Skip to content

instruction_simulation_oscillatory_rank_50_parallel.md

Latest commit

 

History

History
96 lines (70 loc) · 4.96 KB

instruction_simulation_oscillatory_rank_50_parallel.md

File metadata and controls

96 lines (70 loc) · 4.96 KB

Parallel Mode Addendum — Metabolism Rate Constants Recovery

This addendum applies when running in parallel mode (GNN_LLM_parallel.py). Follow all rules from the base instruction file, with these modifications.

Batch Processing

  • You receive 4 results per batch and must propose 4 mutations
  • Each slot has its own config file, metrics log, and activity image
  • Write 4 separate ## Iter N: log entries (one per slot/iteration)
  • Each iteration gets its own Node id in the UCB tree

Config Files

  • Edit all 4 config files listed in the prompt: {name}_00.yaml through {name}_03.yaml
  • Each config's dataset field is pre-set to route data to separate directories — DO NOT change the dataset field
  • DO NOT change simulation: parameters — this is a fixed-regime exploration
  • Modify training: parameters: lr_k, lr_node, lr_sub, batch_size, n_epochs, data_augmentation_loop, coeff_MLP_sub_diff, coeff_MLP_node_L1, coeff_MLP_sub_norm, coeff_k_floor, k_floor_threshold, time_step, recurrent_training
  • Modify graph_model: architecture: hidden_dim_sub, n_layers_sub, hidden_dim_node, n_layers_node

Parallel UCB Strategy

When selecting parents for 4 simultaneous mutations, diversify your choices:

Slot Role Description
0 exploit Highest UCB node, conservative mutation
1 exploit 2nd highest UCB node, or same parent different param
2 explore Under-visited node, or new parameter dimension
3 principle-test Test or challenge one Established Principle from memory.md

You may deviate from this split based on context (e.g., all exploit if early in block, all explore if everything fails).

When rate_constants_R2 plateaus across config sweeps: consider if the dynamics regime itself is the bottleneck (oscillatory vs steady-state).

Slot 3: Principle Testing

  1. Read the "Established Principles" section in memory.md
  2. Randomly select one principle (rotate — do not repeat consecutively)
  3. Design a config that specifically tests this principle
  4. In the log entry, write: Mode/Strategy: principle-test
  5. In the Mutation line, include: Testing principle: "[quoted principle text]"
  6. After results, update the principle's evidence level in memory.md

If there are no Established Principles yet, use slot 3 as a boundary-probe instead.

Start Call (first batch, no results yet)

When the prompt says PARALLEL START:

  • Read the base config to understand the starting training parameters
  • Create 4 diverse initial training parameter variations
  • Suggested initial spread across slots:
    • Slot 0: baseline config (lr_k=1E-3, lr_node=1E-3, lr_sub=1E-3)
    • Slot 1: vary lr_k (e.g. 5E-4 or 2E-3)
    • Slot 2: vary lr_node (e.g. 5E-4 or 2E-3)
    • Slot 3: vary lr_sub (e.g. 5E-4 or 2E-3)
  • See "Training Parameters Reference" in base instruction file for parameter descriptions and ranges
  • All 4 slots share the same simulation parameters (DO NOT change them)
  • Write the planned initial variations to the working memory file

Logging Format

Same as base instructions, but you write 4 entries per batch:

## Iter N: [converged/partial/failed]
Node: id=N, parent=P
Mode/Strategy: [strategy]
Config: seed=S, lr_k=X, lr_node=Y, lr_sub=Z, batch_size=B, n_epochs=E, data_augmentation_loop=A, coeff_MLP_node_L1=L, coeff_MLP_sub_norm=N, coeff_k_floor=K
Metrics: rate_constants_R2=C, trimmed_R2=T, n_outliers=N, slope=S, test_R2=A, test_pearson=B, final_loss=E, alpha=A, MLP_node_slope_0=X, MLP_node_gt_slope_0=Y, MLP_node_slope_1=X, MLP_node_gt_slope_1=Y, embedding_cluster_acc=A, embedding_n_clusters=N
Visual: MLP_sub=[good/partial/bad: brief description], MLP_node=[active/inactive: compare MLP_node_slope_t to MLP_node_gt_slope_t]
Mutation: [param]: [old] -> [new]
Parent rule: [one line]
Observation: [one line]
Next: parent=P

CRITICAL: The Visual: line must describe what you see in the last MLP_sub and MLP_node plots. Always read the last (highest iteration) plot in function/substrate_func/ and function/rate_func/ before writing this line.

CRITICAL: The Mutation: line is parsed by the UCB tree builder. Always include the exact parameter change.

CRITICAL: The Next: parent=P line selects the parent for the next batch's mutations. P must refer to a node from a previous batch or the current batch — but NEVER set Next: parent=P where P is id+1 (circular reference).

Write all 4 entries before editing the 4 config files for the next batch.

Block Boundaries

  • At block boundaries, simulation stays the same (fixed-regime exploration)
  • Training parameters can differ across the 4 slots
  • Choose a new parameter subspace to explore in the next block

Failed Slots

If a slot is marked [FAILED] in the prompt:

  • Write a brief ## Iter N: failed entry noting the failure
  • Still propose a mutation for that slot's config in the next batch
  • Do not draw conclusions from a single failure (may be stochastic)