language	en
tags	azbert pretraining fill-mask
widget	text example_title $f$ $($ $x$ [MASK] $y$ $)$ mathy text example_title $x$ [MASK] $x$ $equal$ $2$ $x$ mathy text example_title Proof by [MASK] that $n$ $fact$ $gt$ $3$ $n$ for $n$ $gt$ $6$ mathy text example_title Proof by induction that $n$ [MASK] $gt$ $3$ $n$ for $n$ $gt$ $6$ mathy text example_title The goal of life is [MASK]. philosophical
license	mit

About

This repository is a boilerplate to push a mask-filling model to the HuggingFace Model Hub.

Checklist

• git-lfs is installed
• tokenizer contains all the files needed: added_tokens.json, special_tokens_map.json, tokenizer_config.json, vocab.txt and tokenizer.json
• no tokenizer_file field in tokenizer_config.json (sometimes it is located locally at ~/.cache)

Upload

1. Put the model checkpoints and optionally log files (*.bin and log files events.out.*) to the ./ckpt directory.
2. Add a branch hgf to point to your huggingface repo. For example git remote add hgf git@hf.co:approach0/mathy-vicuna-13B-FFT
3. Run the upload2hgf.sh script.

Test the MLM task (an example)

pip install pya0 # for math token preprocessing
# testing local checkpoints:
python test.py ./ckpt/math-tokenizer ./ckpt/2-2-0/encoder.ckpt
# testing Model Hub checkpoints:
python test.py approach0/coco-mae-220 approach0/coco-mae-220

Note
Modify the test examples in test.txt to play with it. The test file is tab-separated, the first column is additional positions you want to mask for the right-side sentence (useful for masking tokens in math markups). A zero means no additional mask positions.