Wondering about how to add my dataset/task #3036
-
|
Hi I have some questions regarding how I should add my dataset/task which consists of Historical Swedish patent documents roughly from the year 1885 to 1972. The task is taking the patent claims from the document and making multi-label classification prediction to the cooperative patent classification label. This dataset was curated for my thesis which made prediction at two level subclass and group. Furthermore I also experimented with using the documents older classification symbol that was in another older system alongside the patent claims, thus in total I have 4 different files for my dataset how should I add them ? should I created a separate dataset in hugging face for each one and in that case does that I mean I have to define 4 tasks in mteb ? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Great to see you here @Atheer2104, yes you indeed need to upload the datasets to Huggingface You can see an example of how a dataset is structured here as well as how the resulting task is implemented here. As you see, the task implementation is primarily a HF dataset and some metadata. |
Beta Was this translation helpful? Give feedback.
-
|
Alright this was a clear example, thanks ! |
Beta Was this translation helpful? Give feedback.
Great to see you here @Atheer2104, yes you indeed need to upload the datasets to Huggingface
You can see an example of how a dataset is structured here as well as how the resulting task is implemented here. As you see, the task implementation is primarily a HF dataset and some metadata.