Rickey Bowers Jr.

These build systems are more complex than the programming languages, lol. I'm building the extensions manually. Like the only thing needed to build raymarching is: ``` nvcc -I%BASE%\torch\include -I%BASE%\torch\include\torch\csrc\api\include --expt-relaxed-constexpr...

@flobotics the only way I see that working is if you have build tools on your path for the current user. @pureexe said he was using VS2022 command prompt -...

Found the problem. It's a Windows specific compiler error in NVCC: https://github.com/pytorch/pytorch/issues/69460 This should impact almost all projects using torch\pybind11. Upgrading to CUDA 11.8 solved the two errors I was...

I start with a PowerShell VS2022 command prompt as @pureexe suggested, then I bring my ```python``` into the environment. I can check with ```python --version```. I can also check CUDA...

VS2019 is not honoring the path settings for debugging. I tried a number of settings but just get strange errors. So, I've resulted in using the following batch file in...

For loading from a physical/network drive resharding the larger models to a single file might help, imho. Whereas loading multiple files in parallel would be slower. https://github.com/jankais3r/LLaMA_MPS/blob/main/reshard.py

(comment is more a #202 thing) This is the way I was thinking about it: After the model loads and the prompt is tokenized, create a hash of the context....

I did not intent to expand the meaning of the thread. (2) should probably be addressed elsewhere. #202

https://github.com/basujindal/stable-diffusion/blob/e2aef31f3e1a9b9297a0fc9fbb02a90308c98699/optimizedSD/optimized_txt2img.py#L242 ... if you change the above line, removing the sorting ... ```python data = list(chunk(data, batch_size)) ``` ... you'll get the desired effect. Just gather your prompts in a...

Well, I almost had it. That'll go through the prompts ```opt.n_samples``` times! Change the ```else:``` condition to the following: ```python else: print(f"reading prompts from {opt.from_file}") with open(opt.from_file, "r") as f:...