prompt2model

prompt2model copied to clipboard

Particular retrieved model fails with proto descriptor error

2

When trying a particular task, the model retriever retrieved this model: `mrm8488/t5-base-finetuned-common_gen`. Just before training could start, p2m crashed with the following error log: ``` TypeError: Couldn't build proto file...

saum7800

Support models that require trust_remote_code

1

Some recent models require Hugging Face's [trust_remote_code](https://github.com/huggingface/transformers/issues/22260) to be set to true. This should be off by default, but we should allow users to set it if they wish. Reference:...

neubig

Show good defaults for parameters

2

In `cli_demo.py`, currently there are no defaults shown for "number of generated examples" and "initial temperature" (there is a number for maximum temperature). We should add defaults for the other...

neubig

Smarter selection of Batch size during hyper-parameter selection and training

2

Prompt2Model currently has a static way of defining batch size. The user has to tweak it into the code to either train models faster. Also referencing this issue #315, the...

Anindyadeep

Step wise resume of prompt2model CLI

3

So right now, when we use the prompt2model cli, it gives the option of whether to start from scratch or not. However, sometimes, it could be the case, where we...

Anindyadeep

Many duplicate datasets given by the dataset retriever

1

viswavi

test_encode_model_retriever() gets stuck while running on Windows

1

When using Windows, pytest gets stuck on this test, and it remains stuck till I dont manually abort the command. @viswavi speculates that maybe Tevatron doesn't support windows - I...

ritugala

prompt2model has several components, including: * [prompt parsing](https://github.com/neulab/prompt2model/tree/5ca28583a2e370bad0535a6a486e6a977ab6c763/prompt2model/prompt_parser) * [data generation](https://github.com/neulab/prompt2model/tree/5ca28583a2e370bad0535a6a486e6a977ab6c763/prompt2model/dataset_generator) * upcoming components in autopilot: https://github.com/neulab/prompt2model/issues/285 Most of these are based on prompting API-based models. It would be cool...

neubig

Tuning of model and dataset retrievers

1

Currently our model and dataset retrievers are not perfect, and it would be good to have a way to make them better. One way we can do so is by...

neubig

tiktoken usage

2

https://github.com/neulab/prompt2model/pull/335#issuecomment-1711578817 Another thing is that I am not sure that all the API-based models use tiktoken to compute input tokens. Thus, the num of input tokens may slightly differ between...

zhaochenyang20