MiniCPM icon indicating copy to clipboard operation
MiniCPM copied to clipboard

MiniCPM4: Ultra-Efficient LLMs on End Devices, achieving 5+ speedup on typical end-side chips

Results 152 MiniCPM issues
Sort by recently updated
recently updated
newest added

### Description / 描述 tech report 有这个实验, 那有对比过这样的效果吗 A0 预训练数据退火 B0 预训练数据+SFT数据退火 A1 预训练数据退火 + 4B sft B1 预训练数据+SFT数据退火 -> 4B sft ### Case Explaination / 案例解释 _No response_

badcase

我在使用你们提供的inference_vllm.py进行推理时,发生了以下错误 ERROR 08-19 09:33:57 pynccl.py:53] Failed to load NCCL library from libnccl.so.2 .It is expected if you are not running on NVIDIA/AMD GPUs.Otherwise please set the environment variable VLLM_NCCL_SO_PATH to...

### Description / 描述 import torch from PIL import Image from transformers import AutoModel, AutoTokenizer model = AutoModel.from_pretrained('./model/OpenBMB/MiniCPM-V-2', trust_remote_code=True) model = model.to(device='cuda') tokenizer = AutoTokenizer.from_pretrained('./model/OpenBMB/MiniCPM-V-2', trust_remote_code=True) model.eval() image = Image.open('./img/tmp.jpg').convert('RGB')...

badcase

### Description / 描述 **代码:** `import torch from PIL import Image from transformers import AutoModel, AutoTokenizer model = AutoModel.from_pretrained('openbmb/MiniCPM-V', trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained('openbmb/MiniCPM-V', trust_remote_code=True) model.eval().cuda() image = Image.open('xx.jpg').convert('RGB') question =...

badcase

### Is there an existing issue ? / 是否已有相关的 issue ? - [X] I have searched, and there is no existing issue. / 我已经搜索过了,没有相关的 issue。 ### Describe the bug /...

bug
triage

### Is there an existing issue ? / 是否已有相关的 issue ? - [X] I have searched, and there is no existing issue. / 我已经搜索过了,没有相关的 issue。 ### Describe the bug /...

bug
triage

### Description / 描述 在使用llama.cpp在安卓编译时,显示'execinfo.h'头文件缺失 ### Case Explaination / 案例解释 _No response_

badcase

### Is there an existing issue ? / 是否已有相关的 issue ? - [X] I have searched, and there is no existing issue. / 我已经搜索过了,没有相关的 issue。 ### Describe the bug /...

bug
triage

输入约1134字,错误信息如下 MLCChat failed Stack trace: org.apache.tvm.Base$TVMError: TVMError: Check failed: value->shape[0] shape[0]

### Description / 描述 错误结果:无法复现基于minicpm模型结构进行缩放,得到不同尺寸下的最优学习率一致性结果 ### Case Explaination / 案例解释 作者您好,minicpm是一个非常棒的工作。我在minicpm开源的模型结构基础上进行基于模型的尺寸缩放实验,但是并不能复现博客中的如下图最优学习率一致性的结论 ![image](https://github.com/user-attachments/assets/91033fdf-25d1-404a-831e-47b8bec1e741) 在技术报告中看到了模型整体的缩放参数,在代码中基本都找到了缩放的位置,但是没找到参数初始化和learning rate部分,请问这部分是如何实现的呢?非常感谢您的回复。 ![image](https://github.com/user-attachments/assets/8d7f57cf-d2f7-42ea-9086-ef3405e96e2f) 在代码中看到是正常的初始化方式 ![image](https://github.com/user-attachments/assets/5e89dcd0-7a15-4dc6-8d7c-6e53cc9d259a)

badcase