Instructions to use KSP-NMAI/Boris-250M-Instruct-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use KSP-NMAI/Boris-250M-Instruct-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="KSP-NMAI/Boris-250M-Instruct-GGUF",
	filename="Boris-250M-Instruct-BF16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use KSP-NMAI/Boris-250M-Instruct-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

Use Docker

docker model run hf.co/KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use KSP-NMAI/Boris-250M-Instruct-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "KSP-NMAI/Boris-250M-Instruct-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "KSP-NMAI/Boris-250M-Instruct-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

Ollama
How to use KSP-NMAI/Boris-250M-Instruct-GGUF with Ollama:
```
ollama run hf.co/KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M
```

Unsloth Studio

How to use KSP-NMAI/Boris-250M-Instruct-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for KSP-NMAI/Boris-250M-Instruct-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for KSP-NMAI/Boris-250M-Instruct-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for KSP-NMAI/Boris-250M-Instruct-GGUF to start chatting

Atomic Chat new
Docker Model Runner
How to use KSP-NMAI/Boris-250M-Instruct-GGUF with Docker Model Runner:
```
docker model run hf.co/KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M
```

Lemonade

How to use KSP-NMAI/Boris-250M-Instruct-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull KSP-NMAI/Boris-250M-Instruct-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Boris-250M-Instruct-GGUF-Q4_K_M

List all available models

lemonade list

Boris-250M-Instruct-GGUF

GGUF quantizations of KSP-NMAI/Boris-250M-Instruct for llama.cpp and compatible runtimes (llama-server, LM Studio, koboldcpp, Jan).

The original safetensors weights live in the base repo — use those for finetuning or for any PyTorch-based runtime. GGUF is inference-only.

Which file should I pick?

Use Q8_0, or F16 if you want the exact reference weights.

TQ1_0, TQ2_0 and Q1_0 were built and tested but produced unusable output at this scale, so they are deliberately not published here.

Boris-250M is a small model, and quantization behaves differently at this scale than it does for 7B+ models. The token embedding table is a large fraction of the parameters and is kept at high precision by llama.cpp, which sets a hard floor on file size. The practical result: every file here is between 146 MB and 481 MB. Dropping from Q8_0 to IQ1_S saves you a few tens of megabytes while degrading output substantially. The aggressive quants are provided for completeness, not because they are a good trade.

Files

File	Quant	Size	Notes
`Boris-250M-Instruct-F16.gguf`	F16	481M	Reference. Unquantized conversion of the safetensors weights.
`Boris-250M-Instruct-BF16.gguf`	BF16	481M	Reference, bfloat16.
`Boris-250M-Instruct-Q8_0.gguf`	Q8_0	259M	Effectively lossless. Recommended.
`Boris-250M-Instruct-Q6_K.gguf`	Q6_K	244M	Near-lossless.
`Boris-250M-Instruct-Q5_K_M.gguf`	Q5_K_M	209M	Very good quality.
`Boris-250M-Instruct-Q5_K_S.gguf`	Q5_K_S	198M
`Boris-250M-Instruct-Q5_1.gguf`	Q5_1	202M
`Boris-250M-Instruct-Q5_0.gguf`	Q5_0	191M
`Boris-250M-Instruct-Q4_K_M.gguf`	Q4_K_M	199M	Standard 4-bit default for larger models.
`Boris-250M-Instruct-Q4_K_S.gguf`	Q4_K_S	184M
`Boris-250M-Instruct-Q4_1.gguf`	Q4_1	179M
`Boris-250M-Instruct-Q4_0.gguf`	Q4_0	168M
`Boris-250M-Instruct-IQ4_NL.gguf`	IQ4_NL	168M
`Boris-250M-Instruct-IQ4_XS.gguf`	IQ4_XS	166M
`Boris-250M-Instruct-Q3_K_L.gguf`	Q3_K_L	187M
`Boris-250M-Instruct-Q3_K_M.gguf`	Q3_K_M	176M
`Boris-250M-Instruct-Q3_K_S.gguf`	Q3_K_S	160M
`Boris-250M-Instruct-IQ3_M.gguf`	IQ3_M	168M
`Boris-250M-Instruct-IQ3_S.gguf`	IQ3_S	160M
`Boris-250M-Instruct-IQ3_XS.gguf`	IQ3_XS	160M
`Boris-250M-Instruct-IQ3_XXS.gguf`	IQ3_XXS	157M
`Boris-250M-Instruct-Q2_K.gguf`	Q2_K	160M
`Boris-250M-Instruct-Q2_K_S.gguf`	Q2_K_S	155M
`Boris-250M-Instruct-IQ2_M.gguf`	IQ2_M	153M
`Boris-250M-Instruct-IQ2_S.gguf`	IQ2_S	152M
`Boris-250M-Instruct-IQ2_XS.gguf`	IQ2_XS	151M
`Boris-250M-Instruct-IQ2_XXS.gguf`	IQ2_XXS	149M
`Boris-250M-Instruct-IQ1_M.gguf`	IQ1_M	147M
`Boris-250M-Instruct-IQ1_S.gguf`	IQ1_S	146M

All quantizations below 8-bit were produced with an importance matrix calibrated on 100 chunks of held-out data drawn from the model's own training mixture (60% fineweb-edu / 40% dclm).

Usage

# straight from the Hub
llama-server -hf KSP-NMAI/Boris-250M-Instruct-GGUF:Q8_0 --jinja

# or a local file
llama-server -m Boris-250M-Instruct-Q8_0.gguf --jinja

The Alpaca chat template is embedded in every file, so --jinja applies the correct prompt format automatically.

Prompt format

### Instruction:
{your instruction}

### Response:

Limitations

This is a very small instruction-tuned model. It will produce text that is frequently inaccurate, inconsistent, or offensive, and has received no alignment or safety tuning beyond supervised fine-tuning on Alpaca. Do not rely on it for factual information or deploy it without supervision.

License

Downloads last month: 771

GGUF

Model size

0.3B params

Architecture

gpt2

Hardware compatibility

1-bit

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

16-bit

Model tree for KSP-NMAI/Boris-250M-Instruct-GGUF

Base model

KSP-NMAI/Boris-250M

Finetuned

KSP-NMAI/Boris-250M-Instruct

Quantized

(2)

this model

Dataset used to train KSP-NMAI/Boris-250M-Instruct-GGUF

Collection including KSP-NMAI/Boris-250M-Instruct-GGUF

Boris

Collection

9 items • Updated 5 days ago