Instructions to use Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon with Transformers:

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

pipe = pipeline("text-generation", model="Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon

SGLang

How to use Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon 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 "Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon with Docker Model Runner:
```
docker model run hf.co/Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon
```

LoopLM-Llama3Arch-1_2x9_1-Muon

1 prelude layer, 2 recurrent layers for 9 training loops, and 1 coda layer

Model details


Architecture	`BaseLoopLlamaForCausalLM`
Hidden size	1536
Physical layers	4
Effective depth	20
Vocab size	128256

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon")
model = AutoModelForCausalLM.from_pretrained("Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon", torch_dtype="auto")

Downloads last month: 41

Safetensors

Model size

0.5B params

Tensor type

BF16

Collection including Mihara-bot/LoopLM-Llama3Arch-1_2x9_1-Muon

LoopLM

Collection

Model checkpoints in the paper LoopLM. • 27 items • Updated 7 days ago