GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
nehuronion-dvt
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
nehuronion-dvt
Model Tree
google/gemma-2-9b
Input Modalities
Output Modalities
Supported Functionality
This repository exclusively contains the lightweight adaptive weights (LoRA/PEFT) resulting from the neurodivergent cognitive alignment training. Weighing only a few megabytes, this file stores the executive function correction matrices and the Psychological Mathematical Resolution (RMP) algorithm.
If you are a developer looking to clone this workflow, perform a manual merge, or continue training with new data using unsloth or peft, you can load these adapters directly on top of the base model in your coding environment:
python
from unsloth import FastLanguageModelimport torch MAX_SEQ_LENGTH = 4096 # 1. Load the optimized base model from Google (Gemma 2 9B)model, tokenizer = FastLanguageModel.from_pretrained( model_name = "unsloth/gemma-2-9b-bnb-4bit", max_seq_length = MAX_SEQ_LENGTH, load_in_4bit = True,) # 2. Inject the cognitive adapters from this repositorymodel = FastLanguageModel.get_peft_model( model, r = 64, target_modules = ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"], lora_alpha = 16, lora_dropout = 0, bias = "none",) # Load the Nehuronion-specific weightsmodel.load_adapter("iangel094/nehuronion-dvt-9b-adapters")print("🧬 Neurodivergent cognitive matrix successfully coupled.")