Non-record: GDN-Hybrid (Gated DeltaNet + SWA) — val_bpb 1.209735#1553
Open
Abhishek8108 wants to merge 1 commit intoopenai:mainfrom
Open
Non-record: GDN-Hybrid (Gated DeltaNet + SWA) — val_bpb 1.209735#1553Abhishek8108 wants to merge 1 commit intoopenai:mainfrom
Abhishek8108 wants to merge 1 commit intoopenai:mainfrom
Conversation
…09735 Moves GDN-Hybrid to track_non_record_16mb with corrected BPB calculation. Fixes double-count bug in build_sentencepiece_luts (leading-space +1 was counted in base_bytes and again in the eval loop). Corrected 3-artifact mean: 1.209735 BPB (stride=512 rescore of saved artifacts). Refs PR openai#1545.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 introducing the GDN-Hybrid architecture: a Griffin-style hybrid that replaces the transformer backbone with Gated DeltaNet (delta-rule linear recurrence) + Sliding Window Attention.
Corrected val_bpb: 1.209735 (3-artifact mean, stride=512)
Layout:
[GDN×5] → [SWA] → [GDN×5] → [SWA_shared]— 33.86M params, SP1024, int6 GPTQ + zstd-22. No TTT. Fixed predictor.BPB Correction (from closed PR #1545)
The original submission was closed after a double-counting bug was found in
build_sentencepiece_luts: the leading-space byte was included inbase_bytesAND added again conditionally in the eval loop, inflatingbyte_countand producing an artificially low BPB.Fix: remove
+1frombase_bytesto match the canonicaltrain_gpt.py. The training itself was unaffected. The three saved artifacts were rescored with the corrected formula (EVAL_STRIDE=512) to produce the numbers above. Full results inrescore_results.tsv.Compliance
TTT_ENABLED=0