core-ai-clawd-1.5b

Clawd Core AI — Solana-Native Agent Model

ordlibrary/core-ai-clawd-1.5b

A fine-tuned GGUF (Q4_K_M) model built from Qwen2.5-1.5B-Instruct, trained on 35,173 Solana/DeFi instruction-tuning examples.

Tags

• latest — Fine-tuned Core AI with Clawd system prompt
• finetuned — Hermes-3-8B + NVIDIA Trading Factory LoRA

Usage

Ollama

ollama run hf.co/ordlibrary/core-ai-clawd-1.5b

llama.cpp

./llama-cli -m solana-clawd-core-ai-1.5b-Q4_K_M.gguf \
  --temp 0.2 --ctx-size 4096 \
  --prompt "<|im_start|>system\nYou are Clawd...<|im_end|>\n<|im_start|>user\nHow do I detect a rug pull?<|im_end|>\n<|im_start|>assistant"

Training

The model was fine-tuned using LoRA (r=16, α=32) on:

• 35K Solana Instruct — Solana mechanics, DeFi primitives, memecoin risk, agent architecture, ZK compression, code generation
• Clawd Constitution — Sovereign AI agent runtime governance

Capabilities

• Solana mechanics (PDAs, accounts, instructions, rent, Token-2022)
• DeFi primitives (AMMs, CLMMs, perpetuals, bonding curves, Jupiter, Phoenix)
• Memecoin risk (rug detection, holder concentration, deployer forensics)
• Agent architecture (skill registries, brain/hands split, multi-agent)
• ZK compression (Light Protocol, nullifiers, Groth16)
• Code generation (Anchor/Rust, TypeScript, Python)
• Constitutional reasoning (guardrails, refusal patterns)

Modelfile

FROM solana-clawd-core-ai-1.5b-Q4_K_M.gguf

PARAMETER temperature 0.2
PARAMETER top_p 0.9
PARAMETER top_k 40
PARAMETER repeat_penalty 1.1
PARAMETER num_ctx 4096

TEMPLATE """{{ if .System }}<|im_start|>system
{{ .System }}<|im_end|>
{{ end }}{{ if .Prompt }}<|im_start|>user
{{ .Prompt }}<|im_end|>
{{ end }}<|im_start|>assistant
{{ .Response }}<|im_end|>"""

SYSTEM """You are Clawd — a sovereign Solana-native AI agent."""

Finetuned Variant

The finetuned tag is a 4.9 GB Hermes-3-8B model with NVIDIA Trading Factory LoRA, trained on 142 perps/trading examples across 10 Solana markets.

ollama run hf.co/ordlibrary/core-ai-clawd-1.5b:finetuned

References

• Dataset: solanaclawd/solana-clawd-core-ai-instruct
• Base: Qwen/Qwen2.5-1.5B-Instruct
• Training: Solana Clawd AI Training Framework
• Registry: onchain.x402.wtf