HN 提问:有人愿意回答一个关于机器学习的小调查吗?
1 分•作者: aunekiaru•28 天前
所以……既然你不是大型语言模型(我希望如此!),就不需要太认真地对待,也不需要回答每一个问题,请选择回答以下任何一个问题:
1. 你是否编程/开发机器学习应用?
2. 如果是,你使用什么语言:Python 还是……其他?
3. 你使用的架构是基于 Transformer 的还是其他?
4. 你是自己从头开始训练你使用的模型吗?
5. 你是否熟悉这些比较“冷门”的机器学习相关术语:softmax、卷积、HDC、QKV、SSM、Yudkowsky(哈哈)、隐藏状态衰减、余弦相似度、交叉熵损失?
谢谢!
查看原文
so...since you aren't LLM (i hope!), there's no need to take this seriously and try to address every single question so please reply to any of these: 1. Do you program/develop ML apps? 2. if yes what language do you use: python or...else? 3. Is the architecture you're using transformers based or something else ? 4. do you train the model you're using from scratch? 5. are you familiar with some of these exotic ML-related keywords: softmax, convolution, HDC, QKV, SSM, yudkowsky (lol),hidden statem gradient decay, cosine similarity, cross entropy loss? thanks!