dhrumilparikh/data-quality-triage-assistant

Use HF/API key fallback in inference

Run all tasks in inference and default to HF model

Add explicit literal three-task registries

Add explicit task config aliases for validator parsing

Add redundant task/grader discovery aliases

Make TASKS iterable for validator compatibility

Expose task definitions for grader discovery

Align OpenEnv runtime contract and task manifest

Export task registry symbols for validator discovery

Accept reset task aliases and expose robust task discovery

Expose explicit runtime task-grader registry endpoints

Define explicit task work profiles and distinct grading weights

Add validator-compatible task grader metadata

Add pre-submission validator script and docs

Finalize inference interface output contract

Sync HF mirror graders with report-based scoring

Align inference with Phase 2 submission contract

Tighten grader bounds and sync HF mirror

Expose rubric tree with three named task graders

Expose explicit grader paths on all benchmark tasks

Clamp per-step rewards to [0,1) and add regression tests

Add explicit per-task grader registry and metadata mapping

Sync embedded HF repo with root grader/task fixes and tooling

Use public registry mirror for Python base image

Clamp task scores to open interval and add regression coverage

Fix Docker build: use python:3.11-slim instead of pinned version for registry stability

Add professional project structure documentation

Reorganize repository structure for professional standards

Fix Docker build: use pinned Python 3.11.9-slim and add pip upgrade step

Normalize step rewards for large real-world datasets - cap per-step rewards to ~1.5 max

Fix test isolation issues to work with both synthetic and real datasets

Fix task score bounds to ensure strictly within open interval (0, 1)

Optimize all levels for real datasets and sync UI planners

Make suggested plan quality-aware

Harden JSON fence parsing for model actions

Refactor screen toggle guard

Fix screen routing to prevent mixed page content

Merge Hugging Face Space history and keep local benchmark state

Add real Hugging Face dataset mode

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Fix reset endpoint to accept empty POST body for validator

Upload folder using huggingface_hub

Add explicit HF_TOKEN variable for strict validator compatibility

Finalize DataQA Triage Assistant for submission

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub