Amit Moryossef
Amit Moryossef
### Bug description When training a model with `--valid-metrics chrf ce-mean-words bleu-detok` it seems like the development set is translated fully twice, in order to calculate the `chrf` and `bleu-detok`...
When using `FSQ` with `[8, 5, 5, 5]` levels, and in `pytorch-lightning` specifying `bfloat16` training, the codebook utilization scratches 50% from below, while when training with `float32` it scratches 100%....
In the examples, you report `f"active %: {indices.unique().numel() / num_codes * 100:.3f}"`, for example: https://github.com/lucidrains/vector-quantize-pytorch/blob/master/examples/autoencoder_fsq.py#L76 But this does not generalize when the `num_codebooks` parameters is set to 2 or more....
### [REQUIRED] Environment info **firebase-tools:** 13.1.0 **Platform:** macOS, Ubuntu ### [REQUIRED] Test case ```bash # Step 1: Create a new directory mkdir my-firebase-project cd my-firebase-project # Step 2: Initialize Firebase...
### Problem We would like the system to work on arbitrarily long texts, however the longer the text, the worse the SignWriting translation output. ### Description Instead of sending the...
### Current Behavior In https://sign.mt, https://research.sign.mt etc, we use Google Analytics to track visits, sessions, page views, and events. We do not ask the user for informed consent to collect...
https://web-check.xyz/results/https%3A%2F%2Fsign.mt reports a few issues with the website. It is worth looking over the issues and making sure we are compliant on all relevant ones. Started fixing with: - [x]...
### Problem We have no mechanism for users to provide direct feedback on wrong translations. ### Description When translating from a sign language video to spoken language text, when uploading...
### Problem Users miss announcements when the app is not open or listening. ### Description Use BLE to activate notifications and listening mode for announcements. Tapping the notification displays the...