Skip to main content

Placeholder for BitPolar Fabric — KV cache compression plugin for vLLM. Real release coming soon.

Project description

fabric-kv

⚠️ 0.0.1 is a name-reservation placeholder. The real release is in active development.

BitPolar Fabric is a KV cache compression and transfer layer for LLM inference. The fabric-kv Python package will ship as a vLLM plugin delivering 12.8x KV cache compression at 0.96 quality on Llama-3, with a one-line install:

pip install fabric-kv

Status

Phase Status
Algorithm validated on A100 (833K vec/sec, 12.8x, 0.96) ✅ Done
Wire format v1 freeze 🚧 In progress
PyO3 bindings ⏳ Planned
vLLM plugin ⏳ Planned
Public launch ⏳ ~5 weeks

Track Progress

License

Apache-2.0

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fabric_kv-0.0.1.tar.gz (6.3 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

fabric_kv-0.0.1-py3-none-any.whl (6.5 kB view details)

Uploaded Python 3

File details

Details for the file fabric_kv-0.0.1.tar.gz.

File metadata

  • Download URL: fabric_kv-0.0.1.tar.gz
  • Upload date:
  • Size: 6.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.0

File hashes

Hashes for fabric_kv-0.0.1.tar.gz
Algorithm Hash digest
SHA256 3a48b8b4ca7b4e7ad8d82418cd6e6ac2ef60a7f54a393c52c7a874bc3ac47076
MD5 0b687cef7aa9305fc3c10ddae014376b
BLAKE2b-256 e63919c2a7ba40cad0aad0dad04390591c6718547f77d45fac4ff2a6ff8cf3de

See more details on using hashes here.

File details

Details for the file fabric_kv-0.0.1-py3-none-any.whl.

File metadata

  • Download URL: fabric_kv-0.0.1-py3-none-any.whl
  • Upload date:
  • Size: 6.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.0

File hashes

Hashes for fabric_kv-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 62228d30d0d920cf1b96c939f8c744099bf3c435539e7fe4f3d85de9fadbfadd
MD5 4b296f03ad7c36dd253646643da1df99
BLAKE2b-256 b696b640d7ba0b12f963ccafcb7eecd2770e50d85fe2d5aa941d8cb157abd48b

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page