Instructions to use RayyTien/gemma-4-e2b-uiux-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RayyTien/gemma-4-e2b-uiux-lora with MLX:

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

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("RayyTien/gemma-4-e2b-uiux-lora")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use RayyTien/gemma-4-e2b-uiux-lora with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "RayyTien/gemma-4-e2b-uiux-lora"

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": "RayyTien/gemma-4-e2b-uiux-lora"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use RayyTien/gemma-4-e2b-uiux-lora with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "RayyTien/gemma-4-e2b-uiux-lora"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "RayyTien/gemma-4-e2b-uiux-lora"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "RayyTien/gemma-4-e2b-uiux-lora",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use RayyTien/gemma-4-e2b-uiux-lora 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 "RayyTien/gemma-4-e2b-uiux-lora"

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 RayyTien/gemma-4-e2b-uiux-lora

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use RayyTien/gemma-4-e2b-uiux-lora with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "RayyTien/gemma-4-e2b-uiux-lora"

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 "RayyTien/gemma-4-e2b-uiux-lora" \
  --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"

RayyTien commited on Jun 7

Commit

56cb1e6

verified ·

1 Parent(s): eb2afb0

Add UIUX LoRA adapter and bundled model card

Browse files

Files changed (1) hide show

README.md +46 -13

README.md CHANGED Viewed

@@ -1,25 +1,58 @@
 ---
-library_name: mlx
 license: apache-2.0
 license_link: https://ai.google.dev/gemma/docs/gemma_4_license
-pipeline_tag: any-to-any
 tags:
-- mlx
-base_model: google/gemma-4-e2b-it
 ---
-# mlx-community/gemma-4-e2b-it-4bit
-This model was converted to MLX format from [`google/gemma-4-e2b-it`](https://huggingface.co/google/gemma-4-e2b-it)
-using mlx-vlm version **0.4.3**.
-Refer to the [original model card](https://huggingface.co/google/gemma-4-e2b-it) for more details on the model.
-## Use with mlx
-```bash
-pip install -U mlx-vlm
-```
 ```bash
-python -m mlx_vlm.generate --model mlx-community/gemma-4-e2b-it-4bit --max-tokens 100 --temperature 0.0 --prompt "Describe this image." --image <path_to_image>
 ```

 ---
 license: apache-2.0
 license_link: https://ai.google.dev/gemma/docs/gemma_4_license
+base_model: mlx-community/gemma-4-e2b-it-4bit
+library_name: mlx-vlm
 tags:
+  - mlx
+  - mlx-vlm
+  - lora
+  - gemma4
+  - ui
+  - ux
+  - frontend
+  - design-review
+pipeline_tag: text-generation
 ---
+# Gemma 4 E2B UI/UX LoRA Adapter
+Bundled MLX repo containing the `mlx-community/gemma-4-e2b-it-4bit` base model files plus a UI/UX LoRA adapter.
+This is **not a fused model**. The base weights and LoRA adapter are stored in the same repository so users can download one repo, then pass the same directory as both `--model` and `--adapter-path`.
+The adapter was tuned on UI/UX frontend guidance converted from:
+- `nextlevelbuilder/ui-ux-pro-max-skill` - MIT
+- `pbakaus/impeccable` - Apache-2.0
+## Intended Use
+Frontend UI/UX critique, React/Next.js/Tailwind implementation guidance, design anti-pattern review, accessibility and responsive layout recommendations.
+## Usage
 ```bash
+pip install mlx-vlm
+hf download RayyTien/gemma-4-e2b-uiux-lora --local-dir ./gemma-e2b-uiux
+python -m mlx_vlm.generate \
+  --model ./gemma-e2b-uiux \
+  --adapter-path ./gemma-e2b-uiux \
+  --system "You are a senior frontend engineer and UI/UX reviewer. Give concrete implementation-ready guidance." \
+  --prompt "Review this SaaS dashboard UI: nested cards, purple gradients, tiny gray labels, and no loading state." \
+  --max-tokens 320 \
+  --temperature 0.0
 ```
+## Training Summary
+See `training_args.json` and `dataset_summary.json`.
+Important caveat: this adapter was trained locally with a patched `mlx-vlm` Gemma 4 loader because the unpatched loader did not instantiate K/V projections for weights present in the checkpoint. If vanilla `mlx-vlm` fails to load this model, use the same local patch or wait for upstream Gemma 4 loader support to catch up.
+## Known Limitations
+- This is a bundled base-model-plus-adapter repo, not a fused full fine-tuned checkpoint.
+- The adapter improves UI/UX vocabulary and critique style, but should still be evaluated against real frontend tasks before production use.
+- The converted dataset is guidance-heavy; add more paired bad/good UI implementation examples for stricter behavior.