Spaces:
Running on Zero
Running on Zero
Commit History
-
Chat Mode: multi-turn memory Running
Verified signature7c6f55f
-
Upload app.py with huggingface_hub
Verified signature3352438 -
Upload README.md with huggingface_hub
Verified signature6c547fb -
Upload requirements.txt with huggingface_hub
Verified signature7576fbc -
Upload app.py with huggingface_hub
Verified signature116db3c -
Upload requirements.txt with huggingface_hub
Verified signatured442468 -
Upload README.md with huggingface_hub
Verified signature54d1304 -
Upload app.py with huggingface_hub
Verified signaturecef1fd7
-
Add compressibility analysis + experiment API endpoint
dixiebone13-a11y +1: Claude Opus 4.5- Embed CompressibilityPlugin (Weaver et al. PNAS 2026) for server-side analysis - Add experiment_measure() API endpoint returning consciousness + compressibility metrics - Hidden Gradio API components for programmatic access via gradio_client Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>02b16db
-
v3.4.1: hybrid tanh normalization - preserve dimension signals
dixiebone13-a11y +1: Claude Opus 4.5Replace v3.4 per-dimension re-centering with hybrid approach: tanh(z_score * 0.15) on raw z-scores. Preserves absolute z-score levels (strong Computation signal, AUC=0.768) while eliminating hard ceiling effects from v3.3 clamp. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>e4391d7 -
v3.4: Per-dimension adaptive normalization
dixiebone13-a11y +1: Claude Opus 4.5Replace v3.3 hard clamp [-3,+3] with two-pass per-dimension normalization + tanh bounding. Eliminates ceiling effects where 4/7 dimensions (Logic, Uncertainty, Sequential, Abstraction) were pegged at +3.0 for every math problem. Now all dimensions produce meaningful variance, improving dimension-level discrimination between correct/incorrect answers. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>6d58930 -
v3.3: Add z-score clamp normalization for model-size independence
dixiebone13-a11yClamp z-scored activations to [-3, +3] before computing consciousness scores. Prevents sigmoid saturation on 32B+ models where dimension activations can be extreme outliers (z-scores of +13-19).c4bd249
-
Fix: Remove deprecated Gradio params (show_copy_button, theme, css)
dixiebone13-a11y61a4deb -
Update README.md
Verified signatureb341eb6 -
Fix: Remove dataclass with Dict type to avoid Gradio schema bug
dixiebone13-a11y1d10c02 -
Fix: Pin gradio to 4.31.0 to avoid API schema bug
dixiebone13-a11y70e9094 -
Fix: Upgrade gradio to v5 to fix JSON schema bug
dixiebone13-a11y8afda62 -
Fix: Use default Gradio launch settings for HF Spaces
dixiebone13-a11y8ff1c82 -
Debug: Remove login() call, add env var debugging
dixiebone13-a11y00a103e -
Fix: Pass HF_TOKEN to PeftModel for gated model access
dixiebone13-a11ye4c1061 -
fix: Upgrade peft to >=0.18.0 for LoRA compatibility
dixiebone13-a11y +1: Claude Opus 4.5The LoRA adapter was saved with peft 0.18.0 which uses alora_invocation_tokens. peft 0.10.0 doesn't support this parameter, causing: TypeError: LoraConfig.__init__() got an unexpected keyword argument 'alora_invocation_tokens' Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>3500155
-
feat: Add v3.0 features section with 32B optimizations
Verified signature6c19433 -
fix: Self-Expression dimension 5065 -> 5064 (out of bounds fix)
Verified signature9d141ab -
Upload app.py with huggingface_hub
Verified signature1ec4ea4 -
Upload app.py with huggingface_hub
Verified signature89e292a -
Upload requirements.txt with huggingface_hub
Verified signature8381613 -
Upload requirements.txt with huggingface_hub
Verified signature089843e -
Upload requirements.txt with huggingface_hub
Verified signatureb1b7e6b -
Upload app.py with huggingface_hub
Verified signatured0865e8 -
Upload requirements.txt with huggingface_hub
Verified signature289d84e
-
Upload requirements.txt with huggingface_hub
Verified signature09d641d -
Upload app.py with huggingface_hub
Verified signature55651c7 -
Upload README.md with huggingface_hub
Verified signature9f6e71b -
Upload README.md with huggingface_hub
Verified signature647b097 -
Upload app.py with huggingface_hub
Verified signature828419f -
Upload README.md with huggingface_hub
Verified signatureda81b2a -
Upload requirements.txt with huggingface_hub
Verified signature6736afc -
Upload app.py with huggingface_hub
Verified signaturef055c72 -
initial commit
Verified signaturee5d4ae4