adding report-to-wandb - #9
Merged
Merged
Conversation
Goekdeniz-Guelmez
changed the title
Contributor
Author
|
@awni should be mergable now. |
Contributor
|
It would be great having this integrated OOB 🙏 |
Contributor
Author
|
@awni should be mergeable now! |
angeloskath
requested changes
Apr 5, 2025
angeloskath
left a comment
angeloskath
left a comment
Member
There was a problem hiding this comment.
Hey @Goekdeniz-Guelmez thanks for adding that!
I think it would be better to move this out of lora.py. I would do the following changes
- Add a file
mlx_lm/tuner/callbacks.py - Add
WandBCallbackthere- Use
try: import wandb; except: wandb = Nonepattern instead ofif - Initialize
wandbin the callback constructor - Accept the project name, config and optional wrapped callback in the callback constructor
- Use
- Change the
--report-to-wandbto--wandbor--wandb-projectand allow setting the project or the defaultNoneto disable the callback
- Updated WandB reporting mechanism to use a project name argument instead of a boolean flag. - Removed the old TrainingCallback class definition from trainer.py and imported it from callbacks. - Adjusted argument parsing to accommodate the new WandB configuration.
- Added log_dir parameter to WandBCallback constructor for specifying the logging directory. - Updated lora.py to pass adapter_path as log_dir when initializing WandBCallback.
Contributor
Author
|
Hey @angeloskath your suggestions are really helpful! I did what you said and it works as expected. Also the run config and data files are save in the |
Contributor
Author
|
You can now call it like |
Contributor
Author
|
Hey @awni, this should be mergable now! If you need me to fix or change something, just ping me. |
awni
reviewed
May 9, 2025
awni
reviewed
May 9, 2025
awni
reviewed
May 9, 2025
Member
|
This looks nice! Just a couple minor comments. Could you address then we can merge? |
Contributor
Author
|
Thanks! They have been addressed! |
awni
approved these changes
May 10, 2025
Member
|
@angeloskath you have to approve this in order to land it. |
nitky
pushed a commit
to nitky/mlx-lm
that referenced
this pull request
May 16, 2026
Round 2 fixes from cross-model adversarial review (B/C strata): - B1: drop private workspace path from module docstring - B2: invert network-test polarity (MLX_LM_RUN_NETWORK_TESTS=1 opt-in, default-skip in CI); use public huggingface_hub.snapshot_download instead of private _download; rename test to describe behavior - I1: remove unused List import - I2: rename quant_predicate's _module to _ (matches sibling style) - I3: pass position_ids[-1] as mx.array offset to RoPE (no .item() sync; MTP hot path was paying a GPU->host roundtrip per forward) - I4: build MaskedEmbedder fill on-device (no .item() sync inside the per-draft-step path) - ml-explore#6: rename masked_embedding.token_ordering -> _token_ordering so the int32 gather buffer is excluded from Module.parameters(); install via Model.sanitize side-channel so load_weights (which rejects underscored keys) doesn't trip; eliminates the fragile post-tree_map int32 restore in the synthetic Tier 1 test - ml-explore#9: make_cache() now raises NotImplementedError instead of returning [] (zip-based cache iteration would silently no-op the previous return) 77 model tests pass (10 gemma4 tests + 1 opt-in real-weight test pass).
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
4 participants
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.
No description provided.