Child_TextOnly_Fused (524.9M Parameters)

Model Summary

Child_TextOnly_Fused is a high-efficiency causal language model extracted and reconstructed from an enterprise AI-DNA (.aidna) genetic container. The model weights are serialized in native SafeTensors (model.safetensors) format with 100% loss-free weight preservation (exact 1:1 parameter fidelity).

This repository conforms strictly to standard Hugging Face directory specifications for auto-classes:

• AutoModelForCausalLM.from_pretrained(...)
• AutoTokenizer.from_pretrained(...)
• Hugging Face pipeline("text-generation", ...)
• Open LLM Leaderboard automated evaluation pipelines.

Architectural Specifications

AI-DNA Lineage & Origin Container

• Container Format: AI-DNA Enterprise Binary v2 Specification
• Lineage Metadata: Fused from 5 parents: Parent_Text_SmolLM2, Parent_Text_Qwen2_5_0_5B, Parent_Text_SmolLM2_360M, Parent_Text_TinyLlama_1_1B, Parent_Text_OPT_125M
• Tensor Integrity Check: Exact lossless reconstruction verified (SHA-256 payload verified).

Quickstart & Seamless Usage

Loading with Hugging Face Transformers

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_path = "test_fused_folder"

# 1. Load Tokenizer seamlessly
tokenizer = AutoTokenizer.from_pretrained(model_path)

# 2. Load Model via AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    torch_dtype=torch.bfloat16 if hasattr(torch, "bfloat16") else torch.float32,
    device_map="auto",
)

# 3. Generate with Chat Template
messages = [
    {"role": "user", "content": "Explain the concept of quantum computing in simple terms."}
]
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=128,
    temperature=0.7,
    top_p=0.95,
    do_sample=True,
)

response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
print(response)

Direct Inference Pipeline

from transformers import pipeline

generator = pipeline("text-generation", model="test_fused_folder", device_map="auto")
results = generator("What are the key pillars of evolutionary computation?", max_new_tokens=64)
print(results[0]["generated_text"])

Open LLM Leaderboard Evaluation

This repository is formatted for submission to the Hugging Face Open LLM Leaderboard. All mandatory structural files (config.json, generation_config.json, model.safetensors, tokenizer.json, tokenizer_config.json, README.md) are present and verified.

Verification & Integrity Checksums

• Weights File: model.safetensors
• Total Tensors: 559 tensors
• Tensors by Dtype: {"bfloat16": 559}
• Binary Size: 1001.18 MB

Converted using convert_aidna_to_safetensors.py — AI-DNA to SafeTensors Bridge.