docs: add pipeline concepts page explaining build, reduce, and score#152
Open
luciaquirke wants to merge 3 commits intomainfrom
Open
docs: add pipeline concepts page explaining build, reduce, and score#152luciaquirke wants to merge 3 commits intomainfrom
luciaquirke wants to merge 3 commits intomainfrom
Conversation
Adds docs/pipeline.rst with explanations of what each CLI command does, what it produces, when to use each one, and how they fit together. Also includes a note on preconditioner ordering and a worked LESS-style example. Closes #142 Co-authored-by: Lucia Quirke <luciaquirke@users.noreply.github.com>
Collaborator
Author
|
@claude can you specify that |
|
Claude finished @luciaquirke's task in 27s —— View job Tasks
Updated the overview bullet in
to:
|
Co-authored-by: Lucia Quirke <luciaquirke@users.noreply.github.com>
luciaquirke
commented
Feb 17, 2026
docs/pipeline.rst
Outdated
|
|
||
| Do you want to search a gradient index interactively (e.g. per-prompt)? | ||
| ├── Yes → use build + query | ||
| └── No → Do you want to rank a dataset by influence against a fixed query? |
Collaborator
Author
There was a problem hiding this comment.
This question should be Do you want to use full gradients without random projection or preconditioning?
luciaquirke
commented
Feb 17, 2026
docs/pipeline.rst
Outdated
| --unit_normalize \ | ||
| --projection_dim 0 | ||
|
|
||
| The resulting ``runs/scores/scores.npy`` contains one score per training example. |
Collaborator
Author
|
@claude could you please address my two comments? |
|
Claude finished @luciaquirke's task in 40s —— View job Tasks
Both review comments addressed:
|
Co-authored-by: Lucia Quirke <luciaquirke@users.noreply.github.com>
luciaquirke
commented
Feb 17, 2026
|
|
||
| A directory at ``run_path`` containing: | ||
|
|
||
| - ``data.npy`` — a memory-mapped array of shape ``[num_examples, projection_dim]`` |
Collaborator
Author
There was a problem hiding this comment.
the shape is wrong, it's a structured array
luciaquirke
commented
Feb 17, 2026
|
|
||
| - ``data.npy`` — a memory-mapped array of shape ``[num_examples, projection_dim]`` | ||
| (or ``[num_examples, param_dim]`` if no projection). | ||
| - ``indices.jsonl`` — per-example metadata. |
luciaquirke
commented
Feb 17, 2026
|
|
||
| A directory at ``run_path`` containing: | ||
|
|
||
| - ``data.npy`` — a single aggregated gradient vector of shape ``[1, param_dim]``. |
luciaquirke
commented
Feb 17, 2026
|
|
||
| A directory at ``run_path`` containing: | ||
|
|
||
| - ``scores.npy`` — a memory-mapped array of shape ``[num_examples]`` (or |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
docs/pipeline.rstwith explanations of what each CLI command does, what it produces, when to use each one, and how they fit together. Also includes a note on preconditioner ordering and a worked LESS-style example.Closes #142
Generated with Claude Code