Add YaRN rope adjustment + DeepSeekV3 rope_interleave#2202
Open
ysjprojects wants to merge 6 commits intomainfrom
Open
Add YaRN rope adjustment + DeepSeekV3 rope_interleave#2202ysjprojects wants to merge 6 commits intomainfrom
ysjprojects wants to merge 6 commits intomainfrom
Conversation
ysjprojects
requested review from
KaelanDt,
andyland,
k223kim,
lantiga,
lianakoleva and
t-vi
as code owners
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This pull request introduces comprehensive support for YaRN (Yet another RoPE extensioN) rotary position embedding (RoPE) scaling to the codebase, enabling advanced context extension and compatibility with models like DeepSeek V3. The changes include new configuration options, robust parameter validation, updated RoPE cache construction, and new tensor operations to support interleaved RoPE layouts. Additionally, a new test suite is added to validate LitGPT’s YaRN implementation against HuggingFace’s DeepSeek V3.
rope_interleaveboolean and extendedrope_adjustmentsin theConfigclass to support YaRN and DeepSeekV3 specific RoPE logicbuild_rope_cachefunction, including attention scaling computation, frequency blending, and smooth ramping between extrapolation and interpolation regimes.tests/test_yarn.py, a comprehensive test comparing LitGPT’s DeepSeek V3 block with YaRN RoPE scaling against HuggingFace’s implementation.Motivation