SeedVR2-7B (MLX) — fp16

Runtime-agnostic MLX-format weights (fp16) for SeedVR2-7B, ByteDance's one-step diffusion super-resolution / restoration model (ICLR 2026), for on-device upscaling on Apple Silicon.

Not tied to any single package — these load into:

• mflux (Python MLX, actively maintained; also the parity reference these weights were validated against),
• seedvr2-mlx-swift (MLX-Swift; archived/read-only since Jun 2026 but functional — MIT-licensed and forkable),
• or any MLX code that reconstructs the same module tree (see Format notes below).

int8 variant: SeedVR2-7B-mlx-int8 · sharp checkpoint: SeedVR2-7B-sharp-mlx · 3B family: mlx-community/SeedVR2-3B-mlx

• Files: transformer.safetensors (DiT, fp16, ~16.5 GB) · vae.safetensors (3D-causal-conv VAE, fp16) · pos_emb.safetensors (precomputed text embedding, 58×5120) · config.json.
• Architecture (vs 3B): vid_dim 3072 (2560), 24 heads (20), 36 layers (32), all layers multimodal, plain MLP (SwiGLU), rope_dim 64.
• Precision: fp16. Transformer reload vs the mflux reference (CPU): t_out max_abs 0.00e+00 (bit-exact). End-to-end GPU vs CPU pipeline: 60 dB PSNR.
• Memory: ~17 GB weights resident; recommended 32 GB+ unified memory (below that, use the int8 variant).

Usage — Python (MLX / mflux)

import json, mlx.core as mx
from mlx.utils import tree_unflatten
from mflux.models.seedvr2.model.seedvr2_transformer.transformer import SeedVR2Transformer

cfg = json.load(open("config.json"))
tx = SeedVR2Transformer(**cfg["transformer_overrides"])
tx.update(tree_unflatten(list(mx.load("transformer.safetensors").items())))
mx.eval(tx.parameters())   # bit-exact vs the mflux reference (verified at export)

The full pipeline (VAE, scheduler, pre/post-processing) lives in mflux: mflux-upscale-seedvr2 --model seedvr2-7b --image-path input.png --resolution 2x (note: mflux's built-in downloader fetches the PyTorch source weights and converts on the fly; loading these pre-converted files uses the snippet above).

Usage — Swift

import SeedVR2MLX   // github.com/xocialize/seedvr2-mlx-swift (archived/read-only, MIT — fork to maintain)
let upscaler = try SeedVR2Upscaler(directory: weightsDir)   // downloaded from this repo
let out = upscaler.upscale(processedImage: img, seed: 42)   // [-1,1], dims padded to /16

Format notes (for other MLX runtimes)

• Key naming: mflux module hierarchy, flattened with mlx.utils.tree_flatten (e.g. blocks.17.attn.proj_qkv_vid.weight). Deterministic mapping back to ByteDance's original PyTorch names: mflux src/mflux/models/seedvr2/weights/seedvr2_weight_mapping.py.
• Layouts: MLX conventions throughout — VAE conv weights are (O, *K, I).
• Config: config.json["transformer_overrides"] carries the 7B dims (vid_dim 3072, heads 24, num_layers 36, mm_layers 36, rope_dim 64, …) and must be passed to the transformer constructor.
• Conditioning: pos_emb.safetensors (58×5120, fp16) is the precomputed embedding of the fixed prompt — the text encoder is eliminated from this port, so it is a mandatory txt input.

Provenance & license

Chain: ByteDance Seed — SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training (ICLR 2026, arXiv:2506.05301), ByteDance-Seed/SeedVR, Apache-2.0 → PyTorch fp16 redistribution numz/SeedVR2_comfyUI (seedvr2_ema_7b_fp16.safetensors; independently verified bitwise against ByteDance's original fp32 seedvr2_ema_7b.pth — all 1128 tensors identical after fp32→fp16 cast) → MLX reference impl filipstrand/mflux → export via xocialize/seedvr2-mlx tooling. These are format-converted weight artifacts (not a new model); Apache-2.0 applies. Credit ByteDance Seed (original), cite the paper.