Yerong Li

GLM-iprompt 10B output nonsense

1

``` 9, 1701, 50238, 30995, 27545, 30001, 41489, 4474, 6883, 20045, 5147, 12213, 28588, 15418, 40812, 29381, 47872, 25410, 46473, 1863, 20409, 49002, 15781, 45629, 2637, 23811, 6320, 15177, 15984,...

I am not sure whether there are scripts running for smaller model as `GLM-2b` and `roberta-large`? `model_glm_roberta_large.sh` exists

Could you help us know the specifc version in `requirements.txt` in the folder `engliship`?

How to quantize open-Flamingo? ## Expected Behavior https://github.com/open-mmlab/Multimodal-GPT https://github.com/open-mmlab/Multimodal-GPT/blob/main/mmgpt/models/open_flamingo/builder.py This open_flamingov1 builder did successfully fine-tune with lora. I have a similar builder.py trying to do qlora: ``` ### """Modified from...

微信里面加载不出来网页

1

务必注意下面的事项！ # 阅读主页文档！ **主页的文档里面，尤其是其中【常见问题】一节。** # 描述清楚问题信息 bodhi Linux, Moksha desktop 我的微信里面加载不出来网页 - 网页： 点击聊天记录网页打不开 - 微信搜索结果： 不能搜索公众号 - 扫描公众号二维码加载不出来  

In **FlamingoLayer** the media_location is None. ``` class FlamingoLayer(nn.Module): def __init__(self, gated_cross_attn_layer, decoder_layer): super().__init__() self.gated_cross_attn_layer = gated_cross_attn_layer self.decoder_layer = decoder_layer self.vis_x = None self.media_locations = None self.only_lang_x = False ```...

In the paper you said the following. How to do quantization for Open-Flamingo? ``` Thanks to better generalization, it also achieves good quantization performance for instruction-tuned LMs (e.g., Vicuna) and,...