-
I'm working in a hierarchical multi class problem, and if I flat the labels (flat approach) I have about 1193 classes, which perhaps can already be consider a extreme multi classification problem. F…
-
### Feature : Create Dataset Pipelines
from raw "documents" / nodes / text (and other modalities?)
create NER / QnA pairs / Etc synthetically
### Tasks
- [ ] create NER end-to-end pipeli…
-
I have gone through the code didn't find where the itr_forward function was called
-
Thank you very much for providing this code! I’m interested in understanding the samples_per_label parameter for each task. Specifically, I’d like to know how this parameter is set for each dataset. I…
-
It seems like (from the paper) that the current setup for pair classification relies upon computing similarity metrics for the pairs embeddings and then decide on a _binary_ threshold. Naturally for t…
-
Firstly, thank you for providing code for the rest of your experiments!
May I ask if there are any plans to release code for replicating the FairFace results of the paper? If not, could I at least …
-
### Background
### Solution
DS-MVP-V3 An extended machine learning model that can learn to detect racially biased expressions in context based on input lableled data, without having to specify…
-
Add option to manually classify a pair of source files as plagiarism:
- classification : unclassified, plagiarism, innocent
- confidence: scale from 1 (weak evidence) to 5 (strong evidence)
- evi…
-
- [ ] [[2202.12837] Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?](https://arxiv.org/abs/2202.12837)
# [2202.12837] Rethinking the Role of Demonstrations: What Makes In-…
-
According to your paper you use a large batch size of ~32k samples which means that the raw untrained network initially has a chance of ~1/32k of predicting the correct pair.
I am wondering, how t…