Solana NVIDIA Trading Factory LoRA

LoRA adapter trained from NousResearch/Hermes-3-Llama-3.1-8B on solanaclawd/solana-clawd-nvidia-trading-factory-instruct.

Hub model ID: solanaclawd/solana-nvidia-trading-factory-8b-lora

Training

• Train rows: 127
• Eval rows: 7
• Max sequence length: 4096
• LoRA rank/alpha: r=32, alpha=64
• LoRA target modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
• Output directory: /data/outputs/solana-nvidia-trading-factory-8b-lora

Metrics

Intended Use

This adapter is intended for Solana-native research and execution agents that need paper-first strategy planning over Phoenix/Vulcan perps, Rise read plans, cuFOLIO/cuOpt Mean-CVaR portfolio handoffs, and risk-gated spot/perps trading-factory workflows.

Loading

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model = "NousResearch/Hermes-3-Llama-3.1-8B"
adapter_id = "solanaclawd/solana-nvidia-trading-factory-8b-lora"

tokenizer = AutoTokenizer.from_pretrained(base_model, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(base_model, device_map="auto", trust_remote_code=True)
model = PeftModel.from_pretrained(model, adapter_id)

Safety

The dataset builder runs in public-safe mode by default and excludes common secret filenames, private key/token patterns, binary artifacts, dependency folders, lockfiles, and high-risk security records that are not suitable for public dataset release.

This adapter is a research/developer artifact. Live trading or wallet actions must remain behind separate execution clients, simulation, explicit operator approval, and pre-trade risk gates.