---
license: other
base_model:
- Jackrong/Qwopus3.5-4B-v3
- unsloth/Qwen3.5-4B
private: true
---

# Jackrong/Qwopus3.5-4B-v3-MTP-GGUF

Source model: `Jackrong/Qwopus3.5-4B-v3`
MTP source fallback: `unsloth/Qwen3.5-4B`

Uploaded GGUF variants:

- `Qwopus3.5-4B-v3-MTP-Q2_K.gguf`
- `Qwopus3.5-4B-v3-MTP-Q3_K_S.gguf`
- `Qwopus3.5-4B-v3-MTP-Q3_K_M.gguf`
- `Qwopus3.5-4B-v3-MTP-Q3_K_L.gguf`
- `Qwopus3.5-4B-v3-MTP-IQ4_XS.gguf`
- `Qwopus3.5-4B-v3-MTP-Q4_K_S.gguf`
- `Qwopus3.5-4B-v3-MTP-Q4_K_M.gguf`
- `Qwopus3.5-4B-v3-MTP-Q5_K_S.gguf`
- `Qwopus3.5-4B-v3-MTP-Q5_K_M.gguf`
- `Qwopus3.5-4B-v3-MTP-Q6_K.gguf`
- `Qwopus3.5-4B-v3-MTP-Q8_0.gguf`
- `Qwopus3.5-4B-v3-MTP-BF16.gguf`

The conversion pipeline first verifies whether the source HF model already contains MTP tensors.
If not, it extracts the MTP tensors from the matching unsloth base model and injects them into the safetensors index before GGUF conversion.