-
我在WIN11 下,尝试了在CUDA 11.8~12.6 TORCH 2.4 python3.12各种版本环境下编译都失败了,没办法,请问能不能出WINDOWS的wheel能让我安装?
-
### Your question
I'm using the default provided workflows from the comfyui site [here](https://comfyanonymous.github.io/ComfyUI_examples/flux/). Using the regular ones I get a conversion error (as s…
-
Hi there, got very good results from my trainings according to the validation images of Simpletuner but as soon as I try my lora's ( .safetensors ) on a1111 forge with the Flux fp8 version, it's seems…
-
可否加入对FP8量化模型的支持
vllm0.43中加入了FP8量化模型的支持。
我尝试在xin中注册Qwen2-7B-Instruct-FP8这个模型,启动时报错见附件日志
[xin报错日志.txt](https://github.com/user-attachments/files/15845984/xin.txt)
相同环境下,我直接通过vllm命令行方式是可以启动的python -m…
-
`language/test_core.py::test_abs_fp8[in_dtype2]` is failing because of https://github.com/intel/torch-xpu-ops/issues/570. Let's mark it as xfail so we can enable it straight away once `aten::__lshift_…
-
**TL;DR : very hard to tell if training is starting before getting first samples pics ; either it's just not starting or it fails with some error if trying other GPU external settings ; seems to work …
-
I've updated ComfyUI, and I installed the latest CogVideoXWrapper through ComfyUI manager via this Git's URL. I've loaded the "cogvideox_5b_example_01.json" workflow, and pointed the Load Clip node to…
-
Dear Community,
I have gone through numerous resources and carefully examined the relevant code. However, there is one aspect that I find particularly perplexing and I would greatly appreciate the co…
-
### System Info
x86_64
755G
nvidia T4
ubuntu 22.04
trtllm version : https://github.com/NVIDIA/TensorRT-LLM/archive/9691e12bce7ae1c126c435a049eb516eb119486c.zip
pip install tensorrt-llm==0.11…
-
When testing Triton on the server with the NVIDIA GeForce RTX 2080 Ti GPU, running python/tutorials/06-fused-attention.py, line 458 throws the following CompilationErrror:
```
triton.compiler.errors…