Sanbu
**Describe the bug** Hello, I came across some usage instructions for the qwen model in Jupyter Notebook(254-llm-chatbot), but it seems that there is no specific guide for using qwen 1.8B....
Hi, I want to run the model on fp16, but I found it will shows extremely precision loss, But if I run the model on bf16 , It works well.
 Hi,I really want to use this software, but I still can't compile it, It always says that "Found LLVM 3.8.0 in /usr/local",but i can't find anything about LLVM 3.8.0...
The inference speed of the H model for segmenting anything feels relatively slow. I believe there should be common optimization techniques for transformers. Are there any reference materials available EnergonAI...
imgaug 年久失修,若imshow需要修改源码: python3.10/site-packages/imgaug/imgaug.py ```python fig, ax = plt.subplots(figsize=(w, h), dpi=dpi) fig.canvas.set_window_title("imgaug.imshow(%s)" % (image.shape,)) # cmap=gray is automatically only activate for grayscale images ax.imshow(image, cmap="gray") ``` 其中需要把 fig.canvas.set_window_title 修改为 fig.canvas.manager.set_window_title
图裂了: 在文中部分: 比如下面这个图： https://datawhalechina.github.io/thorough-pytorch/%E7%AC%AC%E5%9B%9B%E7%AB%A0/figures/shortcut2.jpg 左：VGG-19模型，作为参考。 中：一个有34个参数层的普通网络。 右：一个有34个参数层的残差网络（即resnet34）
I want to ask if SWE support use local llm inference backend? because gpt4 will be expensive in multiturn comversation, maybe we can use the local llm to instead of...
Hello friends, I really like the functionality of this program. I encountered an error during the running process:（This problem only occurs after using pycallgraph） `pycallgraph graphviz -- ./mytest.py ` This...
Hi, I would like to explore the model training in BF16 to high performance inference, the model only can run in bf16, if change it to fp16 it will out...
like fp32/fp16 , fp32/bf16. Thank you very much .