Skip to content

docs(readme): the muse-glimmer fixes: format waits for the header, glimmer's KV is priced, explicit image budgets and draft_num_predict on MLX - #474

Merged
glennneuber merged 1 commit into
docs/readme-0402from
docs/readme-glimmer-fixes
Oct 13, 2026
Merged

glennneuber merged 1 commit into
docs/readme-0402from
docs/readme-glimmer-fixes

Conversation

@glennneuber

Copy link
Copy Markdown

The README follows the muse-glimmer fixes. Stacked on #461; merge it after #461, #464, #465, #466 and #468.

row was now follows
format on muse-glimmer (new) — the format waits for the message header in one pass and in two #464
MLX admission — a config.json the estimator cannot parse is named in the warning; glimmer's was admitted on weights alone #466
image_max_tokens, MLX glimmer "reads 1120 (the default) as unset, so 4096" unset: 4096; an explicit 70 or 1120 is honoured (ADR 0048) #468
draft_num_predict, MLX "not read" per request, no reload: 0 off, N caps the depth #465

#470 carries its own README change (the whitespace row). README only; nothing else changes.

Signed: macbook-pro-m5-max-128GB/mlx-metal

🤖 Generated with Claude Code

…immer's KV is priced, explicit image budgets and draft_num_predict on MLX

Follows #464, #465, #466 and #468; stacked on #461.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant