Instructions to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with Transformers:

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

pipe = pipeline("text-generation", model="barbarabhb/nl2sh-qwen25-coder-1.5b-tpu")
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("barbarabhb/nl2sh-qwen25-coder-1.5b-tpu")
model = AutoModelForCausalLM.from_pretrained("barbarabhb/nl2sh-qwen25-coder-1.5b-tpu", 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

llama.cpp

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu 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 barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu: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 barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu: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 barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Use Docker

docker model run hf.co/barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

LM Studio
Jan

vLLM

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

SGLang

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu 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 "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu" \
    --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": "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu",
		"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 "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu" \
        --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": "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with Ollama:
```
ollama run hf.co/barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M
```
Unsloth Desktop

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with Docker Model Runner:
```
docker model run hf.co/barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M
```

Lemonade

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Run and chat with the model

lemonade run user.nl2sh-qwen25-coder-1.5b-tpu-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use barbarabhb/nl2sh-qwen25-coder-1.5b-tpu with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "barbarabhb/nl2sh-qwen25-coder-1.5b-tpu:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

nl2sh-qwen25-coder-1.5b (TPU build)

Natural-language -> shell command model: LoRA fine-tune of Qwen/Qwen2.5-Coder-1.5B-Instruct trained on nl2sh-v3f (228k pairs) on a Kaggle TPU v3-8 with a hand-rolled PyTorch/XLA loop (micro-batch 8 x grad-accum 4, seq 224, assistant-only loss, lr 2e-4 cosine, 1 epoch / 7130 optimizer steps).

Benchmarks (InterCode-ALFA, 300 tasks, temp 0)

variant	size	pass rate
Q4_K_M	0.99 GB	0.6367
Q5_K_M	1.13 GB	0.6233
Q6_K	1.27 GB	0.6200
IQ4_XS	0.90 GB	0.5967
upstream ThorOdinson246/nl2sh-1.5b-Q4_K_M	—	0.620
sibling 120k-pool build (`barbarabhb/nl2sh-qwen25-coder-1.5b-GGUF`, Q4_K_M)	0.99 GB	0.6567

Robustness note: answers bare greetings with plain echo hello (no network-touching commands), addressing whatisit-nl2sh issue #10.

Files

file	use
`model.safetensors` (+ config/tokenizer)	merged fp16 model
`qcoder-tpu-q4_k_m.gguf` / `qcoder-tpu-q6_k.gguf` / `qcoder-tpu-f16.gguf`	llama.cpp quants
`lora-adapter/final_adapter.safetensors`	raw LoRA delta

Chat template as base; serve with --jinja --temp 0.

Downloads last month: 206

Safetensors

Model size

2B params

Tensor type

F16

Model tree for barbarabhb/nl2sh-qwen25-coder-1.5b-tpu

Base model

Qwen/Qwen2.5-1.5B

Finetuned

Qwen/Qwen2.5-Coder-1.5B

Finetuned

Qwen/Qwen2.5-Coder-1.5B-Instruct

Adapter

(187)

this model