Instructions to use SZLHOLDINGS/KHIPU-R2-estate-sft-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SZLHOLDINGS/KHIPU-R2-estate-sft-v1 with Transformers:

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

pipe = pipeline("text-generation", model="SZLHOLDINGS/KHIPU-R2-estate-sft-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SZLHOLDINGS/KHIPU-R2-estate-sft-v1")
model = AutoModelForCausalLM.from_pretrained("SZLHOLDINGS/KHIPU-R2-estate-sft-v1", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SZLHOLDINGS/KHIPU-R2-estate-sft-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SZLHOLDINGS/KHIPU-R2-estate-sft-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SZLHOLDINGS/KHIPU-R2-estate-sft-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SZLHOLDINGS/KHIPU-R2-estate-sft-v1

SGLang

How to use SZLHOLDINGS/KHIPU-R2-estate-sft-v1 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 "SZLHOLDINGS/KHIPU-R2-estate-sft-v1" \
    --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": "SZLHOLDINGS/KHIPU-R2-estate-sft-v1",
		"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 "SZLHOLDINGS/KHIPU-R2-estate-sft-v1" \
        --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": "SZLHOLDINGS/KHIPU-R2-estate-sft-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use SZLHOLDINGS/KHIPU-R2-estate-sft-v1 with Docker Model Runner:
```
docker model run hf.co/SZLHOLDINGS/KHIPU-R2-estate-sft-v1
```

KHIPU-R2-estate-sft-v1

Estate-corpus supervised fine-tune of KHIPU-R2 with the abstention counterweight.


Base	SZLHOLDINGS/KHIPU-R2
Datasets	Full estate corpus (3,176+ examples) (estate knowledge + doctrine + quant + frontier + abstention counterweight)
Method	QLoRA 4-bit (r=16), 2 epochs, lr 2e-4, seqlen 4096, packing
Artifact	Merged 16-bit full model
Hardware	1x RTX A6000 (Massed Compute), Unsloth backend
Ran	2026-11-05

Evidence. [MEASURED] Heartbeat log retained in the private SZLHOLDINGS/wave-logs dataset. Launcher: szl-holdings/szl-forge · training/train_wave.py.

Honest boundaries. Estate-corpus SFT with the abstention counterweight (szl-quant-abstention-sft-v1, built from the measured confabulation failure in the eval receipts). Not an independent quality evaluation; no accuracy or performance claim. Fleet eval pending.

Part of the SZL Holdings governed estate. Control before action. Evidence after.

Downloads last month: 37

Safetensors

Model size

2B params

Tensor type

BF16

Model tree for SZLHOLDINGS/KHIPU-R2-estate-sft-v1

Base model

Qwen/Qwen2.5-1.5B

Finetuned

Qwen/Qwen2.5-1.5B-Instruct

Adapter

SZLHOLDINGS/KHIPU-R2

Finetuned

(1)

this model