Spark · X2.5 · 4B
EXL3 · 6.0 bpw · 3.8 GB · Dense · 36 layers
[!NOTE] An ExLlamaV3 build of
XHToken/Spark-X2.5-4Bat 6.0 bits per weight. See Quants for sibling repos at other bit‑widths or browse the collection.
Quants
| BPW | Size | Status |
|---|---|---|
| 6.0 | 3.8 GB | this repo |
Inference
| Loader | Use it for |
|---|---|
| TabbyAPI | OpenAI‑compatible HTTP server. Drop‑in for OpenAI clients. |
| text‑generation‑webui | Local chat UI. Pick the ExLlamaV3 loader from the model dropdown. |
| ExLlamaV3 | Direct Python API for embedding the model in your own code or pipeline. |
Download
pip install -U huggingface_hub
hf download \
Honkware/Spark-X2.5-4B-exl3-6.0bpw \
--local-dir ./Spark-X2.5-4B-exl3-6.0bpw
quantization_config.json)| Setting | Value |
|---|---|
| Format | EXL3 |
| Bits per weight | 6.0 |
| Head bits | 6 |
| Calibration rows | 250 |
| Calibration data | exllamav3 bundled mix (c4, code, multilingual, technical, tiny, wiki) |
| Codebook | mul1 |
| Out‑scales | always |
| Parallel mode | enabled |
License & use
[!IMPORTANT] Use and license follow the base model. Quantization adds no additional restrictions. Refer to the upstream repository for terms, citation, and safety documentation.
Quantized with BlockQuant · convention
{org}/{model}-exl3-{bpw}bpw