Skip to main content

hypertensor-core — HyperTensor Geometric Core

Riemannian geometry primitives for transformer analysis, compression, hallucination detection, and geodesic trajectory computation.

From the HyperTensor project (Papers I–XVIII).

Install

pip install hypertensor-core

Modules

Module Description
GeodesicMetric Riemannian metric tensor, Christoffel symbols, geodesic integration
HallucinationGuard Four-condition hallucination boundary detection
GenerationMetrics Token-collapse, geodesic half-life, topological compression

Quick Start

from hypercore import GeodesicMetric, HallucinationGuard

# Build a Riemannian metric from hidden states
metric = GeodesicMetric(k_manifold=32)
metric.fit(hidden_states)

# Compute geodesic distance between two points
d = metric.geodesic_distance(h_a, h_b)

# Check if a generation is likely hallucinated
guard = HallucinationGuard(coverage_radius=0.15)
is_hallucination, reason = guard.check(
    query_projection=query_k,
    nearest_trajectories=trajectories,
    jury_confidence=jury_J,
)

Advanced

For the full HyperTensor stack including AxiomGauge, ThermalRank, OnlineOja, TreeDrafter, and safety red-team modules, install the main package:

pip install hypertensor

License

MIT — see LICENSE

Download files

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

Source Distribution

hypertensor_core-1.0.0.tar.gz (16.0 kB view details)

Uploaded Source

Built Distribution

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

hypertensor_core-1.0.0-py3-none-any.whl (16.8 kB view details)

Uploaded Python 3

File details

Details for the file hypertensor_core-1.0.0.tar.gz.

File metadata

  • Download URL: hypertensor_core-1.0.0.tar.gz
  • Upload date:
  • Size: 16.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.15

File hashes

Hashes for hypertensor_core-1.0.0.tar.gz
Algorithm Hash digest
SHA256 80eaf82d9d106c82e5ebac583f8a184bd769298ab44f45313441722115685110
MD5 1030eb0e1329be63ffb4132635da3367
BLAKE2b-256 bf1c2d61a3563e89a4eb8d08a48bb273251f49e086361b93d3a686a6bce2b2f0

See more details on using hashes here.

File details

Details for the file hypertensor_core-1.0.0-py3-none-any.whl.

File metadata

File hashes

Hashes for hypertensor_core-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 90aa850b168a9f967ccd32dc46468b5803701566f5332d7d6adbcdd26f13b102
MD5 94e63fe6144d771672d74617bd1b32b4
BLAKE2b-256 bf9a0b4b9db2ef77325e6d7926b290e8e4c6a77533235d8be29a87663c560d2c

See more details on using hashes here.

Release history Release notifications | RSS feed

This release

1.0.0 This release

2 files

Supported by

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