feat(module): multimodal models infrastructure (DEP-0011) - #43
Open
l1ghtsource wants to merge 1 commit into
Open
feat(module): multimodal models infrastructure (DEP-0011)#43l1ghtsource wants to merge 1 commit into
l1ghtsource wants to merge 1 commit into
Conversation
Adds multimodality as a first-class entity: - MediaSegments + ModalityEncoder trait, MultimodalSequenceInput pipeline input - merge_media_embeddings with the empty-media convention - vision blocks: patch embedding, interpolated position embedding, 2D rope, packed varlen attention, GeluMLP, VisionBlock, SpatialPatchMerger - MultimodalRotaryEmbeddingProvider (interleaved/chunked MRoPE) - VarlenSdpaBackend protocol + FA4/FA2/torch implementations - dataset helpers: compute_multimodal_position_ids, pad_empty_media - fix: probe flash_attn parent before flash_attn.cute in SDPA auto-detection Verified against HF Qwen3.5-MoE reference modules (vision tower fwd/bwd parity, MRoPE parity, position ids bit-exact vs get_rope_index).
l1ghtsource
marked this pull request as draft
July 30, 2026 13:15
l1ghtsource
marked this pull request as ready for review
July 30, 2026 13:15
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds multimodality as a first-class entity:
Verified against HF Qwen3.5-MoE reference modules (vision tower fwd/bwd parity, MRoPE parity, position ids bit-exact vs get_rope_index).