Vijay Viswanathan
Vijay Viswanathan
Hi, Thank you for sharing this dataset with the community! I'm particularly interested in some of the Speaker-control commands other than "PlayMusic", particularly "ResumeMusic", "SpeakerInterrupt" and "VolumeShift". It seems that...
`autofaiss` [currently requires](https://github.com/criteo/autofaiss/blob/1a7b55bad90d56e33f8c6b8fa222e3226429c461/requirements.txt#L4) that `pandas` be installed at a version below 2.0. This is problematic, because many new libraries use pandas>=2.0. Is it possible to change the requirements file to...
Currently, have just moved [the HW1 code from CMU's 11-737 course](https://github.com/neubig/multiling2022-code/tree/main/assign1) in here. In the future, we will: - Add detailed instructions and best practices for improving performance beyond these...
# Description This PR deletes the current "ci.yml" continuous integration workflow and instead creates two workflows, "pull_request_ci.yml" (which runs as a gate for all pull requests) and "daily_ci.yml" (which runs...
We currently have interfaces to interact with prompt2model, in the form of prompt2model_demo.py and prompt2model_demo.ipynb. While the individual components of prompt2model are generally well-tested, these two files are completely untested....
Right now we have an encoded dataset index file, `huggingface_data/huggingface_datasets/huggingface_datasets_datafinder_index`, checked in to the repository. Instead of having a binary in our repo, it would be better to download this...
Our current Prompt2Model pipeline uses a fixed set of hyperparameters for all tasks ([shown here](https://github.com/neulab/prompt2model/blob/0c1f10b52ca093b19a1d4296143b3a03e39f825c/prompt2model/model_trainer/generate.py#L273-L284)). To robustly handle different tasks, we want to implement automated hyperparameter selection by computing metrics...
#34 introduced a prompt parser which uses a long "meta-prompt" to parse a given prompt. Since the "meta-prompt" is long (1.8k tokens), this limits the prompts that can be provided...
In our integration test, we're currently loading a model in the Trainer. this should be mocked to avoid having to download a full transformers model each time this test is...