model-merging-nav

name	model-merging-nav
description	Navigate model merging strategies and theory. Quick reference for Task Arithmetic, DARE, TIES merging methods, alpha selection, and RTX 4090 validation results.
triggers	["model merging","task arithmetic","DARE","TIES","merge models","multi-task merge","alpha parameter","model composition","fine-tune merge"]

Model Merging Navigator

3 Core Strategies

Strategy	Method	Key Parameter	Success Condition	RTX 4090 Result
Task Arithmetic	base + α*Σ(task_vectors)	α=0.75 optimal	α<1.0 safe, α≥1.0 risky	✓ 100% bench-tested
DARE	Random drop p% + rescale 1/(1-p)	drop_rate=0.9	Works on large models	✓ 90% drop OK on large
TIES	Trim top-k% + sign majority + merge	top_k=20%	Minimal interference	✓ best interference score

Task Arithmetic Alpha Selection

α=0.1  → cos_sim=0.90 ✓ GOOD (low signal preservation)
α=0.25 → cos_sim=0.92 ✓ GOOD
α=0.5  → cos_sim=0.91 ✓ GOOD
α=0.75 → cos_sim=0.88 ✓ GOOD ← OPTIMAL (best preservation-safety balance)
α=1.0  → cos_sim=0.83 ✗ RISKY (norm explosion starts)
α=1.5  → cos_sim=0.73 ✗ RISKY
α=2.0  → cos_sim=0.66 ✗ RISKY (severe norm explosion)

Rule: α<1.0 is safe. α≥1.0 causes weight norm ratio >1.5 → unstable.

DARE Drop Rate Selection

drop=0.5  → cos_sim=0.69 ✗ RISKY (too many active → noise)
drop=0.7  → cos_sim=0.57 ✗ RISKY
drop=0.9  → ✓ GOOD on real large models (synthetic small models show different behavior)
drop=0.95 → ✓ OK but sparse → needs more real data
drop=0.99 → ✗ RISKY (only 1% kept → unstable)

Key insight: DARE works well on real large models (many redundant parameters) but poorly on small synthetic matrices. Our RTX 4090 bench confirms 90% drop works in practice.

TIES Algorithm Steps

1. Trim: Keep only top-k% magnitude per task vector (remove noise)
2. Sign: Resolve sign conflicts by majority vote across models
3. Merge: Average magnitudes where signs agree with majority

Result: Lowest interference (-0.04 vs +0.42 for arithmetic) → best for conflicting tasks.

Simulator

• tools/model_merging_simulator.py — NumPy-based validation (no GPU)
• Alpha sweep + DARE drop rate sweep + all 3 strategies comparison
• Results: results/model_merging_simulator_results.json

Key Notes

• notebook/fundamentals/model-merging-theory.md — theoretical foundations
• memory/model-merging-experiment-rtx4090.md — RTX 4090 bench results

Decision Tree

Need to merge N fine-tuned models into one?
  → Tasks similar/compatible? → Task Arithmetic α=0.75 (simplest, proven)
  → Tasks conflicting/different directions? → TIES (best interference control)
  → Very large models (7B+)? → DARE 90% drop (reduces compute, proven on large)
  → All methods? → Try all three, compare cos_sim + preservation + interference