Instructions to use Learn57130/buddy-1.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Learn57130/buddy-1.0 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("Learn57130/buddy-1.0")

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 Learn57130/buddy-1.0 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Learn57130/buddy-1.0"

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": "Learn57130/buddy-1.0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use Learn57130/buddy-1.0 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "Learn57130/buddy-1.0"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "Learn57130/buddy-1.0"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "Learn57130/buddy-1.0",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use Learn57130/buddy-1.0 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 "Learn57130/buddy-1.0"

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 Learn57130/buddy-1.0

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Learn57130/buddy-1.0 with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Learn57130/buddy-1.0"

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 "Learn57130/buddy-1.0" \
  --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"

Buddy 1.0

Buddy 1.0 is the on-device AI model in Languagly, a Mac menu-bar app by Learn Level UP for Thai speakers who write English at work. It translates English → Thai and checks English grammar, on the Mac, with no account and no network once downloaded.

Base model: Gemma 4 E2B (google/gemma-4-E2B-it), packed from the MLX 4-bit conversion mlx-community/gemma-4-e2b-it-4bit.

Website: https://learn57130.github.io/languagly-site

Modified from Gemma

These files are a modified version of Gemma 4 E2B — a Model Derivative under the Gemma Terms of Use. What was changed:

Fine-tuned with LoRA on English → Thai finance translation pairs, plus grammar-check examples taken from the base model's own answers so that its checking behaviour is kept. The adapter was merged into the weights, which were then quantized to 4-bit.
Audio encoder removed.
Image encoder reduced to an empty stub (zero layers). The model is meant for text only.
Vocabulary trimmed to Thai and English: 167,787 of 262,144 tokens kept, with vocab_size set to match in config.json.

languagly-pack.json records the packing so the app does not repack the files.

Use

Made for the Languagly app, which downloads it and loads it with mlx-swift-lm's Gemma 4 multimodal loader. It has not been tested with other runtimes.

On Languagly's own 200-sentence grammar set, Buddy 1.0 corrected 116 of 126 sentences with a mistake and changed 3 of 74 sentences that were already correct. That set is the app's, not a public benchmark.

Gemma Terms of Use

Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms.

Use of Buddy 1.0 is subject to the Gemma Terms of Use and the Gemma Prohibited Use Policy, which apply to anyone who uses or redistributes these files. Buddy and Languagly are not affiliated with or endorsed by Google.

Integrity

model.safetensors SHA-256: a144b71fb00ac2dfd732b1650b6ef8e9921e4ecc34ccbbca50392b576ad59ad7

Downloads last month: 48

Safetensors

Model size

4B params

Tensor type

BF16

U32

MLX

Hardware compatibility

4-bit

Model tree for Learn57130/buddy-1.0

Base model

google/gemma-4-E2B

Finetuned

google/gemma-4-E2B-it

Finetuned

(419)

this model