Skip to content

Release v0.0.1

Choose a tag to compare

View all tags
@github-actions github-actions released this 17 Feb 01:23
· 31 commits to main since this release
v0.0.1
739a9a2
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

What's Changed in v0.0.1

🚀 Changes

  • bypass git leaks license as gitleaks has token generation issues. (22a2828)
  • Removed hardcoded reviewers. (9b9506d)
  • Update .github/workflows/release.yml (fbd55d6)
  • Update docs/SECURITY.md (c30b81d)
  • Update .github/workflows/release.yml (fbfba30)
  • Update .github/workflows/release.yml (06f9a5f)
  • protected by gitleaks (152cb48)
  • Security: add comprehensive security policies and workflows (07f458d)
  • update comprehensive coverage for rtdlm model tests. (eb1ac72)
  • add more coverage tests for sampling (05cef46)
  • fix skipped unit tests and fix critical bugs. (5cbde3d)
  • unit test fixes. (0a1b3b5)
  • test fixes. (619eaaf)
  • unit test fixes. (7f95a1d)
  • file renamed. (8f8e39f)
  • coverage for unit train and rtdlm. (1065a77)
    • test_compute_controller.py: Remove unused imports (Dict, ComputeAction, create_compute_controller_fn, ModuleAdapter, SymbolicReasoningAdapter, ProbabilisticAdapter, create_module_executors) - test_controller_integration.py: Remove unused imports (hk, Dict, ModuleType, ModuleOutput, ComputeState, ControllerLossComputer) (bb4431a)
    • Fix create_rtdlm_agi() default to use_state=True (model uses hk.get_state/set_state) - Fix is_training parameter default to False for deterministic inference - Fix controller loss computation to ADD penalties on top of total_loss (was overwriting) - Remove unused state variable in _forward_with_controller - Add logging to all empty except clauses (12 module executors + orchestration) - Implement compute_multimodal_alignment_loss() with InfoNCE contrastive loss - Aligns text features with audio/video/image features - Includes fusion alignment via cosine similarity (4208106)
    • Fix beam_search early stopping conditions - Changed jnp.all(beam_finished) to bool(jnp.all(beam_finished).item()) - Prevents JAX tracer errors in Python control flow (3cbfd8f)
    • Fix _should_suggest_halt() returning JAX boolean array - Changed bool(confidence.mean() > 0.9) to (confidence.mean() > 0.9).item() (55dbcfe)

Full Changelog: v0.0.0...v0.0.1

Assets 2
Loading