Spaces:
Sleeping
Sleeping
Commit History
-
Update README and add banner image
Verified signaturefd49a3b -
Update README with demo + training space links
Verified signature30ff7e1 -
Update README with demo + training space links
Verified signature22a47c8 -
Remove hf_space_a100 directory - training dashboard moved to separate space
Verified signature210898f -
Deploy demo app (base vs fine-tuned comparison UI)
Verified signature80fd7a4 -
Rename asset images for clarity and update README links
6cf6872 -
Force add documentation images via git lfs
7f4d462 -
Update README with final slide deck images and fix dashboard link
69946ed -
style: fix alignment and whitespace in README architecture diagrams
18efa9a -
Update references: L40S -> A100 across codebase
aed711f -
Final documentation overhaul: Research-grade README and Blog
b3123f8 -
Fix: Restore mandatory HF Space metadata
012ee91 -
Final submission polish: professional README, architecture docs, and plots organization
7f64c0a -
refactor: remove redundant and unused configuration files
8363344 -
feat: add GRPO training script tailored for NVIDIA L40S hardware and ConflictBench evaluation
0abd9ea
-
feat: add GRPO training script for NVIDIA L40S platform integration
0aab4d4 -
feat: implement Gradio dashboard and automated GRPO training pipeline for L40S hardware
b365dea -
feat: implement GRPO training pipeline and visualization suite for Hugging Face Spaces
705f7d6 -
feat: implement Hugging Face Space for automated GRPO training and refine NUM_EPOCHS in production config
a15c1fd -
chore: keep only production training script, remove redundant train_grpo.py
3fc3896 -
chore: remove redundant or unused codebase files
ca758f7 -
feat: 6 bug fixes + production GRPO training script
- Bug 1: verifier rubric3 now 60% pair-F1 + 40% resolution accuracy (was bonus-only) - Bug 2: rebalanced weights (conflict_identification 0.30, no_contradictions 0.15) - Bug 3: 6-tier authority hierarchy (no priority ties between sources) - Bug 4: explicit numbered tier list in prompt - Bug 5: filler count reduced (2,4) from (4,8) - Bug 6: 400 scenarios, curriculum training, beta=0.02, temp=1.0, LoRA r=32855f781 -
feat: wire LoRA adapter loading + update training results
bd19480 -
feat: OpenEnv multi-mode deployment compliance (pyproject.toml, server/app.py, openenv.yaml)
5cbe052 -
feat: Glassmorphism UI redesign and GRPO pipeline fixes
9d99fca -
Enterprise UI Redesign: Implement sleek monochrome glassmorphism, remove emojis, and refine aesthetic hierarchy
0f8c397 -
Fix Gradio TypeError by removing unsupported show_copy_button parameter and move css config to launch()
f974430 -
Redesign Gradio UI with premium glassmorphism, HTML visualizers, and gr.State session handling
9950bcb -
Uncomment accelerate to fix transformers device_map error
7ea4930 -
Remove gradio version pinning to fix huggingface_hub compatibility
933ee02 -
Quote python_version to fix YAML parsing float issue
1a1369e -
Force Python 3.10 in HF Spaces to resolve audioop issues
aa4feaa -
Add pyaudioop for Python 3.13 compatibility
59d4585 -
Disable training dependencies in requirements.txt to fix HF Spaces build
d668ae2 -
Fix Gradio version conflict in metadata
ad4fa1e -
Add HuggingFace Spaces metadata
dedbb92 -
Update HF Space username to Harsh-9209
79ed027 -
Initial commit for ConflictBench hackathon submission
8cc6f0d