Crystalxd

How to test single image or a torch tensor?

1

When I convert the PCB model to .pt(used for c++), I cannot set the batchsize by 1, and I test a tensor such as [1, 3, 384, 192], an error...

belle_7b模型多卡推理结果不对，单卡推理正常

2

 8卡4090推理结果（双卡及以上都是这种乱码的形式）  单卡4090推理结果 以上是使用CUDA_VISIBLE_DEVICES控制卡的数量 请问有可能是什么原因造成的呢？

ARM64 CPU on Linux

2

Can mlc-llm precisely run arm64v8 CPU on Ubuntu? Not just on the Mac M series chip.

[Question] Whether to support running on ARM64v8

3

## ❓ General Questions When I completed the compilation and tried to run `./mlc_chat_cli --local-id ../../dist/vicuna-7b-all-v1.1-q3f16_0`, I got follows error: ` terminate called after throwing an instance of 'tvm::runtime::InternalError' what():...

[BUG] CUDA Out of Memory when eval model.

5

### Required prerequisites - [X] I have read the documentation . - [X] I have searched the [Issue Tracker](https://github.com/baichuan-inc/baichuan-7B/issues) and [Discussions](https://github.com/baichuan-inc/baichuan-7B/discussions) that this hasn't already been reported. (+1 or comment...

logs that inferred custom trained model (LanguageBind + Qwen14B LLM). ```shell (moellava) root@ps:/code/MoE-LLaVA# CUDA_VISIBLE_DEVICES=0 python predict.py [2024-03-18 02:02:14,276] [INFO] [real_accelerator.py:110:get_accelerator] Setting ds_accelerator to cuda (auto detect) /opt/conda/envs/moellava/lib/python3.10/site-packages/torchvision/transforms/_functional_video.py:6: UserWarning: The 'torchvision.transforms._functional_video'...

Fine-tuneing LLM + LanguageBind?

1

How can I combine LanguageBind with LLM to fine-tune my own downstream tasks? Such as Qwen?