Zhou Xin comments

Results 28 comments of


                                            Zhou Xin

If there are raw data for items (original images or texts), rather than extracted features.

Hi, @Du-danger Actually, the raw data can be found at [https://jmcauley.ucsd.edu/data/amazon/links.html](https://jmcauley.ucsd.edu/data/amazon/links.html) And you are free to use other Pre-trained encoders to extract the feature embeddings. Like here: https://github.com/enoche/BM3/issues/9

关于elec数据的问题

嗯，保存的有，应该在google drive那边

关于elec数据的问题

参考这个 https://github.com/enoche/MMRec/issues/8

关于elec数据的问题

> 参考这个 [#8](https://github.com/enoche/MMRec/issues/8) Ok Elec刚查了好像没有，我回头补上！

关于elec数据的问题

@Wmhwxl Hi 你好啊！按照preprocessing的步骤走一下，可以拿到的。应该是跟我之前处理过的一样的。

Ollama Graph Embedding Fails in Local LLM Setup

Marked to follow: error on `no file found at embed output`

Why are the text and image feature inputs treated as unfrozen embeddings?

Hi, @gmeehan96 Thank you for your feedback on this repository. I find your concern intriguing and will look into the issue further. Current code generally follows LATTICE. `https://github.com/enoche/MMRec/blob/master/src/models/lattice.py`

最佳参数不一致

嗯，您好！如果想生成论文中一致结果，请load我们pre-trained的模型，然后直接 evaluation 1、不同的server环境，最佳参数会不一样。 2、不同的python/conda库版本，最佳参数也会不一样。