grandxin

感谢作者的详细讲解！有两个问题希望了解一下： 1. resnet的实现里： layers.append(ResidualBlock(inchannel, outchannel, stride, shortcut)) for i in range(1, block_num): layers.append(ResidualBlock(outchannel, outchannel)) return nn.Sequential(*layers) 这里在循环内部为啥没有shortcut？ 看resnet的原图似乎每个block都有shortcut的啊。 2. residual block的实现里： def forward(self, x): out = self.left(x) residual = x...

Qwen1.5-4b and Qwen1.5-7b model cannot be loaded correctly in ipex-llm version 20240522

9

I save qwen1.5-4b and 7b int4 model in my computer, when loaded these models, there are some errors: Some weights of the model checkpoint at ./models/qwen1.5-4b were not used when...

Process exit when I use "Pyinstaller" to package and run the Python demo code

2

when I use Pyinstaller to package and run the Python demo code, The .exe process exit when `model = model.to('xpu')` There is no problem running the demo directly using Python....

Error response from daemon: Head "https://quay.io/v2/coreos/etcd/manifests/v3.5.5": Get "https://quay.io/v2/auth?scope=repository%3Acoreos%2Fetcd%3Apull&service=quay.io": EOF 请问这是怎么回事？ 另外请问下，backend的千问7B是1代还是2代？

Result is wrong when running Qwen2-1.5B-Instruct on Intel NPU

3

Just follow the example https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/example/NPU/HF-Transformers-AutoModels/LLM/generate.py when I use load_low_bit=sym_int4, the result is wrong. -------------------- Output -------------------- system You are a helpful assistant. user 什么是电子竞技 assistant League League League trail trail...

hi, I didn't see any relevant examples about step2 or step3. It seems to update svd_u and svd_v, but not LORA.