README.md

---
license: apache-2.0
tags: [mergekit, ties, slerp, minicpm, zoder, gguf]
base_model:
  - openbmb/MiniCPM5-1B
  - GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-V2-Thinking
  - GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-Thinking
  - Nasaawakening/Zoder1.0-1B
---

# 🧬 Zoder 2.0-1B (REFINED)

Two-phase refinement: TIES ensemble + SLERP pullback toward the proven Zoder 1.0-1B lineage, with auto-tuned interpolation.

## Merge Architecture

```text
openbmb/MiniCPM5-1B
   ├── V2 (w0.8, d0.9) ─┐
   ── V1 (w0.2, d0.6) ─┤  TIES  →  Phase-1
                        │
Zoder1.0-1B ─── SLERP(t=0.2) ─── Phase-1  →  ZODER 2.0-1B REFINED
```

## Auto-Tune Sweep Results

| t | DNA 1.0 | Quick Score |
|---|---------|-------------|
| 0.2 | 80% | 4/4 |
| 0.3 | 70% | 4/4 |
| 0.4 | 60% | 4/4 |

**Selected t: 0.2** (DNA Zoder 1.0: 80%)

## Full Benchmark

| Test | Status |
|------|--------|
| Basic Conversation | ✅ PASS |
| Logical Reasoning | ✅ PASS |
| Code Generation | ✅ PASS |
| Creative Writing | ✅ PASS |
| Instruction Following | ❌ WEAK |
| Multi-turn Memory | ❌ WEAK |
| Math Problem | ✅ PASS |
| Termux Knowledge | ✅ PASS |

**Score: 6/8 (75%)** | Baseline Zoder 1.0-1B: 8/8 (100%)

## GGUF Quants

| Quant | Size | Recommended For |
|---|---|---|
| Q2_K | 463.3 MB | 1-2GB RAM (fastest) |
| Q3_K_M | 555.9 MB | 2-3GB RAM |
| Q4_K_M | 656.2 MB | Mobile/Termux 4GB+ (sweet spot) |
| Q5_K_M | 664.1 MB | 8GB+ RAM |
| Q6_K | 671.5 MB | 12GB+ RAM |
| Q8_0 | 664.6 MB | 16GB+ RAM (max quality) |


## Identity
- Name: Zoder
- Model: Zoder2.0-1B