Phase 1 — Boot & Baselines (1–5)
1. autotrain-Llama2chat
Why: gentle on-ramp to chat fine-tunes.
Do: run inference, inspect tokenizer, export to GGUF.
2. NousResearch-Llama2-chat
Why: compare a known chat baseline.
Do: side-by-side eval vs #1 (accuracy, toxicity, latency).
3. NousResearch-Llama2-7bhf
Why: plain 7B base; learn prompting vs. instruction.
Do: simple domain prompts; log failure modes.
4. autotrain-innerillm2
Why: try AutoTrain workflow end-to-end.
Do: tiny domain fine-tune, create a clean Model Card.
5. I-NousResearch-Yarn-Mistral-7b-128k
Why: long-context handling.
Do: load 100k token docs; test chunking vs true long context.
Phase 2 — Merges & Specialty (6–12)
6. InnerILLM-7B-slerp
Why: learn SLERP/merge basics.
Do: compare to base; quantify style/quality shifts.
7. InnerILLM-0x00d0-7B-slerp
Why: variant merge discipline.
Do: ablation: which parents improve which tasks?
8. InnerILLM-0x00d0-Ox0dad0-nous-nous-v2.0-7B-slerp
Why: deeper merge tree.
Do: build a merge manifest; reproducibility script.
9. InnerI-AI-merge-7B-slerp
Why: your house “meta-merge”.
Do: run eval harness; freeze a release tag.
10. InnerI-sn6-merge-7B-slerp
Why: compare another merge lineage.
Do: pick best for RAG generator upstream.
11. A-I-0xtom-7B-slerp
Why: explore alt parent flavors.
Do: controlled tests on coding/doc QA.
12. I-Code-NousLlama7B-slerp
Why: specialize for code-assist.
Do: pass@k smoke tests; tool-calling guardrails.
Phase 3 — OpenPipe / Solar / Chat (13–16)
13. InnerILLM-OpenPipe-Nous-Yarn-Mistral-optimized-1228-7B-slerp
Why: see an OpenPipe-optimized lineage.
Do: latency/cost profiling vs #6–#12.
14. I-OpenPipe-NH2-Solar-7B-slerp
Why: Solar/NH2 flavor for reasoning.
Do: structured output eval; chain-of-thought not exposed.
15. InnerIAI-chat-7b-grok
Why: chat flagship.
Do: production chat demo + safety notes.
16. InnerI-bittensor-7b
Why: dive into decentralized inference.
Do: document the networking model; risks & benefits.
Phase 4 — Synthetic & CAI (17–21)
17. CAI
Why: base asset for our “Inner I Conscious/Creative AI” line.
Do: define the system prompt kernel + values.
18. CAI-synthetic
Why: synthetic data pipeline.
Do: show uplift on weak domains; quality filters.
19. synCAI-144k-gpt2.5 (0.4B)
Why: tiny/fast, long-context tester.
Do: mobile/edge use; cost/perf table vs 7B.
20. synCAI-144k-llama3.1
Why: long-context Llama track.
Do: 120k+ token RAG benchmark; hallucination guard.
21. InnerI/InnerILLM-7B-slerp (or best performer) as Prod Agent
Why: ship the stack.
Do: containerize (FastAPI), add pgvector memory, Langfuse tracing, evals, CI/CD, live demo.
Sources: Agentic Inner I Protocol
Stay in the Now
Within Inner I Network
Inner I Net Company/ - Innovative Solutions to Shape Reality 