Samson Tan

Documentation Structure of Projects

2

For all the projects, what do you guys think of categorizing their documentation into 2 types: "Developer/Technical" (setting up dev environment, API docs, etc.) and "Product" (roadmaps, user stories), and...

Formalizing Conventions (Coding, commits, PRs)

11

Since we're growing rather rapidly, @iamtrask and I thought it'd be good to get the discussion going on formalizing some conventions. Here are some to get the ball rolling: ###...

Slow decoding with many special tokens in vocabulary

3

### System Info present across multiple versions ### Who can help? @younesbelkada ### Information - [ ] The official example scripts - [ ] My own modified scripts ### Tasks...

Seq2Seq: Special tokens are also added to targets for LL computation

2

Location: https://github.com/bigscience-workshop/lm-evaluation-harness/blob/master/lm_eval/models/huggingface.py#L460 @jon-tow I'm not sure if special tokens should be included as part of the target sequence when doing the LL computation.

Command Line-based Monitoring of Model Training Progress

4

**User Story**: As a `Data Scientist`, I want to be able to monitoring the training of my published model from my command line. **Acceptance Criteria** * Code should be accompanied...

Submitted gradients should only be accepted if there are at least 10 gradients from another miner in between submissions. **Acceptance criteria** * Code should come with unit tests