Ritian-Li

torch版本loss降不下去，另外经常报error calling function

5

1. loss 基本保持在5.1-5.8之间 2. 出现的回复都是毫无章法的话   3. 使用测试案例“宝塔镇河妖”，显示error calling  求解1！！

why self.relu(x + 3) / 6 ?

3

https://github.com/Andrew-Qibin/CoordAttention/blob/8ab857a3b3a149dfe1ce018a90c5c23816bbf860/coordatt.py#L12

cannot import name 'load_pretrained'

2

from timm.models.resnet import default_cfgs, load_pretrained, BasicBlock, Bottleneck ImportError: cannot import name 'load_pretrained' and in timm.models.resnet , there is no ‘load_pretrained’ Can someone tell me how to do

Can this work on Chinese to English?

1

I've tried, but at the step of building binary dict, but too much Chinese word replaced by unk word.

planner可以正常出结果，但是_execute_agent_step中报错400

5

感谢开源，报错信息如下： ```bash ================================== Ai Message ================================== Name: planner { "locale": "zh-CN", "has_enough_context": false, "thought": "用户想要了解大模型时代的技术发展情况，需从多个方面进行研究，以呈现全面且深入的技术发展态势。", "title": "大模型时代技术发展的研究计划", "steps": [ { "need_web_search": true, "title": "大模型时代技术发展的历史与现状研究", "description": "收集大模型从起步到当前阶段的关键时间节点及标志性技术突破；各阶段代表性大模型的参数规模、训练数据量、应用领域等数据；当前主流大模型在自然语言处理、计算机视觉等主要领域的性能指标与应用成果。", "step_type": "research" }, {...