Spaces:
Sleeping
Sleeping
Commit History
-
inference file change
23cd9a7 -
python file
403d77b -
fix: move type ignore to the correct position for OpenAI call and suppress requests stub warning
c446324 -
fix: resolve IDE warnings and add safety checks for requests library
9b0fc9d -
fix: enforce strict scoring normalization in inference.py
68223ad -
fix: enforce strict (0,1) on ALL score fields including penalties
- grader.py: triple-safe wrap every value with safe_score before return - models.py: remove ge/le constraints on penalties, add to auto-clamp validator - inference.py: catch-all numeric safe_score loop in _sanitize_task_result - inference.py: pre-wrap avg_reward with safe_score in aggregation1a071c6 -
fix: rewrite grader with reference scoring pattern and clamp all reward fields
- Adopted additive scoring: category(+0.3), empathy(+0.1/0.2), helpfulness(+0.3), resolution(+0.2), efficiency bonus - Added penalties: angry(-0.25), generic(-0.1), repetition(-0.2), escalation(-0.1) - CRITICAL: clamped observation.reward to avg instead of raw cumulative (was exceeding 1.0) - CRITICAL: clamped cumulative_reward in info dict - Pass action_type, step_count, max_steps to grader for resolution/efficiency scoring7acbefe -
fix: resolve Pyrefly linter errors - add search-path config and type guards
562c0a4 -
fix(ide): suppress IDE false positive import warnings in server/app.py
11c71eb -
pending
6e4bdde -
last 0 and 1 error update
6c591d0
-
fix: add pyright extraPaths to resolve IDE import warnings for models, grader, tasks, server.environment
3932d4b -
Harden score outputs to strict open interval
4191feb -
fix: clamp scores in environment and validate to strict open interval
3e4c834 -
fix: clamp all task scores to strict (0,1) range - never 0.0 or 1.0 - grader.py: add _clamp() helper, apply to all sub-scores and total - models.py: update Pydantic fields to gt/lt strict bounds - inference.py: clamp avg_reward and final_score Fixes Phase 2 'task scores out of range' validation error
11e6068 -
fix: clamp task scores strictly between 0 and 1 to pass Phase 2 validation
1a235af -
fix: strictly pin dependencies in pyproject.toml and use python:3.10-slim to bypass HF docker image cache rate limits
cd189b9 -
fix: update openai pinned version to 2.30.0 to resolve conflict with openenv-core 0.2.0
e3ef386 -
fix: pin dependencies and change docker base image to prevent HF cache error
cc0cc48 -
fix: align port to 7860 for HF Spaces and harden inference.py error handling
40b0e9f -
inference change
b1da698 -
Fix inference script validation crashes
0d5ce63
-
dockerize the project to work
Mathiyazhagan_NTL45e91a8 -
final all correct checklist update
9b26357 -
mac update
Mathiyazhagan_NTL6c741e5 -
check list
f5c7702 -
pre submission checklist
bfda67d -
completed the setup
e52fa02 -
completed submission setup
919d11e -
update the validate files
2d15ceb -
Add openenv-core dependency and uv.lock
8b5be14 -
Update URLs and add validate script
b079cfc -
configure for HuggingFace Spaces deployment (port 7860, non-root user, HF frontmatter)
7b09dd1 -
project upload
81aa69d