zwukong
zwukong
Yes,pbr supporting is really amazing~ 3A game now can use AI models.looking forword to the pbr model~
> wow this is a billion times difference yeah ~
> It would help if you added what models you're using. all tried,similar
I think it supports all platforms, no requirements.txt 😄
I think flashpack is just a tensor loader. model need to be transformed to model.flashpack first,then loaded from their own api,that 's all. PS,they are working on the offload feature😄
I think it will end up like a 'GGUF loader ' node, using the same way like gguf, convert to gguf first then load gguf 😄.Almost all my models are...
**I find that their codes still need ram，but need much less ram,very ram friendly. they really have done a lot of optimizations** 使用 np.memmap 直接映射文件（零或低拷贝读取），避免反复的 Python 层缓冲与解析开销。memmap 允许按切片读取连续区块。 预计算合理的 chunk...
because of you i tried new params thinking may be get some boost, however many exceptions and oom then come . I have to revert to a version a month...
may be you are right ,i will try to use a newest packaged version
i think it is the pytorch's problem, 2.8 uses more vram. i updated before and oom, so i had to revert to 2.6