bigscience-workshop/metadata

View on GitHub

Experiments on including metadata such as URLs, timestamps, website descriptions and HTML tags during pretraining.

Stars: 29
Forks: 11
Open beginner issues: 0
Indexed issues: 25

Dominant language: Python
License: Apache-2.0
Last GitHub push: Jun 12, 2023
Latest indexed: Sep 14, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: No beginner labels indexed
PR merge metrics: No merged PRs in 30d

Beginner friendly Everything

Open All

25 open issues indexed

Common simple eval function to calculate ppl Open

bigscience-workshop/metadata#138 · 1 assignee · Opened Feb 23, 2022
multi gpu training Open

bigscience-workshop/metadata#129 · 1 assignee · Opened Jan 21, 2022
Create Dataset with metadata Open

#dataset Epic

Difficulty 4/5 3-5 days Newbie friendliness 25/100

bigscience-workshop/metadata#124 · 3 reactions · Opened Jan 18, 2022
feat: find a solution to load the dataset Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

bigscience-workshop/metadata#122 · Opened Jan 13, 2022
Which HTML tags should be used during training? Open

bigscience-workshop/metadata#113 · 1 assignee · Opened Dec 21, 2021
slurm script test multi-gpu training Open

bigscience-workshop/metadata#109 · 3 assignees · Opened Dec 21, 2021
Evaluation bias Open

bigscience-workshop/metadata#108 · 1 comment · 1 assignee · Opened Dec 21, 2021
Discuss style evaluation for website description and data source with Anna Open

bigscience-workshop/metadata#100 · 1 assignee · Opened Dec 14, 2021
Evaluation toxicity for website description and data source Open

bigscience-workshop/metadata#99 · 1 comment · 1 assignee · Opened Dec 14, 2021
data analysis: website description (quality and yield) Open

bigscience-workshop/metadata#98 · 1 assignee · Opened Dec 14, 2021
Start joint training Open

bigscience-workshop/metadata#97 · 2 assignees · Opened Dec 13, 2021
eval hyperparameters: occupied tokens Open

bigscience-workshop/metadata#96 · 1 assignee · Opened Dec 13, 2021
estimate amount of data Open

bigscience-workshop/metadata#94 · 2 assignees · Opened Dec 13, 2021
eval hyperparameters: amount of metadata Open

bigscience-workshop/metadata#93 · 1 assignee · Opened Dec 13, 2021
method to sample global metadata Open

bigscience-workshop/metadata#92 · 1 assignee · Opened Dec 13, 2021
method to sample local metadata Open

bigscience-workshop/metadata#91 · 1 assignee · Opened Dec 13, 2021
explore hyperparameters: Open

Difficulty 5/5 Over a week Newbie friendliness 10/100

bigscience-workshop/metadata#90 · Opened Dec 13, 2021
simple zero-shot eval function: time stamps Open

bigscience-workshop/metadata#89 · 1 assignee · Opened Dec 13, 2021
simple zero-shot eval function: website description Open

bigscience-workshop/metadata#88 · 1 comment · 1 assignee · Opened Dec 13, 2021
simple zero-shot eval function: datasource Open

bigscience-workshop/metadata#87 · 1 assignee · Opened Dec 13, 2021
simple zero-shot eval function: entity tags Open

bigscience-workshop/metadata#86 · 1 assignee · Opened Dec 13, 2021
simple zero-shot eval function: HTML tags Open

bigscience-workshop/metadata#85 · 1 assignee · Opened Dec 13, 2021
simple zero-shot eval function: generation length Open

bigscience-workshop/metadata#84 · 1 comment · 1 assignee · Opened Dec 13, 2021
feat: refine dependency imports for preprocessing Open

enhancement

Difficulty 4/5 3-5 days Newbie friendliness 35/100

bigscience-workshop/metadata#71 · Opened Nov 25, 2021
feat: save the model and stop training based on `exit-duration-in-mins` Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

bigscience-workshop/metadata#67 · 2 comments · Opened Nov 23, 2021

bigscience-workshop/metadata

Get new issues in your inbox