--- base_model: cerebras/Qwen3-Coder-REAP-25B-A3B base_model_relation: quantized quantized_by: ArtusDev language: - en library_name: transformers tags: - qwen-coder - MOE - pruning - compression - exl3 license: apache-2.0 license_link: https://huggingface.co/cerebras/Qwen3-Coder-REAP-25B-A3B/blob/main/LICENSE pipeline_tag: text-generation ---
EXL3 quants of cerebras/Qwen3-Coder-REAP-25B-A3B using exllamav3 for quantization.
| Quant | BPW | Head Bits | Size (GB) |
|---|---|---|---|
| 2.5_H6 | 2.5 | 6 | 8.57 |
| 3.0_H6 | 3.0 | 6 | 10.10 |
| 3.14_H6 (optimized) | 3.14 | 6 | 10.43 |
| 3.5_H6 | 3.5 | 6 | 11.60 |
| 3.63_H6 (optimized) | 3.63 | 6 | 11.90 |
| 4.0_H6 | 4.0 | 6 | 13.13 |
| 4.5_H6 | 4.5 | 6 | 14.63 |
| 5.0_H6 | 5.0 | 6 | 16.16 |
| 6.0_H6 | 6.0 | 6 | 19.19 |
| 8.0_H8 | 8.0 | 8 | 25.32 |
You can download quants by targeting specific size using the Hugging Face CLI.
pip install -U "huggingface_hub[cli]"
2. Download a specific quant:
huggingface-cli download ArtusDev/cerebras_Qwen3-Coder-REAP-25B-A3B-EXL3 --revision "5.0bpw_H6" --local-dir ./
EXL3 quants can be run with any inference client that supports EXL3, such as TabbyAPI. Refer to documentation for set up instructions.
Made possible with cloud compute from lium.io