Instructions to use build-small-hackathon/deku with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use build-small-hackathon/deku with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct")
model = PeftModel.from_pretrained(base_model, "build-small-hackathon/deku")

Transformers

How to use build-small-hackathon/deku with Transformers:

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

pipe = pipeline("text-generation", model="build-small-hackathon/deku")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("build-small-hackathon/deku", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use build-small-hackathon/deku with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/build-small-hackathon/deku

SGLang

How to use build-small-hackathon/deku 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 "build-small-hackathon/deku" \
    --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": "build-small-hackathon/deku",
		"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 "build-small-hackathon/deku" \
        --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": "build-small-hackathon/deku",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use build-small-hackathon/deku with Docker Model Runner:
```
docker model run hf.co/build-small-hackathon/deku
```

deku / README.md

frankyy03

Upload folder using huggingface_hub

7b84259 verified 1 day ago

preview code

raw

history blame contribute delete

2.54 kB

	---
	base_model: Qwen/Qwen2.5-0.5B-Instruct
	library_name: peft
	pipeline_tag: text-generation
	tags:
	- base_model:adapter:Qwen/Qwen2.5-0.5B-Instruct
	- lora
	- transformers
	- knowledge-distillation
	- cka
	license: mit
	---

	# Deku — One for All Student

	Qwen2.5-0.5B-Instruct fine-tuned via gated CKA geometry distillation from 5 heterogeneous teacher LLMs. The student learns to absorb the representation geometry of multiple teachers simultaneously through a learned routing gate.

	## Teachers

	\| Model \| Strength \|
	\|---\|---\|
	\| Qwen2.5-1.5B-Instruct \| code, structured reasoning \|
	\| SmolLM2-1.7B-Instruct \| curated quality \|
	\| Phi-3.5-mini-instruct \| instruction following, CoT \|
	\| gemma-2-2b-it \| long context \|
	\| MiniCPM-2B-sft-bf16 \| multilingual, efficiency \|

	## Method

	Path B — geometry-only, tokenizer-agnostic distillation.

	Each teacher has a different tokenizer and hidden dimension, making token-level KL divergence ill-defined across the ensemble. Instead, the student learns to align its hidden-state geometry with each teacher via CKA (Centered Kernel Alignment), weighted by a learned gating network that routes each input to the most relevant teacher.

	The objective is:

	```
	L = λ1·L_task + λ2·L_KL(Qwen1.5B) + λ3·L_geo(gate)
	```

	- `L_task` — next-token cross-entropy on the training mix
	- `L_KL` — KL divergence from Qwen2.5-1.5B (same tokenizer, zero friction)
	- `L_geo` — gated CKA loss: `1 - mean_i gate_i · CKA(H_student, Pi_i · H_teacher_i)`

	Lambdas follow a three-phase curriculum: task-only warmup → KL ramp-in → geometry ramp-in.

	## Training

	- Base: Qwen/Qwen2.5-0.5B-Instruct
	- Adapter: LoRA r=64, α=128 on all attention + MLP projections
	- Data: OpenHermes-2.5 (70%) + GSM8K (20%) + ARC-Challenge (10%)
	- Steps: 5 000 · batch 8 · seq 512
	- Hardware: A100-80GB via Modal
	- Precision: bfloat16

	## Usage

	```python
	from transformers import AutoTokenizer, AutoModelForCausalLM
	from peft import PeftModel

	base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct")
	model = PeftModel.from_pretrained(base, "build-small-hackathon/deku")
	tok = AutoTokenizer.from_pretrained("build-small-hackathon/deku")

	inputs = tok("Explain what a hash map is.", return_tensors="pt")
	out = model.generate(**inputs, max_new_tokens=200)
	print(tok.decode(out[0], skip_special_tokens=True))
	```

	## Demo

	Live soul space + probe interface: [build-small-hackathon/one-for-all](https://huggingface.co/spaces/build-small-hackathon/one-for-all)

	---
	PEFT 0.19.1