Instructions to use YuRiVeRTi/Tensora-Curious-135M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use YuRiVeRTi/Tensora-Curious-135M with Transformers:

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

pipe = pipeline("text-generation", model="YuRiVeRTi/Tensora-Curious-135M")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("YuRiVeRTi/Tensora-Curious-135M")
model = AutoModelForCausalLM.from_pretrained("YuRiVeRTi/Tensora-Curious-135M", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use YuRiVeRTi/Tensora-Curious-135M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "YuRiVeRTi/Tensora-Curious-135M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YuRiVeRTi/Tensora-Curious-135M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/YuRiVeRTi/Tensora-Curious-135M

SGLang

How to use YuRiVeRTi/Tensora-Curious-135M 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 "YuRiVeRTi/Tensora-Curious-135M" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YuRiVeRTi/Tensora-Curious-135M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "YuRiVeRTi/Tensora-Curious-135M" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YuRiVeRTi/Tensora-Curious-135M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use YuRiVeRTi/Tensora-Curious-135M with Docker Model Runner:
```
docker model run hf.co/YuRiVeRTi/Tensora-Curious-135M
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Tensora Curious 135M (TSC-135M)

"Curiosity is the beginning of intelligence."

Overview

Tensora Curious 135M (TSC-135M) is an experimental instruction-following language model developed by Tensoramax Lab.

Rather than pursuing model scale alone, Tensora Curious focuses on creating an AI that is lightweight, transparent, privacy-respecting, and practical for everyday developers. The project serves as the first public milestone toward the long-term Tensoramax Lab vision of building an open ecosystem of local AI models.

The model is built upon HuggingFaceTB/SmolLM2-135M-Instruct, with future iterations being progressively adapted through supervised fine-tuning and LoRA training.

Vision

Tensoramax Lab believes capable AI should not be limited to cloud infrastructure or expensive hardware.

Our long-term goals include:

Offline-first AI
Privacy-focused deployment
Lightweight local models
Transparent development
Open research
Community collaboration
Developer-first tooling
Continuous model improvement

Why Tensora Curious?

Large language models continue to grow in size and hardware requirements.

TSC explores another direction:

Can a compact model remain genuinely useful?
Can developers fine-tune it on modest hardware?
Can students learn AI without enterprise infrastructure?
Can privacy-first assistants run entirely on-device?

This project exists to explore those questions.

Planned Capabilities

Current roadmap includes:

Natural conversation
Programming assistance
Code explanation
Mathematics
Scientific reasoning
Writing assistance
Documentation generation
Educational tutoring
Local AI assistant
Offline inference
Tensoramax ecosystem integration

Technical Overview

Property	Value
Model Name	Tensora Curious 135M
Alias	TSC-135M
Developer	Tensoramax Lab
Base Model	HuggingFaceTB/SmolLM2-135M-Instruct
Architecture	SmolLM2
Parameters	135 Million
Training Method	LoRA (planned)
Current Status	Alpha
License	Apache-2.0 (base model)

Current Development Status

✅ Repository created
✅ Base model duplicated
✅ Documentation prepared
⏳ Dataset construction
⏳ LoRA fine-tuning
⏳ Local evaluation
⏳ GGUF export
⏳ Alpha release

Intended Use

Recommended:

Education
Coding experiments
AI research
Offline assistants
Personal projects
Rapid prototyping

Not recommended:

High-risk decision making
Medical advice
Legal advice
Financial decisions
Safety-critical environments

Attribution

This repository contains a derivative work based on:

HuggingFaceTB/SmolLM2-135M-Instruct

Please refer to the original repository, paper, and license for details regarding the base model architecture and pretraining.

Tensoramax Lab claims authorship only over future fine-tuning, datasets, documentation, branding, and improvements added in this repository.

Acknowledgements

Special thanks to the Hugging Face team and the SmolLM2 researchers for releasing an open base model that enables independent research and experimentation.

Built with ❤️ by Tensoramax Lab

Downloads last month: 87

Safetensors

Model size

0.1B params

Tensor type

BF16

Model tree for YuRiVeRTi/Tensora-Curious-135M

Base model

HuggingFaceTB/SmolLM2-135M

Quantized

HuggingFaceTB/SmolLM2-135M-Instruct

Finetuned

(361)

this model

Quantizations

1 model