feat: Initial block diffusion strategy support - #7
Merged
Merged
Conversation
… into fast_dllm_v2
…2f_engine` into `diffuserve/legacy`
refactor: renaming the project to diffulex (diffusion + flex)
…nager, ModelRunner, Scheduler, and Sequence classes
…h diffulex; adjust launch configurations and import paths
fix: fix the launching errors of refactored d2f strategy
…ed parameter in model_runner
Contributor
Author
10 tasks
…ment dllm prefill/decode tilelang kernels
- Add SDARConfig in diffulex/model/config/sdar/configuration_sdar.py - Implement SDARForDiffusionLM with native Diffulex KV cache integration - Register SDAR model to AutoModelForDiffusionLM - Update model __init__.py to trigger SDAR registration
- Add test_sdar_dvllm.py in examples/ - Support automatic weight key conversion from HF format to Diffulex-native - Test SDAR model forward pass with minimal attention metadata setup - Use .venv Python environment for execution
…ionality; fix launch configurations and improve kernel integration; test: add dllm_flash_attn_decode_kernel test file
…tn prefilling kernel; refactor: renaming imgs to assets
drewjin
changed the title
…ipt bug fixed, all test passed.
…remove redundant assertions and enhance comments for clarity; update test to utilize einops for tensor rearrangement
…t; refactor: Move diffulex_legacy to a separate module to prevent import conflicts;
feat: add SDAR model support for fast_dllm_v2 branch
…update model cache class definition
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Motivation
This PR introduces a flexible framework to support multiple inference paradigms for Diffusion Language Models (dLLMs). Specifically, it implements a Strategy Registration Mechanism, which decouples the decoding logic from the model architecture. This design allows researchers and developers to rapidly prototype, register, and experiment with novel inference algorithms without modifying the core system.
Implementation Status
Features included in this PR:
ShiftingandNon-Shifting, and inherit model-specific samplers based on this.Features to be implemented in future PRs:
Our From-Scratch Diffusion Language Models
Related Issues
#5
#10
Contributors