# EMEP

- [EMEP: Evolutionary Model Engineering Platform](https://doc.blueapi.ir/index.md): EMEP creates new high-performing open-weight LLMs via compatibility analysis, tensor-level model merging, automated evaluation, and evolutionary optimization.
- [Quickstart: Understand EMEP in Five Minutes](https://doc.blueapi.ir/quickstart.md): Trace a single candidate through EMEP end to end: import, compatibility, merge, evaluate, evolve. A five-minute map of the whole platform.
- [EMEP: Business and Technical Context](https://doc.blueapi.ir/overview/context.md): Understand who EMEP serves, what triggered the project, and where it sits among open-weight LLM tooling like MergeKit and LM Evaluation Harness.
- [EMEP Vision and Scope Boundaries](https://doc.blueapi.ir/overview/vision-and-scope.md): What EMEP is, what it is not, and the phase boundaries that govern its development from specification to production deployment.
- [EMEP Project Constitution: Scientific Rules](https://doc.blueapi.ir/overview/project-constitution.md): The 20 non-negotiable scientific rules governing EMEP design, evaluation integrity, and reproducibility. Each rule includes a one-line rationale.
- [EMEP Canonical Glossary of Terms and Components](https://doc.blueapi.ir/overview/glossary.md): Alphabetical definitions of every EMEP component, state, lifecycle stage, and technical term used across the specification.
- [EMEP Architectural Decision Records](https://doc.blueapi.ir/overview/decisions.md): ADR-001 through ADR-012 covering modular architecture, safetensors, offline-first mode, evidence-tagged docs, and reference runtime assumptions.
- [EMEP Research Reference Cards](https://doc.blueapi.ir/research/references.md): Annotated bibliography of 18 papers underpinning EMEP merge strategies, evaluation, evolution, and PEFT methods. Each card links to a component and experiment ID.
- [EMEP Evidence Tag Classification Map](https://doc.blueapi.ir/research/evidence-map.md): Map every EMEP research area to its evidence tag: Established research, Research hypothesis, Engineering assumption, Experimental result, or Future goal.
- [Research to EMEP Component Justification Map](https://doc.blueapi.ir/research/component-mapping.md): Table and graph mapping each foundational paper to the EMEP components it justifies, from MergeEngine to EvaluationEngine.
- [EMEP Executive System Dependency Map](https://doc.blueapi.ir/maps/system-dependency-map.md): High-level dependency chain from base models through merge, evolution, evaluation, and deployment, with optional fine-tuning and RAG side loops.
- [EMEP Internal Component Dependency Graph](https://doc.blueapi.ir/maps/component-dependency-graph.md): Runtime dependency graph showing which EMEP components call which, with a legend explaining edge types and grouping.
- [EMEP Data Flow and Class Flow Map](https://doc.blueapi.ir/maps/data-flow-map.md): Data-flow diagram showing Weights, Tokenizer, Config, Task Vector, Metric, Manifest, and Signature flows, with explicit benchmark split boundaries.
- [EMEP Experiment Lifecycle State Machine](https://doc.blueapi.ir/maps/experiment-lifecycle.md): State diagram for Experiment states, valid and invalid transitions, and operator actions permitted at each stage.
- [EMEP System Architecture: Three-Level Design](https://doc.blueapi.ir/architecture/system-architecture.md): Three-level architecture for EMEP: executive blocks, logical services, and internal component graphs for TensorEngine and EvolutionEngine.
- [EMEP Technical Requirements: Functional and Non-Functional](https://doc.blueapi.ir/architecture/technical-requirements.md): Numbered FR and NFR requirements for EMEP across correctness, performance, reliability, security, reproducibility, observability, offline operation, and portability.
- [EMEP Component Architecture: 16 Internal Abstractions](https://doc.blueapi.ir/architecture/component-architecture.md): Detailed specification of all 16 EMEP internal abstractions: purpose, interface, responsibilities, dependencies, data ownership, persistence, and failure modes.
- [EMEP Runtime Architecture: Processes, Workers, and Queues](https://doc.blueapi.ir/architecture/runtime-architecture.md): Runtime topology for EMEP: API server, scheduler, GPU workers, CPU workers, evaluation workers, and message flow. Includes merge job sequence diagram.
- [EMEP Security Architecture: Trust Boundaries and Zones](https://doc.blueapi.ir/architecture/security-architecture.md): Trust boundaries, principals, secrets management, artifact signing hierarchy, and network zones for EMEP research, enterprise, and air-gap deployments.
- [EMEP Storage Architecture: Tiers, Layout, and Retention](https://doc.blueapi.ir/architecture/storage-architecture.md): Storage tiers for EMEP: hot NVMe, warm object, cold archive. Layout for model artifacts, datasets, experiment DB, metrics, logs, and signing keys with per-tier retention.
- [EMEP GPU Orchestration: Scheduling and Allocation](https://doc.blueapi.ir/architecture/gpu-orchestration.md): GPU scheduling for EMEP: queue design, priority, preemption, fair-share, per-experiment allocation, multi-node support, and topology awareness for NVLink groups.
- [EMEP Observability: Logs, Metrics, Traces, and Alerts](https://doc.blueapi.ir/architecture/observability.md): Observability pipeline for EMEP: logs, metrics, traces, model-level metrics, evaluation metrics, and alerting hooks with producer-collector-storage-dashboard flow.
- [EMEP Master Diagram Index and Quality Gates](https://doc.blueapi.ir/diagrams/diagram-index.md): Master index of all 30+ mandatory diagram types across EMEP specifications, with type, location, phase, and quality gate checklist.
- [EMEP Architecture Diagrams: Levels 1-3 and Runtime Views](https://doc.blueapi.ir/diagrams/architecture-diagrams.md): Mermaid source for Level 1 Executive, Level 2 Logical, Level 3 Internal, Runtime, Storage, GPU, Trust, Supply Chain, and Deployment topology diagrams.
- [EMEP Algorithm Flowchart Index and Standard Template](https://doc.blueapi.ir/diagrams/algorithm-flowchart-index.md): Table of every mandated algorithm with links to specification pages, plus the standard flowchart template with START/INPUT/VALIDATION/PROCESSING/DECISION/BRANCHES/OUTPUT/END.
- [Model Compatibility: Full Decision Tree and States](https://doc.blueapi.ir/compatibility/model-compatibility.md): Complete compatibility decision tree for EMEP model pairs. Covers architecture, tokenizer, tensor shapes, and metadata checks down to COMPATIBLE, CONDITIONALLY_COMPATIBLE, and INCOMPATIBLE states.
- [Model Import Algorithm and Sequence](https://doc.blueapi.ir/compatibility/model-import.md): Specification for importing models into EMEP. Covers the ModelLoader algorithm, hash verification, ArtifactStore staging, and the full interaction sequence with external sources.
- [Tokenizer Compatibility Algorithm and Checks](https://doc.blueapi.ir/compatibility/tokenizer-compatibility.md): Tokenizer compatibility specification for EMEP. Covers vocab overlap, special token matching, BPE and SentencePiece family checks, and the decision flow to COMPATIBLE, CONDITIONALLY_COMPATIBLE, or INCOMPATIBLE.
- [Architecture Compatibility Algorithm and Checks](https://doc.blueapi.ir/compatibility/architecture-compatibility.md): Architecture compatibility specification for EMEP. Checks model family, layer count, hidden size, attention heads, RoPE, and context config. Cross-family merges are INCOMPATIBLE unless franken-merge.
- [Tensor Shape Validation Algorithm and Rules](https://doc.blueapi.ir/compatibility/tensor-shape-validation.md): Tensor shape validation specification for EMEP. Iterates over state_dict, compares shapes and dtypes per parameter, reports first mismatch, and references broadcasting rules from Tensor Math.
- [MergeEngine: Pipeline, Interface, and Failure Modes](https://doc.blueapi.ir/merge/merge-engine.md): MergeEngine specification for EMEP. Defines the full merge pipeline from model sources through strategy selection, tensor operations, numerical validation, and candidate registration. Includes failure modes.
- [Merge Strategies: Selection, Parameters, and Limitations](https://doc.blueapi.ir/merge/merge-strategies.md): Overview of EMEP merge strategies: Linear Merge, SLERP, TIES, DARE, DARE+TIES, Task Arithmetic, Franken-Merge, and Passthrough. Includes decision tree for strategy selection and per-strategy specs.
- [Tensor Operations: Low-Level Merge Primitives](https://doc.blueapi.ir/merge/tensor-operations.md): Specification of every low-level tensor operation used by EMEP merge strategies. Covers input/output shapes, dtype rules, broadcasting, and numerical stability.
- [Merge Validation: Post-Merge Checks and Outcomes](https://doc.blueapi.ir/merge/merge-validation.md): Post-merge validation specification for EMEP. Covers NaN/Inf scanning, dtype invariants, norm envelopes, tokenizer round-trip, and small-batch forward pass with PASS, FAIL, and QUARANTINE outcomes.
- [Merge Math: Notation, Shapes, and Precision](https://doc.blueapi.ir/math/merge-math.md): Foundational merge math notation for EMEP. Defines scalars, vectors, matrices, tensors, LLM weight shapes, parameter deltas, task vectors, norms, and precision effects. Base of all merge math.
- [SLERP Math: Formula, Boundaries, and Fallback](https://doc.blueapi.ir/math/slerp-math.md): SLERP mathematical specification for EMEP. Defines the spherical linear interpolation formula, shape expectations, boundary cases near theta = 0 and theta = pi, and fallback to lerp. Reference: Shoemake 1985.
- [TIES Math: Trim, Elect Sign, and Disjoint Merge](https://doc.blueapi.ir/math/ties-math.md): TIES mathematical specification for EMEP. Defines the three-step trim-by-magnitude, elect-sign, and disjoint-merge formulas with density parameter and rescaling. Reference: Yadav et al. 2023.
- [DARE Math: Drop and Rescale for Task Vectors](https://doc.blueapi.ir/math/dare-math.md): DARE mathematical specification for EMEP. Defines random sparsification with drop probability p, rescaling by 1/(1-p), and application per task vector. Reference: Yu et al. 2023.
- [Task Arithmetic: Task Vectors and Linear Combinations](https://doc.blueapi.ir/math/task-arithmetic.md): Task Arithmetic mathematical specification for EMEP. Defines task vectors as theta_ft minus theta_pre, linear combinations, skill addition, forgetting/negation, and combination rules. Reference: Ilharco et al. 2022.
- [Tensor Math: Broadcasting, Operations, and Precision](https://doc.blueapi.ir/math/tensor-math.md): Tensor math specification for EMEP. Defines broadcasting rules, elementwise operations, reductions, matmul shape rules for transformer weights, and precision ranges for fp32, bf16, fp16, fp8, int8, and int4.
- [Optional Fine-Tuning in EMEP](https://doc.blueapi.ir/finetuning/optional-finetuning.md): Fine-tuning is an optional, secondary loop in EMEP that produces specialist models as inputs to the MergeEngine.
- [PEFT Overview for EMEP Fine-Tuning](https://doc.blueapi.ir/finetuning/peft.md): Parameter-Efficient Fine-Tuning methods in EMEP including LoRA, prefix-tuning, prompt-tuning, and trade-offs versus full fine-tuning.
- [LoRA Specification for EMEP Fine-Tuning](https://doc.blueapi.ir/finetuning/lora.md): LoRA (Hu et al. 2021) math, hyperparameters, merging into base weights, and the EMEP training flowchart.
- [QLoRA Specification for EMEP Fine-Tuning](https://doc.blueapi.ir/finetuning/qlora.md): QLoRA (Dettmers et al. 2023) with NF4 quantization and LoRA adapters for memory-efficient fine-tuning in EMEP.
- [Catastrophic Forgetting Controls in EMEP](https://doc.blueapi.ir/finetuning/catastrophic-forgetting-controls.md): Controls to mitigate catastrophic forgetting during fine-tuning in EMEP, with mandatory regression evaluation before and after any FT operation.
- [EvolutionEngine Specification for EMEP](https://doc.blueapi.ir/evolution/evolution-engine.md): Specification for the EvolutionEngine component in EMEP, covering the full evolutionary optimization loop for model merging and candidate generation.
- [Genome Encoding for Model Merge Candidates](https://doc.blueapi.ir/evolution/genome.md): Specification for the genome encoding used by EMEP to represent merge candidate configurations, including structure, fields, and JSON example.
- [Population Management in EMEP Evolution](https://doc.blueapi.ir/evolution/population.md): Specification for population initialization, diversity preservation, and archival elite pool management in the EMEP EvolutionEngine.
- [Mutation Operators for Merge Genome Evolution](https://doc.blueapi.ir/evolution/mutation.md): Specification for mutation operators in EMEP, including alpha jitter, per-layer perturbation, strategy switch, and density adjustment with rate scheduling.
- [Crossover Operators for Merge Genome Combination](https://doc.blueapi.ir/evolution/crossover.md): Specification for crossover operators in EMEP, including uniform, one-point on layer index, and arithmetic blend with constraint enforcement.
- [Selection Strategies for Evolutionary Search](https://doc.blueapi.ir/evolution/selection.md): Specification for selection operators in EMEP, including tournament, truncation, elitism, and roulette with decision rationale and configuration.
- [Fitness Engine and Candidate Scoring](https://doc.blueapi.ir/evolution/fitness.md): Specification for the FitnessEngine in EMEP, covering scalar single-objective aggregation and multi-objective vector fitness computation.
- [Multi-Objective Optimization with NSGA-II](https://doc.blueapi.ir/evolution/multi-objective-optimization.md): Specification for multi-objective optimization in EMEP using NSGA-II, covering non-dominated sorting, crowding distance, and Pareto front management.
- [EvaluationEngine and BenchmarkEngine Framework](https://doc.blueapi.ir/evaluation/framework.md): Specification for the EMEP evaluation framework, covering the full candidate evaluation pipeline and explicit split isolation between Optimization, Validation, and Hidden Test sets.
- [Benchmark Record Schema and Configuration](https://doc.blueapi.ir/evaluation/benchmark-specification.md): Specification for benchmark records in EMEP, covering schema fields, versioning, seeds, prompt templates, decoding parameters, and metric definitions.
- [Benchmark Catalog by Category and Purpose](https://doc.blueapi.ir/evaluation/benchmark-catalog.md): Catalog of EMEP benchmark categories covering Persian, English, reasoning, code, safety, and enterprise tasks with split policies and primary metrics.
- [Persian Language Evaluation Concerns and Benchmarks](https://doc.blueapi.ir/evaluation/persian-evaluation.md): Specification for Persian-specific evaluation in EMEP, covering script coverage, diacritics, ZWNJ, transliteration, register variation, and cultural knowledge.
- [Safety Evaluation and Harmful Content Detection](https://doc.blueapi.ir/evaluation/safety-evaluation.md): Specification for safety evaluation in EMEP, covering harmful content categories, jailbreak resistance, refusal appropriateness, and PII leakage detection.
- [Regression Detection and Statistical Protocol](https://doc.blueapi.ir/evaluation/regression-evaluation.md): Specification for regression evaluation in EMEP, covering statistical protocols for detecting performance regression against baselines using bootstrap confidence intervals.
- [Benchmark Integrity and Data Contamination Prevention](https://doc.blueapi.ir/evaluation/benchmark-integrity.md): Specification for data contamination detection, leakage classification, and Hidden Test Set access control enforcement in EMEP benchmarks.
- [ExperimentTracker Event Stream and Responsibilities](https://doc.blueapi.ir/tracking/experiment-tracking.md): Specification for the ExperimentTracker in EMEP, covering event stream semantics, immutable append-only logging, and integration with the evolution and evaluation pipeline.
- [Complete Experiment Record Schema](https://doc.blueapi.ir/tracking/experiment-schema.md): Specification for the full experiment record schema in EMEP, including all fields for reproducibility, provenance, and result storage with JSON example.
- [ModelRegistry API and State Management](https://doc.blueapi.ir/tracking/model-registry.md): Specification for the ModelRegistry in EMEP, covering API operations, storage model, and state transitions aligned with the model lifecycle.
- [DatasetRegistry for Versioned Data Management](https://doc.blueapi.ir/tracking/dataset-registry.md): Specification for the DatasetRegistry in EMEP, covering versioned datasets, split manifests, checksums, and per-dataset license tracking.
- [ArtifactStore for Content-Addressed Model Storage](https://doc.blueapi.ir/tracking/artifact-registry.md): Specification for the ArtifactStore in EMEP, covering content-addressed storage, tiering, garbage collection, and digital signatures.
- [Model Lineage Graph and Ancestry Tracking](https://doc.blueapi.ir/tracking/model-lineage.md): Specification for model lineage tracking in EMEP, covering parent-child relationships, merge metadata, and lineage graph visualization.
- [Reproducibility Requirements for EMEP Experiments](https://doc.blueapi.ir/tracking/reproducibility.md): Specification for what is required to reproduce an EMEP experiment, covering code, model, dataset, benchmark, environment, seed, and configuration versions.
- [Step-by-Step Experiment Reproduction Protocol](https://doc.blueapi.ir/tracking/reproducibility-protocol.md): Specification for the EMEP experiment reproduction protocol, including step-by-step flowchart, failure conditions, and verification checks.
- [Model Provenance Record and Verification Chain](https://doc.blueapi.ir/tracking/model-provenance.md): Specification for model provenance in EMEP, covering provenance record schema, signed provenance chains, and the verification algorithm with flowchart.
- [Quantization Engine: Precision Reduction for Inference](https://doc.blueapi.ir/deployment/quantization.md): QuantizationEngine reduces model weights to FP16, BF16, FP8, INT8, or INT4. This page specifies the decision tree, flowchart, and quality gates.
- [Inference Backend: Serving Engine Selection](https://doc.blueapi.ir/deployment/inference.md): InferenceBackend specifies PyTorch, vLLM, TGI, llama.cpp, and ONNX Runtime. Backend selection, batching, KV cache, and speculative decoding are covered.
- [Deployment Manager: Environment and Strategy Specification](https://doc.blueapi.ir/deployment/deployment-specification.md): DeploymentManager orchestrates staging, canary, and production rollouts. This page specifies deployment strategies, approval gates, and state lifecycle.
- [Offline Deployment: Air-Gapped Transfer Protocol](https://doc.blueapi.ir/deployment/offline-deployment.md): Full air-gapped deployment flow from research model to production in an isolated environment. Sequence diagram, integrity checks, and acceptance gates.
- [Performance Validation: Latency and Throughput Gates](https://doc.blueapi.ir/deployment/performance-validation.md): PerformanceValidation measures latency, throughput, memory, tokens per second, and TTFT. Statistical protocol, regression gate, and flowchart specified.
- [Enterprise Model: Versioning and Lineage Specification](https://doc.blueapi.ir/enterprise/enterprise-model.md): Enterprise Model versioning rules, semantic versioning, lineage to research candidates, compatibility guarantees, and deprecation policy.
- [Offline Operations: Air-Gapped Environment Runbook](https://doc.blueapi.ir/enterprise/offline-operations.md): Air-gapped operations with no egress, signed bundles, one-way log transfer, and forbidden operations. Runbook flowchart and security rules.
- [Model Update Protocol: Controlled Upgrade Pipeline](https://doc.blueapi.ir/enterprise/model-update-protocol.md): Controlled upgrade pipeline from current production version to the next. Evaluation, merge, validation, security review, and approval gates.
- [Model Rollback: Regression Recovery Protocol](https://doc.blueapi.ir/enterprise/model-rollback.md): Rollback flowchart for production regression. Detect, freeze, restore, verify, resume, and post-mortem. Automated and manual triggers.
- [Secure Model Packaging: Signed Release Manifest](https://doc.blueapi.ir/enterprise/secure-model-packaging.md): Signed release manifest schema, JSON example, dependency pinning, offline bundle, and signing chain diagram for secure model distribution.
- [Optional RAG Architecture: Downstream Retrieval Pipeline](https://doc.blueapi.ir/rag/optional-rag-architecture.md): Retrieval-Augmented Generation pipeline for EMEP. Ingestion, embeddings, retrieval, reranking, and citation. Downstream and optional.
- [Model Supply Chain Security](https://doc.blueapi.ir/security/model-supply-chain.md): End-to-end model supply chain with checkpoint gates, attack vectors, and mitigations for tampered weights, poisoned models, and malicious tokenizers.
- [License Compliance Matrix and Rules](https://doc.blueapi.ir/security/license-compliance.md): License compatibility matrix, attribution requirements, redistribution rules, and a decision tree for merging models under different open-source licenses.
- [Security Requirements Specification](https://doc.blueapi.ir/security/security-requirements.md): Numbered security requirements for authentication, encryption, secrets, audit logging, malicious model detection, dependency scanning, and offline operation.
- [Artifact Signing and Key Hierarchy](https://doc.blueapi.ir/security/artifact-signing.md): Key hierarchy, signing algorithms, verification flow, rotation schedule, revocation, and air-gapped signing options for model artifacts.
- [Dependency Safety and SBOM Management](https://doc.blueapi.ir/security/dependency-safety.md): Dependency pinning, SBOM generation, CVE monitoring, offline mirror maintenance, and blocked package lists with automated build-time enforcement.
- [Air-Gapped and Offline Security Posture](https://doc.blueapi.ir/security/offline-security.md): Air-gap guarantees, physical media handling, incident containment, one-way log egress, offline HSM custody, and offline signing ceremonies for enterprise deployments.
- [Operations Runbook: Startup, Shutdown, and Tasks](https://doc.blueapi.ir/operations/runbook.md): Cold start, warm start, graceful shutdown, forced shutdown, and common operational tasks for EMEP production deployments.
- [Disaster Recovery Plan and Procedures](https://doc.blueapi.ir/operations/disaster-recovery.md): Failure classification, recovery procedures, RPO/RTO targets, and quarantine protocols for total site loss, storage loss, registry loss, and GPU cluster loss.
- [Backup Strategy and Restore Procedures](https://doc.blueapi.ir/operations/backup-strategy.md): Backup targets, schedules, retention policies, and restore drill procedures for model artifacts, experiment databases, datasets, metrics, and signing keys.
- [Monitoring Pipeline and Metrics Catalog](https://doc.blueapi.ir/operations/monitoring.md): Metrics catalog, dashboard definitions, and the OpenTelemetry pipeline for GPU health, experiment throughput, evaluation queues, and storage capacity.
- [Alert Catalog and Escalation Procedures](https://doc.blueapi.ir/operations/alerting.md): Alert definitions by severity, routing rules, acknowledgment requirements, and the escalation flow for unacknowledged alerts in EMEP operations.
- [GPU Health Monitoring and Remediation](https://doc.blueapi.ir/operations/gpu-health.md): GPU health metrics, automated health checks, quarantine procedures, and auto-remediation for ECC errors, thermal events, XID errors, and memory failures.
- [Storage Operations: Cleanup, Retention, and Tiering](https://doc.blueapi.ir/operations/storage-operations.md): Artifact cleanup, retention enforcement, cold tier migration, capacity thresholds, and the automated storage lifecycle flow for model weights and experiment data.
- [EMEP REST API Specification](https://doc.blueapi.ir/interfaces/api-specification.md): REST API surface for EMEP including endpoints for models, compatibility, merge, experiments, candidates, evolution, benchmarks, and artifacts.
- [EMEP CLI Specification](https://doc.blueapi.ir/interfaces/cli-specification.md): emep CLI commands for model registration, compatibility checks, merge execution, experiments, evolution, evaluation, artifacts, deployment, and offline packaging.
- [EMEP Configuration Specification](https://doc.blueapi.ir/interfaces/configuration-specification.md): YAML configuration schema for EMEP platform, registry, merge, evolution, evaluation, storage, GPU, deployment, security, and observability settings.
- [EMEP Testing Strategy and Pyramid](https://doc.blueapi.ir/testing/testing-strategy.md): Testing pyramid for EMEP covering unit, integration, numerical, merge regression, benchmark regression, performance, reproducibility, failure recovery, and security tests.
- [EMEP Unit Test Plan Per Component](https://doc.blueapi.ir/testing/unit-test-plan.md): Unit test coverage targets and test categories for every EMEP component from ModelRegistry to DeploymentManager.
- [EMEP Integration Test Plan and Scenarios](https://doc.blueapi.ir/testing/integration-test-plan.md): End-to-end integration test scenarios for EMEP covering register, merge, evaluate, and promote workflows.
- [EMEP Model Merge Tests and Golden Outputs](https://doc.blueapi.ir/testing/model-merge-tests.md): Tensor correctness tests per MergeStrategy in EMEP with golden outputs and tolerance policies for SLERP, TIES, DARE, Task Arithmetic, and Franken-Merge.
- [EMEP Numerical Testing and Tolerance Thresholds](https://doc.blueapi.ir/testing/numerical-testing.md): Tolerance thresholds per dtype, NaN/Inf policies, and numerical stability validation for EMEP tensor operations and merge strategies.
- [EMEP Evaluation Testing and Benchmark Harness](https://doc.blueapi.ir/testing/evaluation-testing.md): Benchmark harness tests, deterministic seeds, and cross-backend validation for EMEP EvaluationEngine and BenchmarkEngine.
- [EMEP Performance Testing and Metrics](https://doc.blueapi.ir/testing/performance-testing.md): Latency, throughput, memory, tokens per second, and TTFT performance tests for EMEP with cross-links to deployment validation.
- [EMEP Security Testing and Verification](https://doc.blueapi.ir/testing/security-testing.md): Signature verification, tampered manifest, malicious model detection, and dependency CVE tests for EMEP security validation.
- [EMEP Failure Injection and Recovery Testing](https://doc.blueapi.ir/testing/failure-injection.md): Failure injection cases for EMEP including GPU crash, storage 500, corrupted weights, invalid tokenizer, and network partition with recovery expectations.
- [EMEP Failure Modes Catalog and Response](https://doc.blueapi.ir/risk/failure-modes.md): Failure modes catalog for EMEP with detection, containment, recovery, retry, rollback, logging, alerting, and final status for each mode.
- [EMEP Risk Management Process](https://doc.blueapi.ir/risk/risk-management.md): Risk identification, scoring, ownership, and mitigation cadence for EMEP with a process flowchart.
- [EMEP Risk Register](https://doc.blueapi.ir/risk/risk-register.md): Risk register for EMEP with risk IDs, descriptions, likelihood, impact, owners, mitigations, and status aligned with failure modes.
- [EMEP Roadmap: Phases 0 Through 9](https://doc.blueapi.ir/project/roadmap.md): Executive roadmap for the Evolutionary Model Engineering Platform covering all ten phases from research through enterprise deployment and optional RAG integration.
- [EMEP Project Management: Governance and Cadence](https://doc.blueapi.ir/project/project-management.md): Governance model for EMEP including docs-first development, ADR-driven changes, phase gates, RACI assignments, and team cadence.
- [EMEP Master Task Index: All Epics and Deliverables](https://doc.blueapi.ir/project/tasks.md): Master task index for EMEP spanning all phases. Every task is tagged with an epic, phase, priority, owner, and current status.
- [EMEP KPI Specification: Metrics and Targets](https://doc.blueapi.ir/project/kpi-specification.md): Key performance indicators for EMEP including documentation coverage, traceability, reproducibility, regression rate, throughput, latency, and deployment success metrics.
- [Phase 0: Research, Architecture, and Documentation Foundation](https://doc.blueapi.ir/project/phases/phase-0.md): Establish the foundational research, system architecture, and documentation for EMEP. This phase delivers core conceptual assets, architectural decisions, and a comprehensive glossary before any implementation begins.
- [Phase 1: Model Registry and Compatibility Infrastructure](https://doc.blueapi.ir/project/phases/phase-1.md): Build the foundational model management layer for EMEP. Phase 1 delivers the ModelRegistry, ModelLoader, and ModelCompatibilityAnalyzer, plus API and CLI specifications for model operations.
- [Phase 2: Basic Merge Engine with Linear, SLERP, and Task Arithmetic](https://doc.blueapi.ir/project/phases/phase-2.md): Implement the foundational merge engine for EMEP. Phase 2 delivers TensorEngine, MergeEngine, and three core strategies: Linear Merge, SLERP (Shoemake 1985), and Task Arithmetic (Ilharco et al. 2022).
- [Phase 3: Advanced Merge Strategies with TIES, DARE, and Franken-Merge](https://doc.blueapi.ir/project/phases/phase-3.md): Extend EMEP with advanced model merging strategies. Phase 3 delivers TIES (Yadav et al. 2023), DARE (Yu et al. 2023), DARE+TIES, Franken-Merge, and the merge validation framework.
- [Phase 4: Evaluation Engine, Benchmarking, and Fitness Assessment](https://doc.blueapi.ir/project/phases/phase-4.md): Build the evaluation and benchmarking layer for EMEP. Phase 4 delivers EvaluationEngine, BenchmarkEngine, FitnessEngine, benchmark catalog, Persian evaluation, safety evaluation, and split enforcement.
- [Phase 5: Evolution Engine for Automated Model Optimization](https://doc.blueapi.ir/project/phases/phase-5.md): Phase 5 delivers the EvolutionEngine with genome encoding, population management, mutation, crossover, selection, and NSGA-II multi-objective optimization to discover superior merged models.
- [Phase 6: Quantization and Inference Optimization for EMEP](https://doc.blueapi.ir/project/phases/phase-6.md): Phase 6 delivers the QuantizationEngine and InferenceBackend, enabling FP16, BF16, FP8, INT8, and INT4 quantization with performance validation for efficient model serving.
- [Phase 7: Offline Deployment and Secure Model Packaging](https://doc.blueapi.ir/project/phases/phase-7.md): Phase 7 delivers offline deployment capabilities with the DeploymentManager, artifact signing, secure model packaging, and security validation for air-gapped production environments.
- [Phase 8: Enterprise Pilot with Monitoring and Operations](https://doc.blueapi.ir/project/phases/phase-8.md): Phase 8 delivers the Enterprise Model v1 pilot with update protocols, rollback procedures, runbooks, monitoring, alerting, and disaster recovery drills for production readiness.
- [Phase 9: Optional RAG Integration for Enhanced Retrieval](https://doc.blueapi.ir/project/phases/phase-9.md): Phase 9 is an optional RAG integration phase that delivers retrieval-augmented generation architecture and downstream services without modifying model weights or the core EMEP loop.
- [Consistency Audit Report: EMEP Documentation](https://doc.blueapi.ir/audit/consistency-report.md): Comprehensive audit template for verifying terminology, architecture, API, database, task alignment, dependencies, research claims, math, benchmarks, security, and versioning across EMEP documentation.
- [Traceability Matrix for EMEP Requirements and Implementation](https://doc.blueapi.ir/audit/traceability-matrix.md): Maps EMEP functional and non-functional requirements to architecture components, specifications, implementation tasks, tests, and KPIs in a single traceable table.
- [Open Questions Register for EMEP Development](https://doc.blueapi.ir/audit/open-questions.md): Tracks unresolved decisions across EMEP phases, including ownership, decision deadlines, impact if unresolved, and current status for each open question.