Instructions to use vtava/Laya-Integrated-Memory-V22 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use vtava/Laya-Integrated-Memory-V22 with Transformers:

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

pipe = pipeline("text-generation", model="vtava/Laya-Integrated-Memory-V22")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("vtava/Laya-Integrated-Memory-V22", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use vtava/Laya-Integrated-Memory-V22 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "vtava/Laya-Integrated-Memory-V22"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "vtava/Laya-Integrated-Memory-V22",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/vtava/Laya-Integrated-Memory-V22

SGLang

How to use vtava/Laya-Integrated-Memory-V22 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 "vtava/Laya-Integrated-Memory-V22" \
    --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": "vtava/Laya-Integrated-Memory-V22",
		"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 "vtava/Laya-Integrated-Memory-V22" \
        --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": "vtava/Laya-Integrated-Memory-V22",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use vtava/Laya-Integrated-Memory-V22 with Docker Model Runner:
```
docker model run hf.co/vtava/Laya-Integrated-Memory-V22
```

Laya-Integrated-Memory-V22 / loader.py

vtava

Upload Laya Integrated Memory V2.2 adapter; accepted layers=[0, 3, 6, 9, 18, 21, 24, 27]

37e177f verified 14 days ago

Raw History Blame Contribute Delete

1.03 kB

	from pathlib import Path
	import torch
	from huggingface_hub import snapshot_download
	import laya
	from tinycenn_lm.laya_lab.core import LayaLabConfig
	from tinycenn_lm.laya_lab.factory import make_replacement


	def load_model(repo_id, device=None, token=None):
	root = Path(snapshot_download(
	repo_id,
	repo_type="model",
	token=token,
	allow_patterns=["adapter.pt", "report.json", "model_meta.json"],
	))
	payload = torch.load(root / "adapter.pt", map_location="cpu", weights_only=False)
	cfg = LayaLabConfig(**payload["lab_config"])
	agent = laya.load(cfg.model_id, device=device, token=token)

	for layer_id, spec in payload["adapters"].items():
	idx = int(layer_id)
	layer = agent.model.encoder.layers[idx]
	replacement = make_replacement(layer.attn, cfg)
	replacement.load_state_dict(spec["state_dict"], strict=True)
	replacement.to(agent.device).eval().requires_grad_(False)
	layer.attn = replacement

	agent.model.eval()
	return agent