Instructions to use arealicehole/trinity-nano-hermes-qlora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use arealicehole/trinity-nano-hermes-qlora with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="arealicehole/trinity-nano-hermes-qlora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("arealicehole/trinity-nano-hermes-qlora", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use arealicehole/trinity-nano-hermes-qlora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "arealicehole/trinity-nano-hermes-qlora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arealicehole/trinity-nano-hermes-qlora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/arealicehole/trinity-nano-hermes-qlora

SGLang

How to use arealicehole/trinity-nano-hermes-qlora with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "arealicehole/trinity-nano-hermes-qlora" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arealicehole/trinity-nano-hermes-qlora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "arealicehole/trinity-nano-hermes-qlora" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arealicehole/trinity-nano-hermes-qlora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use arealicehole/trinity-nano-hermes-qlora with Docker Model Runner:
```
docker model run hf.co/arealicehole/trinity-nano-hermes-qlora
```

Trinity Nano Hermes QLoRA

Fine-tuned MoE Adapter — Trinity Nano (Trinity-Nano-Preview) fine-tuned on Hermes function-calling data via QLoRA + Axolotl.

Model Details

Base Model: arcee-ai/Trinity-Nano-Preview (AfmoeForCausalLM, MoE)
Format: QLoRA LoRA adapter (safetensors)
Fine-tune Data: NousResearch/hermes-function-calling-v1
Training Stack: Axolotl + Akash Network
Fine-tuned by: arealicehole
License: Apache 2.0

LoRA Configuration

{
  "peft_type": "LORA",
  "r": 64,
  "lora_alpha": 128,
  "use_rslora": true,
  "target_modules": ["q_proj", "k_proj", "v_proj", "o_proj", "gate_up_proj"],
  "lora_dropout": 0.0
}

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

tokenizer = AutoTokenizer.from_pretrained("arealicehole/trinity-nano-hermes-qlora")
model = AutoModelForCausalLM.from_pretrained(
    "arealicehole/trinity-nano-hermes-qlora",
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

messages = [{"role": "user", "content": "What's 84 * 3 / 2?"}]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to("cuda")
outputs = model.generate(inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0]))

Merge with Base (Transformers)

from transformers import AutoModelForCausalLM
from peft import PeftModel

base = AutoModelForCausalLM.from_pretrained(
    "arcee-ai/Trinity-Nano-Preview",
    torch_dtype=torch.bfloat16,
    device_map="auto"
)
model = PeftModel.from_pretrained(base, "arealicehole/trinity-nano-hermes-qlora")
merged = model.merge_and_unload()

Model Sources

Base Model: arcee-ai/Trinity-Nano-Preview
Training Dataset: NousResearch/hermes-function-calling-v1

Cite

@misc{trinity-nano-hermes-qlora,
  author = {arealicehole},
  title = {Trinity Nano Hermes QLoRA},
  year = {2026},
  url = {https://huggingface.co/arealicehole/trinity-nano-hermes-qlora}
}

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for arealicehole/trinity-nano-hermes-qlora

Base model

arcee-ai/Trinity-Nano-Base-Pre-Anneal

Finetuned

arcee-ai/Trinity-Nano-Base

Finetuned

arcee-ai/Trinity-Nano-Preview

Finetuned

(5)

this model