ml-explore / ml-explore/mlx-examples

Command-R-Plus, Context Window Limitations

Open
#660 42 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
9k
Forks
1.2k
PR merge metrics
No merged PRs in 30d

Description

Cohere's new Command-R-Plus model reportedly features a 128k context window. However, testing with progressively longer prompts reveals it begins producing nonsensical output (e.g., "<PAD><PAD>...") after 8192 tokens, aligning with the "max_position_embeddings" value in the config.json file. The config also lists a "rope_theta" value, suggesting its role in achieving the large context window. Is "rope" supported in MLX?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by comparing the Command-R-Plus config.json values, especially max_position_embeddings and rope_theta, with the model-loading path used by the examples. Reproduce the progressively longer prompts and determine whether the model remains coherent beyond 8192 tokens; done requires a decided approach for the reported context-window limitation.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.