Feat/scaling laws notebook setup #612
Open
+1,214
−720
Conversation
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
This commit completes the setup of the "scaling_laws.ipynb" notebook
as per the issue requirements.
Key changes include:
- Creation of a dummy "gutenberg_poetry.txt" and its processed
version in "data/gutenberg_poetry/" for initial experimentation.
- Added a notebook cell to calculate and display non-embedding
parameters (N) for predefined GPT models, following Chinchilla
guidelines (N = Total - WTE - WPE, accounting for tied lm_head).
- Added notebook cells to derive the formula for training steps (S)
and to calculate S and total tokens (D) for various model sizes
and compute budgets.
- Replaced the original Task 3 placeholder with a comprehensive
markdown guide detailing how to:
- Perform model training using `train.py`.
- Record final validation losses.
- Plot L vs. N.
- Extract N_opt and D_opt.
- Fit scaling laws (N_opt vs. C, D_opt vs. C) to derive
parameters N0, a, D0, and b.
The notebook is now structured for you to replace the dummy
dataset, run the preparatory calculations, and follow the guide
to perform the full scaling law analysis.
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Outdated suggestions cannot be applied.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.
No description provided.