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

Libraries

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

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

pipe = pipeline("text-generation", model="KSP-NMAI/Boris-250M-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("KSP-NMAI/Boris-250M-Instruct")
model = AutoModelForCausalLM.from_pretrained("KSP-NMAI/Boris-250M-Instruct", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use KSP-NMAI/Boris-250M-Instruct 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"
# 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",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

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

SGLang

How to use KSP-NMAI/Boris-250M-Instruct 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 "KSP-NMAI/Boris-250M-Instruct" \
    --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": "KSP-NMAI/Boris-250M-Instruct",
		"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 "KSP-NMAI/Boris-250M-Instruct" \
        --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": "KSP-NMAI/Boris-250M-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use KSP-NMAI/Boris-250M-Instruct with Docker Model Runner:
```
docker model run hf.co/KSP-NMAI/Boris-250M-Instruct
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Boris-250M-Instruct

Boris-250M-Instruct is the instruction-tuned variant of KSP-NMAI/Boris-250M, a 250 million-parameter language model created by New Millennium Artificial Intelligence (NMAI). It was fine-tuned on tatsu-lab/alpaca.

Prompt format

This model uses the Alpaca format. A chat template is included in tokenizer_config.json, so apply_chat_template produces the correct prompt automatically:

from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("KSP-NMAI/Boris-250M-Instruct")
model = AutoModelForCausalLM.from_pretrained("KSP-NMAI/Boris-250M-Instruct")

messages = [{"role": "user", "content": "What is the capital of France?"}]
prompt = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

ids = tok(prompt, return_tensors="pt").input_ids
out = model.generate(ids, max_new_tokens=64, do_sample=True, top_p=0.95, temperature=0.7)
print(tok.decode(out[0][ids.shape[1]:], skip_special_tokens=True))

If you are building the prompt by hand, the layout is:

### Instruction:
{your instruction}

### Response:

Generation should stop at ### Instruction: (or end-of-text, token id 0).

Details


Architecture	GPT-2 (pre-LN, learned positional embeddings, tied embeddings)
Layers / heads / d_model	12 / 18 / 1152
Context length	1024
Vocab	50304 (GPT-NeoX-20B BPE, padded)
Tokenizer	`EleutherAI/gpt-neox-20b`
Precision	trained in bf16 autocast with fp32 master weights

Base model training

Trained on 5.01B tokens for ~105:56:17 on one RTX 3060.


Final loss	3.0693
Final grad norm	0.225
Final learning rate	6.00e-05

The table above describes the base model's pretraining run; the instruction tuning was applied on top of that checkpoint.

Limitations

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

Copyright & License

This project and all associated files (the "Work") are licensed under the Apache License, Version 2.0 (the "License"); you may not use this project except in compliance with the License. You may obtain a copy of the License at:

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.