Instructions to use Grimlabs/GrimBoost-01 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Grimlabs/GrimBoost-01 with Transformers:

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

pipe = pipeline("text-generation", model="Grimlabs/GrimBoost-01")
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("Grimlabs/GrimBoost-01")
model = AutoModelForCausalLM.from_pretrained("Grimlabs/GrimBoost-01", 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 Grimlabs/GrimBoost-01 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Grimlabs/GrimBoost-01"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Grimlabs/GrimBoost-01",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Grimlabs/GrimBoost-01

SGLang

How to use Grimlabs/GrimBoost-01 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 "Grimlabs/GrimBoost-01" \
    --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": "Grimlabs/GrimBoost-01",
		"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 "Grimlabs/GrimBoost-01" \
        --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": "Grimlabs/GrimBoost-01",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Grimlabs/GrimBoost-01 with Docker Model Runner:
```
docker model run hf.co/Grimlabs/GrimBoost-01
```

GrimBoost-01 / tests /test_decode_dispatch.py

cheezeborger

Claude Code

Publish GrimBoost-01 model card, FP32 drafter, and runtime

60e7bb5 verified 14 days ago

Raw History Blame Contribute Delete

1.45 kB

from types import SimpleNamespace

import pytest

from grimboost.targets.hf import HFTarget


@pytest.mark.parametrize('mode,budget,enabled', [
    ('auto', 256, False), ('auto', 8191, False), ('auto', 8192, True),
    ('off', 50000, False), ('on', 256, True),
])
def test_optimization_dispatch_uses_requested_generation_budget(mode, budget, enabled):
    calls = []
    target = HFTarget.__new__(HFTarget)
    target.decode_optimizations = mode
    target.decode_optimizations_fallback = None
    target._decode = SimpleNamespace(enable_optimizations=lambda: calls.append(True))
    target.prepare_generation(budget)
    assert bool(calls) is enabled


def test_auto_optimization_falls_back_once_on_unsupported_configuration():
    calls = []
    def unsupported():
        calls.append(True)
        raise ValueError('unsupported quantization kernel')
    target = HFTarget.__new__(HFTarget)
    target.decode_optimizations = 'auto'
    target.decode_optimizations_fallback = None
    target._decode = SimpleNamespace(enable_optimizations=unsupported)
    target.prepare_generation(50000)
    target.prepare_generation(50000)
    assert calls == [True]
    assert target.decode_optimizations_fallback == 'unsupported quantization kernel'
    target.decode_optimizations = 'on'
    target.decode_optimizations_fallback = None
    with pytest.raises(ValueError, match='unsupported quantization kernel'):
        target.prepare_generation(50000)

	from types import SimpleNamespace

	import pytest

	from grimboost.targets.hf import HFTarget


	@pytest.mark.parametrize('mode,budget,enabled', [
	('auto', 256, False), ('auto', 8191, False), ('auto', 8192, True),
	('off', 50000, False), ('on', 256, True),
	])
	def test_optimization_dispatch_uses_requested_generation_budget(mode, budget, enabled):
	calls = []
	target = HFTarget.__new__(HFTarget)
	target.decode_optimizations = mode
	target.decode_optimizations_fallback = None
	target._decode = SimpleNamespace(enable_optimizations=lambda: calls.append(True))
	target.prepare_generation(budget)
	assert bool(calls) is enabled


	def test_auto_optimization_falls_back_once_on_unsupported_configuration():
	calls = []
	def unsupported():
	calls.append(True)
	raise ValueError('unsupported quantization kernel')
	target = HFTarget.__new__(HFTarget)
	target.decode_optimizations = 'auto'
	target.decode_optimizations_fallback = None
	target._decode = SimpleNamespace(enable_optimizations=unsupported)
	target.prepare_generation(50000)
	target.prepare_generation(50000)
	assert calls == [True]
	assert target.decode_optimizations_fallback == 'unsupported quantization kernel'
	target.decode_optimizations = 'on'
	target.decode_optimizations_fallback = None
	with pytest.raises(ValueError, match='unsupported quantization kernel'):
	target.prepare_generation(50000)