Text Generation
Transformers
Safetensors
English
qwen3
muon
gqa
rope
swiglu
chatml
conversational
text-generation-inference
Instructions to use 8BitStudio/Sekai-1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use 8BitStudio/Sekai-1 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="8BitStudio/Sekai-1") 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("8BitStudio/Sekai-1") model = AutoModelForCausalLM.from_pretrained("8BitStudio/Sekai-1", 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 8BitStudio/Sekai-1 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "8BitStudio/Sekai-1" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "8BitStudio/Sekai-1", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/8BitStudio/Sekai-1
- SGLang
How to use 8BitStudio/Sekai-1 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 "8BitStudio/Sekai-1" \ --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": "8BitStudio/Sekai-1", "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 "8BitStudio/Sekai-1" \ --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": "8BitStudio/Sekai-1", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use 8BitStudio/Sekai-1 with Docker Model Runner:
docker model run hf.co/8BitStudio/Sekai-1
|
Download README.md from 8BitStudio/Sekai-1: direct link, hf CLI and curl.
- Browser
- Download file 3.75 kB
-
https://huggingface.co/8BitStudio/Sekai-1/resolve/main/README.md
- Command line
-
hf download hf://8BitStudio/Sekai-1/README.md
-
curl -L -o README.md https://huggingface.co/8BitStudio/Sekai-1/resolve/main/README.md
3.75 kB
| license: apache-2.0 | |
| language: | |
| - en | |
| pipeline_tag: text-generation | |
| tags: | |
| - muon | |
| - gqa | |
| - rope | |
| - swiglu | |
| - chatml | |
| library_name: transformers | |
| datasets: | |
| - karpathy/climbmix-400b-shuffle | |
| - HuggingFaceTB/smollm-corpus | |
| - HuggingFaceTB/finemath | |
| - HuggingFaceTB/cosmopedia | |
| - openbmb/UltraX-Preview | |
| - allenai/dolma3_dolmino_mix-100B-1025 | |
| - allenai/dolma3.5_pool | |
| - allenai/dolmino-mix-1124 | |
| - IFM/TxT360-v2 | |
| - nvidia/OpenMathInstruct-2 | |
| - nvidia/Nemotron-MIND | |
| - openbmb/UltraData-Math | |
| - microsoft/orca-math-word-problems-200k | |
| - meta-math/MetaMathQA | |
| - OpenCoder-LLM/opc-annealing-corpus | |
| - HuggingFaceTB/smoltalk | |
| - HuggingFaceTB/smoltalk2 | |
| - HuggingFaceH4/ultrachat_200k | |
| - allenai/tulu-3-sft-mixture | |
| - bigcode/commitpackft | |
|  | |
| Sekai-1 is the first model in the Sekai series: a small language model trained from scratch on a single RTX 5060 Ti. | |
| The model has 95.6M parameters, with 23 layers, a hidden size of 576 and a 2,048-token context window. It uses a Qwen3-style decoder-only architecture (grouped-query attention with 9 query heads and 3 key-value heads, a SwiGLU MLP, RMSNorm, RoPE, tied embeddings) and its own 24,576-token BPE tokenizer, trained on Sekai’s data mix, that splits numbers into single digits. | |
| Sekai-1 is pretrained on 58.5B tokens of web, educational, math, code and conversation data, using the Muon and AdamW optimizers with a warmup-stable-decay learning-rate schedule. | |
| ## Evaluation | |
|  | |
| The model scores number two in the under 100M parameters, just behind [KLM-1.5](https://huggingface.co/MinimaLabs/KLM-1.5) on the [SLM leaderboard](https://huggingface.co/spaces/AxiomicLabs/Open_SLM_Leaderboard) as seen in the image above. | |
| All scores are zero-shot. Sekai-1's scores were measured locally on the same tasks the leaderboard uses; the other models' scores are taken from the leaderboard (November 7, 2026). | |
| ## Training | |
| **Pretraining:** 58.5B tokens of web, educational, math, code and conversation data, with a warmup-stable-decay schedule. The final ~12B tokens were a decay phase on a higher-quality mix. | |
| **Chat fine-tuning (SFT):** about 400M tokens of instruction following, everyday chat, short and long answers, summarizing and rewriting, code writing and code fixing, plus a small set of identity examples, so the model knows it is Sekai-1. It does not use chain-of-thought (a model this size doesn't benefit from it), and refusals were filtered out of the data. | |
| ## How to use | |
| ```bash | |
| pip install transformers torch | |
| ``` | |
| ```python | |
| from transformers import AutoModelForCausalLM, AutoTokenizer | |
| model_id = "8BitStudio/Sekai-1" | |
| tokenizer = AutoTokenizer.from_pretrained(model_id) | |
| model = AutoModelForCausalLM.from_pretrained(model_id) | |
| messages = [{"role": "user", "content": "Who are you?"}] | |
| inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_dict=True, return_tensors="pt") | |
| outputs = model.generate(**inputs) | |
| print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True)) | |
| ``` | |
| The model is small enough to run on a CPU. The default sampling settings (temperature 0.5, top-p 0.9, top-k 40, repetition penalty 1.1) are stored in `generation_config.json`; pass `do_sample=False` to `generate()` for deterministic answers. | |
| Sekai-1 uses the ChatML format: | |
| ``` | |
| <|im_start|>user | |
| What is the capital of France?<|im_end|> | |
| <|im_start|>assistant | |
| Paris<|im_end|> | |
| ``` | |
| ## Limitations | |
| Sekai-1 is a 95.6M-parameter model. It makes factual mistakes, loses track in long conversations, and is weak at multi-step reasoning and math beyond simple arithmetic. | |
| It has no safety tuning or refusal behavior, so don't use it anywhere that needs either. It understands English only. Check responses for the things that matter. |