Instructions to use oswaldoludwig/kappatune-lora-tinyllama-agnews with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use oswaldoludwig/kappatune-lora-tinyllama-agnews with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("TinyLlama/TinyLlama-1.1B-Chat-v1.0")
model = PeftModel.from_pretrained(base_model, "oswaldoludwig/kappatune-lora-tinyllama-agnews")

Transformers

How to use oswaldoludwig/kappatune-lora-tinyllama-agnews with Transformers:

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

pipe = pipeline("text-generation", model="oswaldoludwig/kappatune-lora-tinyllama-agnews")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("oswaldoludwig/kappatune-lora-tinyllama-agnews", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use oswaldoludwig/kappatune-lora-tinyllama-agnews with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/oswaldoludwig/kappatune-lora-tinyllama-agnews

SGLang

How to use oswaldoludwig/kappatune-lora-tinyllama-agnews 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 "oswaldoludwig/kappatune-lora-tinyllama-agnews" \
    --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": "oswaldoludwig/kappatune-lora-tinyllama-agnews",
		"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 "oswaldoludwig/kappatune-lora-tinyllama-agnews" \
        --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": "oswaldoludwig/kappatune-lora-tinyllama-agnews",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use oswaldoludwig/kappatune-lora-tinyllama-agnews with Docker Model Runner:
```
docker model run hf.co/oswaldoludwig/kappatune-lora-tinyllama-agnews
```

kappatune-lora-tinyllama-agnews / README.md

oswaldoludwig

Update README.md

addac3c verified 7 months ago

preview code

Raw History Blame Contribute Delete

3.56 kB

	---
	library_name: peft
	license: apache-2.0
	base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
	tags:
	- base_model:adapter:TinyLlama/TinyLlama-1.1B-Chat-v1.0
	- lora
	- transformers
	- kappaTune
	pipeline_tag: text-generation
	model-index:
	- name: kappatune-lora-tinyllama-agnews
	results: []
	datasets:
	- sh0416/ag_news
	- fancyzhx/ag_news
	- SetFit/ag_news
	language:
	- en
	metrics:
	- perplexity
	---

	<!-- This model card has been generated automatically according to the information the Trainer had access to. You
	should probably proofread and complete it, then remove this comment. -->

	# kappatune-lora-tinyllama-agnews

	This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0) on the AG News dataset. It was trained to demonstrate the KappaTune fine-tuning methodology, which selectively adapts neural network layers based on their condition number to mitigate catastrophic forgetting.

	## Model description

	This model is a LoRA (Low-Rank Adaptation) adapter applied to TinyLlama-1.1B. Unlike standard LoRA, which targets manually specified module types (e.g., all `q_proj` or `v_proj` layers), this adapter was trained using the KappaTune PEFT Integration.
	Before training, the KappaTune algorithm performed a Singular Value Decomposition (SVD) on all candidate weight matrices to calculate their Condition Number (kappa).
	* High-kappa tensors (highly specialized, anisotropic weights containing pre-trained knowledge) were frozen.
	* Low-kappa tensors (numerically stable, general-purpose weights of higher output entropy acting as a "raw marble block") were selected for LoRA adaptation.

	This targeted approach ensures the model learns the new domain efficiently while preserving its foundational conversational capabilities.

	* Paper: The Condition Number as a Scale-Invariant Proxy for Information Encoding in Neural Units ([arXiv:2506.16289](https://arxiv.org/abs/2506.16289))
	* Repository: [oswaldoludwig/kappaTune](https://github.com/oswaldoludwig/kappaTune)

	## Intended uses & limitations

	* Methodological Proof of Concept: This model serves primarily as a demonstration of the KappaTune automated layer-selection strategy.
	* Domain Adaptation: It illustrates how a general-purpose chat model can be steered toward a specific formatting and vocabulary style (News Reporting) using a tiny fraction of trainable parameters.

	## Training and evaluation data

	The model was fine-tuned on the AG News dataset, a curated collection of news articles categorized into four topics: World, Sports, Business, and Sci/Tech.

	## Training procedure

	The model was trained using the Hugging Face `Trainer` API combined with the `kappaTune` optimizer wrapper. Although AG News is traditionally a classification benchmark, the data was formatted as a continuous text stream for this experiment. The model was trained on a Causal Language Modeling task (next-token prediction) to adapt its generation style to news reporting, rather than predicting a categorical label.

	### Training hyperparameters

	The following hyperparameters were used during training:
	- learning_rate: 5e-05
	- train_batch_size: 4
	- eval_batch_size: 8
	- seed: 42
	- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
	- lr_scheduler_type: linear
	- num_epochs: 3

	### Training results



	### Framework versions

	- PEFT 0.17.1
	- Transformers 4.55.4
	- Pytorch 2.6.0+cu124
	- Datasets 4.1.1
	- Tokenizers 0.21.4