Instructions to use AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3.5-9B")
model = PeftModel.from_pretrained(base_model, "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history")

Transformers

How to use AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history with Transformers:

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

pipe = pipeline("text-generation", model="AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history

SGLang

How to use AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history 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 "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history" \
    --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": "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history",
		"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 "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history" \
        --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": "AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history with Docker Model Runner:
```
docker model run hf.co/AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history
```

Qwen3.5-9B-EldenRing-LoRA-video-history

LoRA adapter for Qwen/Qwen3.5-9B trained on Elden Ring gameplay with video history.

This repo is not the last-step adapter. It is the closest surviving checkpoint to the 1-epoch point from the video-history run:

Source checkpoint: checkpoint-6048
Saved epoch: 1.0921
Base model: Qwen/Qwen3.5-9B
LoRA: rank 64, alpha 128, LLM-only
Input format: 10 history frames at 5 fps + current frame + instruction
Output format: raw gamepad action string inside <|act_start|> ... <|act_end|>

Use

Load this adapter on top of Qwen/Qwen3.5-9B with PEFT. The tokenizer and processor files included here match the training run and contain the added action boundary tokens.

Notes

This is the video-history variant.
It was selected as the best available approximation to the 1-epoch checkpoint after older checkpoints were pruned during training.
The exact 1.0-epoch checkpoint was no longer present in remote storage, so checkpoint-6048 was chosen as the nearest surviving checkpoint.

Downloads last month: 1

Model tree for AntimLabs/Qwen3.5-9B-EldenRing-LoRA-video-history

Base model

Qwen/Qwen3.5-9B-Base

Finetuned

Qwen/Qwen3.5-9B

Adapter

(514)

this model