Non-record: Autoresearch Heads4 + Step-based LR + Sliding Window (1xH100)#344
Open
aryanbhosale wants to merge 1 commit intoopenai:mainfrom
Open
Non-record: Autoresearch Heads4 + Step-based LR + Sliding Window (1xH100)#344aryanbhosale wants to merge 1 commit intoopenai:mainfrom
aryanbhosale wants to merge 1 commit intoopenai:mainfrom
Conversation
Built on SOTA (10L, int5/int6, BigramHash, SmearGate, SWA) with 75+ automated experiments across Mac MLX and 1xH100 CUDA. Key findings: - NUM_HEADS=4 with head_dim=128: -0.095 BPB relative improvement - Step-based LR schedule: -0.483 BPB vs wallclock-based - BigramHash(16384): -0.025 BPB vs 10240 - MATRIX_LR=0.03: -0.003 BPB Tested on 1xH100 (800 steps, 600s). Post-quant val_bpb: 1.2756 with sliding window eval stride=256. Known issue: artifact is 17.4MB (over 16MB) due to head_dim=128 increasing params. Needs int4/int5 MLP compression to fit budget.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Non-record submission exploring automated architecture search for Parameter Golf.
Built on the current SOTA (10L, int5/int6, BigramHash, SmearGate, SWA) with 75+ automated experiments across Mac MLX and 1xH100 CUDA using an autoresearch loop inspired by Karpathy's autoresearch methodology.
Key Findings
Results (1xH100, 800 steps)
Known Issues
Negative Results (also valuable)