Text Generation
PyTorch
English
ScalarPotentialLMSARFMassLNMultiXi
research
semsimula
conservative-language-model
scalar-potential
lagrangian-mechanics
energy-based-model
physics-informed
splm
multi-channel-xi
tinystories
non-transformer
attention-free
constant-memory-inference
riemannian-geometry
riemannian-geodesics
damped-riemannian-geometry
Eval Results (legacy)
Commit History
-
Rename Fock-Attention to 'Fock Attention (MLP V_theta)', fix stale pre-causal-leak-fix 9.30 PPL comparisons, add V_theta-vs-exchange-mechanism insight
Verified signature774c51f -
Remove deprecated depth-cond OpenWebText row (retired due to causal leak) from SPLM Family Overview table
Verified signature3a20ea9 -
Move Verlet-instability row to the bottom of the SPLM Family Overview table
Verified signature8f8d020 -
Reorder SPLM Family Overview: all TinyStories rows before all OpenWebText rows
Verified signature0b50f44 -
Sync SPLM Family Overview table across the collection (order, completeness, corpus column)
Verified signaturec67304c
-
add effective damping value to parameter table row
Verified signature932384d -
fix: use Unicode γ + HTML <sub> for gamma_eff (HF inline-math brace rendering issue)
Verified signature04e1d2a -
clarify effective damping: LayerNorm makes gamma_eff << gamma_nominal (heavily underdamped)
Verified signatureb39259c -
Add riemannian-geodesics tag
Verified signature8ae8466
-
fix: remove outer paren before \( math delimiter in CoT bullet
Verified signaturee4ea6fd -
fix: remove outer parens before math delimiters; revert to bare underscore
Verified signature9b758a6 -
fix: escape _ as \_ in prose inline math to prevent italic parsing
Verified signature98ecb0a -
fix: use $...$ delimiters for inline math in bullet points
Verified signature91a99c6
-
docs: add damped Riemannian geometry framework (June 2026 diagnostic battery results)
Verified signature7c185a2
-
Upload README.md with huggingface_hub
Verified signatureda98831 -
Add links to Colab rerun notebook and training results in companion repo
Verified signatured66d26b -
Update Multi-Xi SPLM PPL to 11.51 (16k rerun) and add checkpoint docs
Verified signaturef5a827a -
Update Multi-Xi SPLM PPL to 11.51 (16k rerun) and add checkpoint docs
Verified signatureec659d7 -
Upload experiment_report.json
Verified signature3b4ca0e -
Upload alpha_evolution.png
Verified signature07b6944 -
Upload convergence_curves.png
Verified signature5801334 -
Upload comparison_chart.png
Verified signature666ffdb -
Upload training_summary_8k.md
Verified signaturec3a2229 -
Upload loss_curve_8k.png
Verified signature4fff77a -
Upload training_log_8k.jsonl
Verified signature3a273e8 -
Upload checkpoint/model_8k.pt
Verified signatured2e9b88 -
Upload training_summary_16k.md
Verified signature0d18565 -
Upload loss_curve_16k.png
Verified signature8632292 -
Upload training_log_16k.jsonl
Verified signature728306c -
Upload checkpoint/model_16k.pt
Verified signature283ce39
-
Add 'riemannian-geometry' tag
Verified signature5a78d08 -
Add 'constant-memory-inference' tag
Verified signature74dd2c3 -
Add 'non-transformer' tag
Verified signaturee873fb6 -
Add 'attention-free' tag
Verified signature249a484 -
Add Geometric Capabilities section: Riemannian metric, geodesic reasoning, hallucination detection, native CoT
Verified signature4ec0df3 -
Fix KaTeX: restructure prose to avoid ( before \\( delimiter
Verified signature93c72c6 -
Fix KaTeX rendering: remove bold wrapping and adjacent parens around math in table cells
Verified signaturead1fc7a -
Add 'Why Not a Transformer?' section with O(1) complexity, parameter comparison, and runtime capacity figure; fix KaTeX inline math; add Inference column to family table
Verified signaturea5d3e76 -
Add runtime information capacity figure
Verified signature31b453f -
Fix inline math: use \\(...\\) for KaTeX on HuggingFace
Verified signature37429a2 -
Fix math notation: convert plain text to LaTeX
Verified signaturea45d844 -
Initial model card and config for semsimula-splm-multixi
Verified signature7d59234 -
initial commit
Verified signaturede9a6b9