| | --- |
| | license: mit |
| | language: |
| | - en |
| | quantized_by: TheMelonGod |
| | pipeline_tag: text-generation |
| | tags: |
| | - quantized |
| | - safetensors |
| | - exllamav2 |
| | - phi |
| | - phi4 |
| | base_model: |
| | - microsoft/phi-4 |
| | base_model_relation: quantized |
| | --- |
| | **Orignal Model by:** [Microsoft](https://huggingface.co/microsoft) |
| | **Orignal Model:** [Phi-4](https://huggingface.co/microsoft/phi-4) |
| |
|
| | For more information about the model, I highly recommend checking out the original model page and the creator while you're at it. |
| |
|
| | **ExLlamaV2 Quantizations:** |
| | **8.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-8.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-8.0bpw) |
| | **7.5bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-7.5bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-7.5bpw) |
| | **7.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-7.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-7.0bpw) |
| | **6.5bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-6.5bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-6.5bpw) |
| | **6.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-6.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-6.0bpw) |
| | **5.5bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-5.5bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-5.5bpw) |
| | **5.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-5.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-5.0bpw) |
| | **4.5bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-4.5bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-4.5bpw) |
| | **4.25bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-4.25bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-4.25bpw) |
| | **4.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-4.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-4.0bpw) |
| | **3.75bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-3.75bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-3.75bpw) |
| | **3.5bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-3.5bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-3.5bpw) |
| | **3.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-3.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-3.0bpw) |
| | **2.75bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-2.75bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-2.75bpw) |
| | **2.5bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-2.5bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-2.5bpw) |
| | **2.25bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-2.25bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-2.25bpw) |
| | **2.0bpw**: [8hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/8hb-2.0bpw) | [6hb](https://huggingface.co/TheMelonGod/phi-4-exl2/tree/6hb-2.0bpw) |
| |
|
| | [Measurement File](https://huggingface.co/TheMelonGod/phi-4-exl2/blob/main/phi-4-measurement.json) _(Default/built-in calibration dataset was used)_ |
| |
|
| | If you need a specific model quantized or particular bits per weight, please let me know. I’m happy to help. |
| |
|
| | Your feedback and suggestions are always welcome! They help me improve and make quantizations better for everyone. |
| |
|
| | Special thanks to [turboderp](https://huggingface.co/turboderp) for developing the tools that made these quantizations possible. Your contributions are greatly appreciated! |
