Open2-250B-W4A8 \
  --served-model-name solar-open2-250b \
  --tensor-parallel-size 4 --enable-expert-parallel \
  --kv-cache-dtype fp8 \
  --enable-prefix-caching \
  --speculative-config '{"method":"dspark","model":"vessl/Solar-Open2-250B-DSpark","num_speculative_tokens":4}' \
  --reasoning-parser solar_open2 --tool-call-parser solar_open2 \
  --enable-auto-tool-choice \
  --port 8000
```

## License

Distributed under the same **Upstage Solar License** as the Solar Open 2 base model and its W4A8 derivative,
per the license's requirements for derivative AI models (name prefixed with "Solar," "Built with Solar"
attribution).