Instructions to use micrictor/LFM2.5-350M-ShellAI with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use micrictor/LFM2.5-350M-ShellAI with Transformers:

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

pipe = pipeline("text-generation", model="micrictor/LFM2.5-350M-ShellAI")
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("micrictor/LFM2.5-350M-ShellAI")
model = AutoModelForCausalLM.from_pretrained("micrictor/LFM2.5-350M-ShellAI", 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 micrictor/LFM2.5-350M-ShellAI with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "micrictor/LFM2.5-350M-ShellAI"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "micrictor/LFM2.5-350M-ShellAI",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/micrictor/LFM2.5-350M-ShellAI

SGLang

How to use micrictor/LFM2.5-350M-ShellAI 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 "micrictor/LFM2.5-350M-ShellAI" \
    --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": "micrictor/LFM2.5-350M-ShellAI",
		"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 "micrictor/LFM2.5-350M-ShellAI" \
        --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": "micrictor/LFM2.5-350M-ShellAI",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use micrictor/LFM2.5-350M-ShellAI with Docker Model Runner:
```
docker model run hf.co/micrictor/LFM2.5-350M-ShellAI
```

LFM2.5-350M-ShellAI / README.md

micrictor

Publish distilled ShellAI BF16 checkpoint

4666b42 verified about 1 month ago

preview code

Raw History Blame Contribute Delete

1.88 kB

	---
	license: other
	license_name: lfm1.0
	base_model: LiquidAI/LFM2.5-350M
	tags:
	- lfm2
	- distillation
	- bash
	- shellai
	- transformers
	---

	# LFM2.5-350M-ShellAI

	ShellAI Bash-command model distilled at the response level from
	`LiquidAI/LFM2.5-2.6B` into `LiquidAI/LFM2.5-350M`.

	This repository contains the merged BF16 training checkpoint.

	The model is trained to emit exactly one command inside:

	```text
	<shellai-command>command</shellai-command>
	```

	General chat anchors, assistant-only loss, LoRA, one training epoch, early stopping,
	and a pre-publication retention gate are used to reduce catastrophic forgetting.
	No generated command was executed during dataset construction or evaluation.

	## Distillation

	The two models have different vocabularies (128K teacher versus 65,536 student),
	so this uses sequence-level response distillation rather than token-logit KL.
	Teacher candidates that fail the command envelope, primary-utility, or reference-
	similarity gates fall back to the verified dataset reference.

	## Evaluation

	Held-out repository test split (300 examples, BF16 Transformers):

	\| Model \| Exact \| Utility match \| Token F1 \| Valid envelope \|
	\|---\|---:\|---:\|---:\|---:\|
	\| Base 350M \| 5.7% \| 23.0% \| 0.225 \| 100.0% \|
	\| Distilled 350M \| 4.0% \| 39.7% \| 0.305 \| 100.0% \|

	Chat retention used 50 non-shell prompts: shell-envelope leakage remained 0.0%; anchor similarity retained 84.9% of baseline.

	Q8_0 llama.cpp CPU test (same 50-example subset):

	\| Threads \| Token F1 \| Utility match \| Median latency \| Decode \|
	\|---:\|---:\|---:\|---:\|---:\|
	\| 1 \| 0.268 \| 36.0% \| 1278 ms \| 16.6 tok/s \|
	\| 2 \| 0.268 \| 36.0% \| 641 ms \| 32.5 tok/s \|

	## License

	This is a modified derivative of Liquid AI's LFM2.5 weights and is distributed
	under the included LFM Open License v1.0. See `NOTICE` for modification details.