Pinned Loading

1. ai-agent-scenario-qc ai-agent-scenario-qc Public

   Python toolkit for quality-controlling AI agent training scenarios — JSON schema validation, rubric-based scoring, defect detection, structured feedback reports

   Jupyter Notebook

2. content-policy-rater content-policy-rater Public

   Rubric-based content moderation reviewer — per-criterion PASS/FLAG/BLOCK decisions with reasoning, configurable aggregation (worst_wins / majority / weighted), per-criterion inter-rater agreement.

   Python

3. mlops-training-pipeline mlops-training-pipeline Public

   Composable MLOps toolkit: training loops, experiment tracking, HuggingFace model evaluation, JAX trainer, distributed training config (DDP/FSDP)

   Python

4. python-perf-optimizer python-perf-optimizer Public

   Python performance profiling, latency/throughput benchmarking, memory optimization, and optimization advisor with severity-tagged recommendations

   Python

5. rlhf-pairwise-rater rlhf-pairwise-rater Public

   Pairwise rating CLI for AI responses — per-axis scoring (helpfulness/harmlessness/accuracy/instruction-following), JSONL in/out, inter-rater Cohen's kappa

   Python

6. triton-kernel-lab triton-kernel-lab Public

   Custom GPU kernels in OpenAI Triton — fused softmax, GELU, tiled matmul — with benchmarks vs PyTorch baselines

   Python