-
Flatten.jl provided `constructor_of` for Unitful.jl in requires, because flattening models with units is very common and ConstructorBase didn't exist. I'm going to put in a pull request at unitful to …
-
Hello!
I have reconstructed a model with ~1500 images and resulting point cloud 180k points.
It looks like this in the viewer:
Now as you can see, the frames/cameras sizes are extremely l…
-
Hi @chenhsuanlin , thanks for the great work! I have few questions about your Pascal evaluation. I realize you mentioned in the paper that you applied ICP for Pascal pointcloud comparison but not shap…
-
currently, TPTP contains no consistency checkers. Add those consistency checkers that are available for SoftFOL. It should be relatively straightforward to adapt them for TPTP.
-
Hello, I run the pipeline on my dataset, which contains **18,384 images**. The resolution of the images is **6000*4000**. The total time consumption is **49 hours** on my ubuntu server(Intel(R) Xeon(R…
-
Hi Ellen,
Thanks for sharing the data! I want to check the experiment outputs, and I see in README you write
> The outputs of the experiments (trained models and reconstructed volumes), which ar…
-
Dear Francisco,
Recently, I have tried to use your pipeline to analyse some metagenome data. For the bin, i found I can only get about 30 MAGs of high quality (completeness > 90% by checkM). So I am …
-
- [x] Check roberta's event files
- [ ] Truncate images to remove no reading content
- [x] Preprocess w/ fmriprep
- [ ] Merge events with original dataset
- [x] Check for BIDS complicance
- [x] L…
-
I'm opening an issue following a suggestion from @juliasilge on StackOverflow (the [question](https://stackoverflow.com/q/63372731/4653440), her [suggestion](https://stackoverflow.com/a/63474617/46534…
-
请问一下，在训练完之后，得到了bert分类模型，类别数是self.num_labels。
在测试时，我看代码中是这样处理：先用bert提取测试集的embedding，然后用kmeans聚类，类数是self.num_labels，得到聚类结果以后去和真实标签计算最优匹配。
那么请问测试时为什么不直接对测试集分类呢？直接分类应该也可以得到对应的label