Files
mlx-knife/benchmarks/reports
The BROKE Cluster Team d3f7d091bc Release 2.0.4-beta.3: Dependency compatibility + Documentation
Bugfixes and compatibility improvements. No new features.

Core fixes:
- Framework detection for web API models (Issue #48)
- Video-only model filtering from vision capability
- Page size detection for memory metrics (macOS)
- Model switch log timing (after load completion)

Compatibility:
- hub 1.x + transformers 5.0 support
- Python 3.9-3.14 verified (494 tests passing)

Testing infrastructure:
- Benchmark schema v0.2.0 (hardware profiling, system health)
- Benchmark template v1.0 (automated JSONL→Markdown reports)
- Memory timeline visualization (memplot.py)
- Unified model filter (build_model_object single source)

Documentation:
- Multi-Modal Support section in README (Vision subsection)
- JSON API 0.1.5-0.1.6 marked Stable
- Vision promoted from alpha to beta status
- Removed conceptual drift and outdated references

See CHANGELOG.md for complete details.
2025-12-23 12:19:04 +01:00
..
2025-11-17 22:54:06 +01:00
2025-11-17 22:54:06 +01:00

Test Reports

This directory contains JSONL test reports from mlx-knife's E2E test suite.

File Naming Convention

YYYY-MM-DD-vX.Y.Z.jsonl

Example: 2025-11-16-v2.0.3.jsonl

Format

  • One JSON object per line (JSONL)
  • Schema version: Each report has schema_version field
  • Appending: New releases append new files (never edit old ones)

Historical Data

Reports are git-tracked to preserve historical trends:

  • Performance changes over releases
  • Model compatibility evolution
  • Stop token workaround stability

Analysis Examples

# Count tests by outcome
jq -r '.outcome' benchmarks/reports/*.jsonl | sort | uniq -c

# Average tokens/sec by model family
jq -r 'select(.performance) | "\(.model.family) \(.performance.tokens_per_sec)"' \
  benchmarks/reports/*.jsonl | \
  awk '{sum[$1]+=$2; count[$1]++} END {for (f in sum) print f, sum[f]/count[f]}'

# List models with workarounds
jq -r 'select(.stop_tokens.workaround != "none") | "\(.model.id): \(.stop_tokens.workaround)"' \
  benchmarks/reports/*.jsonl | sort -u

# Performance regression detection
jq -r 'select(.performance) | "\(.timestamp) \(.model.id) \(.performance.tokens_per_sec)"' \
  benchmarks/reports/*.jsonl | sort

Schema Version

Current: 0.1.0 (Phase 0 - Experimental)

See ../schemas/report-v0.1.schema.json for details.