RAM requirements?
i tried running this locally with 16GB RAM, intel i7 11th, geforce RTX 1060 Ti and got BSOD with error message MEMMORY_MANAGEMENT_ERROR
there was a there was a split second of an warning message of some github repo that was required for a better loader. does anyone know which one that is so i can test if that's the issue.
if thats not the issue id like to know if someone with similar specs can run this model and how they did this.
i tried running this locally with 16GB RAM, intel i7 11th, geforce RTX 1060 Ti and got BSOD with error message MEMMORY_MANAGEMENT_ERROR
there was a there was a split second of an warning message of some github repo that was required for a better loader. does anyone know which one that is so i can test if that's the issue.
if thats not the issue id like to know if someone with similar specs can run this model and how they did this.
Hello, it is best and most convenient to run it through LM Studio. I have 32 GB of RAM and 4 GB of video memory, and I use the Q4_K_M 35B A3B options. 21 GB + the context window for communicating with AI takes up space. Overall, I have 8-10 tokens per second.
In your case, considering the density of the model, the generation speed will be low. You can try the IQ4_XS version at 14.8 GB or Q3_K_L at 14 GB. 6 GB can be offloaded to the video card, and the rest will go to RAM.
i tried running this locally with 16GB RAM, intel i7 11th, geforce RTX 1060 Ti and got BSOD with error message MEMMORY_MANAGEMENT_ERROR
there was a there was a split second of an warning message of some github repo that was required for a better loader. does anyone know which one that is so i can test if that's the issue.
if thats not the issue id like to know if someone with similar specs can run this model and how they did this.
It would be convenient to use GPT-OSS without censorship on your PC. From HauHauCS User.
Alternatively, use the QWEN 3.5 9B Q8_0 or Q6 dense model, which will fit well with your PC.
or use quant with a small bit per weight. IQ3_XS 27B but the results with such quant may be worse.
i tried running this locally with 16GB RAM, intel i7 11th, geforce RTX 1060 Ti and got BSOD with error message MEMMORY_MANAGEMENT_ERROR
there was a there was a split second of an warning message of some github repo that was required for a better loader. does anyone know which one that is so i can test if that's the issue.
if thats not the issue id like to know if someone with similar specs can run this model and how they did this.
Here is an example