issues Search Results · language:Dune language:JavaScript language:JavaScript language:JavaScript language:Python linked:pr
Filter by
4.5M results
🔍 설명
YOLO11n 모델 학습 결과(v4, v5)를 정리하여 프로젝트에 추가합니다.
TO-DO
- [x] v4 학습 결과 업로드
- [x] v5 학습 결과 업로드
비고
- Model: YOLO11n
- 결과 저장 위치: results/
- 버전별 폴더(v4, v5)로 관리
- 모델 파일(.pt)은 별도 공유 예정
enuyhs39
Problem
The context token indicator at the bottom of the chat area shows \N / 32,768 tok\ with a thin progress bar. The bar
never appears to move because:
1. Poor token estimator — \Math.ceil(str.length ...
hasitpbhatt
Part of the perf-engineering epic #864. A batch of correctness + ergonomics fixes to the gold-workload runner found
while dogfooding it on large corpora.
1. Buffered logging (don t contaminate the numbers) ...
performance
lathrys-at
Finding
Reported by: code-reviewer + security-reviewer (PR #643 review)
File: bootstrap/get_app_token.sh line 41-44 Severity: LOW — permission gate bypass on stat failure
Description
_env_mode=$(stat ...
bug
needs-ai
ScottThurlow
Finding
Reported by: security-reviewer (PR #643 review)
File: bootstrap/get_app_token.sh line 67 Severity: LOW — CWE-59, fail-open fallback
Description
_resolved_pem=$(python3 -c import os,sys; print(os.path.realpath(sys.argv[1])) ...
bug
needs-ai
Problem
Only http/llm/callable/demo adapters exist; integrating an existing agent/RAG pipeline needs hand-written boilerplate
even when teams already expose a .query/.invoke method.
Proposal
- Add ...
extensibility
integration
sev:minor
v1.1
justinkastil
Problem
Docs are hand-written .md files with no mkdocs/Sphinx config and no docstring-generated API reference; rich config
surfaces (config.yaml, rubric.yaml, personas.yaml, redteam profiles) are hard ...
docs
dx
sev:minor
v1.1
Problem
Custom-code assertion scorers support a safe AST-checked expr mode and a subprocess mode. In service mode subprocess has
no seccomp/AppArmor, no read-only rootfs, no capability drops, so a malicious ...
hardening
security
sev:major
v1.1
Problem
ci.yml and publish.yml emit no SBOM, no pip-audit/grype CVE scan, no license audit, and no provenance attestation
(confirmed: NONE in workflows). Transitive trees (anthropic/openai/jinja2, optional ...
ci
procurement
security
sev:blocker
v1.1
Problem
The breach/rubric judge is an LLM calibrated once; runs do not pin judge model/version/temperature/system-prompt hash,
there is no canary set that must always breach, and cross-run comparisons ...
calibration
measurement
sev:major
trust
v1.1
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.