Instructions to use cagataydev/loopl-2b-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cagataydev/loopl-2b-4bit with MLX:

# Make sure mlx-vlm is installed
# pip install --upgrade mlx-vlm

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("cagataydev/loopl-2b-4bit")
config = load_config("cagataydev/loopl-2b-4bit")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use cagataydev/loopl-2b-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "cagataydev/loopl-2b-4bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "cagataydev/loopl-2b-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use cagataydev/loopl-2b-4bit with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "cagataydev/loopl-2b-4bit"

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 cagataydev/loopl-2b-4bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use cagataydev/loopl-2b-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "cagataydev/loopl-2b-4bit"

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 "cagataydev/loopl-2b-4bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

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

loopl-2b-4bit

Commit History

Oct 7 28 days ago

11:06
card: loopl 2B v1.2 (mlx) — train/cards.py

cagataydev

Verified signature
8ead044
11:05
card: loopl 2B v1 (mlx) — train/cards.py

cagataydev

Verified signature
38cc1a7
11:05
assets: brand banner + app captures — train/brand_assets.py

cagataydev

Verified signature
192ac68
10:54
loopl v1.2-c — promoted from cagataydev/loopl-2b-v1.2-c-4bit @ 95b350df (DATASET v2.2 rev 47d27b44: tool-trajectory family t over all 30 app tools — mount/wake/tap/pressure, render_ui, ask_user, system_prompt — on top of v2.1's identity/bio families); the previous version is the preceding commit

cagataydev

Verified signature
7539b6f
07:26
card: loopl 2B v1.1 (mlx) — train/cards.py

cagataydev

Verified signature
fc74cef

Oct 6 29 days ago

19:07
card: loopl 2B v1.1 (mlx) — train/cards.py

cagataydev

Verified signature
e26f67a
19:07
assets: brand banner + app captures — train/brand_assets.py

cagataydev

Verified signature
1a27c40
18:10
card: loopl 2B v1.1 (mlx) — train/cards.py

cagataydev

Verified signature
c34864e
15:40
loopl v1.1 — promoted from cagataydev/loopl-2b-v1.1-4bit @ 0ef19414 (dataset v2.1 rev db6fa452: identity/bio family, file cards, repair rows); v1 is the previous commit

cagataydev

Verified signature
5a7c7d7
15:29
card: loopl 2B v1 (mlx) — train/cards.py

cagataydev

Verified signature
e5a6b91
13:01
vision-preserving re-export: mlx_vlm 0.7.6 4-bit g64 from cagataydev/loopl-2b (297 vision_tower tensors, vision_config, preprocessor configs); tokenizer normalised from mlx-community/Qwen3.5-2B-4bit

cagataydev

Verified signature
8c2afdb
12:26
tokenizer.json: the mlx-community base's serialization (identical vocab/merges/added tokens; transformers-5 pre_tokenizer/decoder shapes are unknown to swift-transformers)

cagataydev

Verified signature
3ff8064
12:25
tokenizer: class Qwen2Tokenizer (swift-transformers has no Qwen3_5Tokenizer) + chat_template.jinja/vocab from the mlx-community base — fixes 'Qwen3_5Tokenizer is not supported' on iOS

cagataydev

Verified signature
144738f
05:10
Add files using upload-large-folder tool

cagataydev

Verified signature
5bc0fd1
05:09
initial commit

cagataydev

Verified signature
cbb241f

Commit History

Oct 7 28 days ago

card: loopl 2B v1.2 (mlx) — train/cards.py

card: loopl 2B v1 (mlx) — train/cards.py

assets: brand banner + app captures — train/brand_assets.py

card: loopl 2B v1.1 (mlx) — train/cards.py

Oct 6 29 days ago

card: loopl 2B v1.1 (mlx) — train/cards.py

assets: brand banner + app captures — train/brand_assets.py

card: loopl 2B v1.1 (mlx) — train/cards.py

loopl v1.1 — promoted from cagataydev/loopl-2b-v1.1-4bit @ 0ef19414 (dataset v2.1 rev db6fa452: identity/bio family, file cards, repair rows); v1 is the previous commit

card: loopl 2B v1 (mlx) — train/cards.py

vision-preserving re-export: mlx_vlm 0.7.6 4-bit g64 from cagataydev/loopl-2b (297 vision_tower tensors, vision_config, preprocessor configs); tokenizer normalised from mlx-community/Qwen3.5-2B-4bit

tokenizer.json: the mlx-community base's serialization (identical vocab/merges/added tokens; transformers-5 pre_tokenizer/decoder shapes are unknown to swift-transformers)

tokenizer: class Qwen2Tokenizer (swift-transformers has no Qwen3_5Tokenizer) + chat_template.jinja/vocab from the mlx-community base — fixes 'Qwen3_5Tokenizer is not supported' on iOS

Add files using upload-large-folder tool

initial commit