Duplicated from  empero-ai/Qwen3.5-9B-Claude-Code-GGUF

Atomic-Germ

/

Qwen3.5-9B-Claude-Code-NPU2

Like 0

hey, i wanted to make a replacement for the qwen3.5:9b model with https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B, but at the end when trying to run the model I end up with

qwen3_5vl_model_def.hpp:396: Assertion w_ptr + w.size() <= proj_weights.end() failed

What I did is:

1. Download the original repository
2. Convert safetensors into main gguf and mmproj gguf
3. Quantize main gguf into Q4_1
4. Use the convert.py script for main.gguf and mmproj.gguf separately and move vision_weights.q4nx where main.gguf lies
5. Use flm-add.py script

Am I missing something?