Agent skills
PyTorch Low-VRAM Block Streaming (Byte-Identical Output)
A model whose weights are bigger than the GPU memory you may use (a shared card, a leased slice, a memory-capped job) can still run at full fidelity if its blocks visit the GPU one at a time.
Inside the ZIP
SKILL.mdscripts/gpu_block_streaming/memory.pyscripts/gpu_block_streaming/shards.pyscripts/gpu_block_streaming/__init__.py
What you get
A model whose weights are bigger than the GPU memory you may use (a shared card, a leased slice, a memory-capped job) can still run at full fidelity if its blocks visit the GPU one at a time. This package does that for any model that runs its blocks in order: transformer encoders, diffusion transformers. The cost is speed; paired runs produce byte-identical artifacts.
- SKILL.md workflow and the supporting files listed in the package.
- Includes eligible updates within this major version. Your purchased version does not expire.
Changelog
First APMSG edition of the pinned skill package.
Package checks
Archive structure and file hashes were checked against the release manifest. These checks are not a guarantee of security.
