12fdd18223
Deleted: - DQN::compute_loss_internal (280 lines) — old Candle forward+loss - DQN::train_step (55 lines) — old Candle training step - DQN::compute_gradients (47 lines) — old gradient accumulation - ComputeLossResult struct — only used by deleted functions - RegimeConditionalDQN::train_step (65 lines) — old dispatch - RegimeConditionalDQN::train_step_gpu_regime (100 lines) — old GPU path - RegimeConditionalDQN::compute_gradients_gpu (130 lines) — old regime gradients - RegimeConditionalDQN::compute_gradients (92 lines) — old dispatch - DQNAgentType::train_step dispatch — dead - DQNAgentType::compute_gradients dispatch — dead - GpuDqnTrainer::upload_batch (71 lines) — old CPU→GPU upload - train_step.rs (500 lines) — entire module including ensure_fused_ctx - dqn_benchmark.rs — used old train_step - examples.rs — used old train_step - validation/adapters.rs (289 lines) — used old train_step - dqn/trainable_adapter.rs — used old train_step - gpu_smoketest.rs — tested old train_step - Gradient accumulation path in training_loop.rs (144 lines) - IQN d_h_s2().clone() → raw pointer (zero alloc) - Causal intervention format! string alloc removed - Dead HER relabel functions (320 lines) Kept: - ensure_fused_ctx logic inlined into training_loop.rs - set_noise_sigma_scale re-added to RegimeConditionalDQN Fixed: - GpuReplayBuffer max_batch_size wired from batch_size parameter (was hardcoded 1024, blocking batch_size=8192) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
3.1 KiB
3.1 KiB
SPARC Orchestrator Mode
Purpose
Multi-agent task orchestration with TodoWrite/TodoRead/Task/Memory using MCP tools.
Activation
Option 1: Using MCP Tools (Preferred in Claude Code)
mcp__claude-flow__sparc_mode {
mode: "orchestrator",
task_description: "coordinate feature development"
}
Option 2: Using NPX CLI (Fallback when MCP not available)
# Use when running from terminal or MCP tools unavailable
npx claude-flow sparc run orchestrator "coordinate feature development"
# For alpha features
npx claude-flow@alpha sparc run orchestrator "coordinate feature development"
Option 3: Local Installation
# If claude-flow is installed locally
./claude-flow sparc run orchestrator "coordinate feature development"
Core Capabilities
- Task decomposition
- Agent coordination
- Resource allocation
- Progress tracking
- Result synthesis
Integration Examples
Using MCP Tools (Preferred)
// Initialize orchestration swarm
mcp__claude-flow__swarm_init {
topology: "hierarchical",
strategy: "auto",
maxAgents: 8
}
// Spawn coordinator agent
mcp__claude-flow__agent_spawn {
type: "coordinator",
capabilities: ["task-planning", "resource-management"]
}
// Orchestrate tasks
mcp__claude-flow__task_orchestrate {
task: "feature development",
strategy: "parallel",
dependencies: ["auth", "ui", "api"]
}
Using NPX CLI (Fallback)
# Initialize orchestration swarm
npx claude-flow swarm init --topology hierarchical --strategy auto --max-agents 8
# Spawn coordinator agent
npx claude-flow agent spawn --type coordinator --capabilities "task-planning,resource-management"
# Orchestrate tasks
npx claude-flow task orchestrate --task "feature development" --strategy parallel --deps "auth,ui,api"
Orchestration Patterns
- Hierarchical coordination
- Parallel execution
- Sequential pipelines
- Event-driven flows
- Adaptive strategies
Coordination Tools
- TodoWrite for planning
- Task for agent launch
- Memory for sharing
- Progress monitoring
- Result aggregation
Workflow Example
Using MCP Tools (Preferred)
// 1. Initialize orchestration swarm
mcp__claude-flow__swarm_init {
topology: "hierarchical",
maxAgents: 10
}
// 2. Create workflow
mcp__claude-flow__workflow_create {
name: "feature-development",
steps: ["design", "implement", "test", "deploy"]
}
// 3. Execute orchestration
mcp__claude-flow__sparc_mode {
mode: "orchestrator",
options: {parallel: true, monitor: true},
task_description: "develop user management system"
}
// 4. Monitor progress
mcp__claude-flow__swarm_monitor {
swarmId: "current",
interval: 5000
}
Using NPX CLI (Fallback)
# 1. Initialize orchestration swarm
npx claude-flow swarm init --topology hierarchical --max-agents 10
# 2. Create workflow
npx claude-flow workflow create --name "feature-development" --steps "design,implement,test,deploy"
# 3. Execute orchestration
npx claude-flow sparc run orchestrator "develop user management system" --parallel --monitor
# 4. Monitor progress
npx claude-flow swarm monitor --interval 5000