Is this model obtained by direct quantification by prithivMLmods/QwQ-LCoT-7B-Instruct?
#1
by
liss0510
- opened
By comparing the model, I found that this new model is completely consistent with the original QwQ-LCoT-7B-Instruct network structure. Now I need to transplant this model to relevant devices, and directly use QwQ-LCoT-7B-Instruct to quantify whether it is feasible.
If you have time, please reply to me as soon as possible.
Yes, the QwQ-4B is the compressed 4B version with 4-bit precision for the QwQ-7B-LCoT.