Instructions to use robbiemu/MobileLLM-R1-950M-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use robbiemu/MobileLLM-R1-950M-MLX 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("robbiemu/MobileLLM-R1-950M-MLX")

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 robbiemu/MobileLLM-R1-950M-MLX with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "robbiemu/MobileLLM-R1-950M-MLX"

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": "robbiemu/MobileLLM-R1-950M-MLX"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use robbiemu/MobileLLM-R1-950M-MLX with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "robbiemu/MobileLLM-R1-950M-MLX"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "robbiemu/MobileLLM-R1-950M-MLX"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "robbiemu/MobileLLM-R1-950M-MLX",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use robbiemu/MobileLLM-R1-950M-MLX 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 "robbiemu/MobileLLM-R1-950M-MLX"

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 robbiemu/MobileLLM-R1-950M-MLX

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use robbiemu/MobileLLM-R1-950M-MLX with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "robbiemu/MobileLLM-R1-950M-MLX"

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 "robbiemu/MobileLLM-R1-950M-MLX" \
  --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"

MobileLLM-R1-950M-MLX / MobileLLM-R1-950M-mixed-4bit-mlx /config.json

robbiemu

mlx-lm conversion and quantization artifacts

1f30715 about 1 year ago

Raw History Blame Contribute Delete

9.4 kB

	{
	"architectures": [
	"Llama4ForCausalLM"
	],
	"attention_bias": false,
	"attention_chunk_size": 32768,
	"attention_dropout": 0.0,
	"attn_scale": 0.1,
	"attn_temperature_tuning": false,
	"bos_token_id": 128000,
	"eos_token_id": [
	128001,
	128008,
	128009
	],
	"floor_scale": 8192,
	"for_llm_compressor": false,
	"head_dim": 64,
	"hidden_act": "silu",
	"hidden_size": 1536,
	"initializer_range": 0.02,
	"interleave_moe_layer_step": 0,
	"intermediate_size": 8192,
	"intermediate_size_mlp": 6144,
	"layer_types": [
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention",
	"full_attention"
	],
	"max_position_embeddings": 32768,
	"model_type": "llama4_text",
	"moe_layers": [],
	"no_rope_layers": [
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1,
	1
	],
	"num_attention_heads": 24,
	"num_experts_per_tok": 0,
	"num_hidden_layers": 22,
	"num_key_value_heads": 6,
	"num_local_experts": 0,
	"output_router_logits": false,
	"rms_norm_eps": 1e-05,
	"rope_scaling": null,
	"rope_theta": 8000000.0,
	"router_aux_loss_coef": 0.001,
	"router_jitter_noise": 0.0,
	"tie_word_embeddings": true,
	"torch_dtype": "float32",
	"transformers_version": "4.55.0",
	"use_cache": true,
	"use_qk_norm": true,
	"vocab_size": 128256,
	"quantization": {
	"group_size": 64,
	"method": "mixed_precision_dynamic",
	"per_layer_bits": {
	"layers.0.attention.q_proj": 4,
	"layers.0.attention.k_proj": 4,
	"layers.0.attention.v_proj": 4,
	"layers.0.attention.o_proj": 8,
	"layers.0.feed_forward.gate_proj": 4,
	"layers.0.feed_forward.up_proj": 4,
	"layers.0.feed_forward.down_proj": 4,
	"layers.1.attention.q_proj": 4,
	"layers.1.attention.k_proj": 4,
	"layers.1.attention.v_proj": 4,
	"layers.1.attention.o_proj": 4,
	"layers.1.feed_forward.gate_proj": 4,
	"layers.1.feed_forward.up_proj": 4,
	"layers.1.feed_forward.down_proj": 4,
	"layers.2.attention.q_proj": 4,
	"layers.2.attention.k_proj": 4,
	"layers.2.attention.v_proj": 4,
	"layers.2.attention.o_proj": 4,
	"layers.2.feed_forward.gate_proj": 4,
	"layers.2.feed_forward.up_proj": 4,
	"layers.2.feed_forward.down_proj": 4,
	"layers.3.attention.q_proj": 4,
	"layers.3.attention.k_proj": 4,
	"layers.3.attention.v_proj": 4,
	"layers.3.attention.o_proj": 4,
	"layers.3.feed_forward.gate_proj": 4,
	"layers.3.feed_forward.up_proj": 4,
	"layers.3.feed_forward.down_proj": 4,
	"layers.4.attention.q_proj": 4,
	"layers.4.attention.k_proj": 4,
	"layers.4.attention.v_proj": 4,
	"layers.4.attention.o_proj": 4,
	"layers.4.feed_forward.gate_proj": 4,
	"layers.4.feed_forward.up_proj": 4,
	"layers.4.feed_forward.down_proj": 4,
	"layers.5.attention.q_proj": 4,
	"layers.5.attention.k_proj": 4,
	"layers.5.attention.v_proj": 4,
	"layers.5.attention.o_proj": 4,
	"layers.5.feed_forward.gate_proj": 4,
	"layers.5.feed_forward.up_proj": 4,
	"layers.5.feed_forward.down_proj": 4,
	"layers.6.attention.q_proj": 4,
	"layers.6.attention.k_proj": 4,
	"layers.6.attention.v_proj": 4,
	"layers.6.attention.o_proj": 4,
	"layers.6.feed_forward.gate_proj": 4,
	"layers.6.feed_forward.up_proj": 4,
	"layers.6.feed_forward.down_proj": 4,
	"layers.7.attention.q_proj": 4,
	"layers.7.attention.k_proj": 4,
	"layers.7.attention.v_proj": 4,
	"layers.7.attention.o_proj": 4,
	"layers.7.feed_forward.gate_proj": 4,
	"layers.7.feed_forward.up_proj": 4,
	"layers.7.feed_forward.down_proj": 4,
	"layers.8.attention.q_proj": 4,
	"layers.8.attention.k_proj": 4,
	"layers.8.attention.v_proj": 4,
	"layers.8.attention.o_proj": 4,
	"layers.8.feed_forward.gate_proj": 4,
	"layers.8.feed_forward.up_proj": 4,
	"layers.8.feed_forward.down_proj": 4,
	"layers.9.attention.q_proj": 4,
	"layers.9.attention.k_proj": 4,
	"layers.9.attention.v_proj": 4,
	"layers.9.attention.o_proj": 4,
	"layers.9.feed_forward.gate_proj": 4,
	"layers.9.feed_forward.up_proj": 4,
	"layers.9.feed_forward.down_proj": 4,
	"layers.10.attention.q_proj": 4,
	"layers.10.attention.k_proj": 4,
	"layers.10.attention.v_proj": 4,
	"layers.10.attention.o_proj": 4,
	"layers.10.feed_forward.gate_proj": 4,
	"layers.10.feed_forward.up_proj": 4,
	"layers.10.feed_forward.down_proj": 4,
	"layers.11.attention.q_proj": 4,
	"layers.11.attention.k_proj": 4,
	"layers.11.attention.v_proj": 4,
	"layers.11.attention.o_proj": 4,
	"layers.11.feed_forward.gate_proj": 4,
	"layers.11.feed_forward.up_proj": 4,
	"layers.11.feed_forward.down_proj": 4,
	"layers.12.attention.q_proj": 4,
	"layers.12.attention.k_proj": 4,
	"layers.12.attention.v_proj": 4,
	"layers.12.attention.o_proj": 4,
	"layers.12.feed_forward.gate_proj": 4,
	"layers.12.feed_forward.up_proj": 4,
	"layers.12.feed_forward.down_proj": 4,
	"layers.13.attention.q_proj": 4,
	"layers.13.attention.k_proj": 4,
	"layers.13.attention.v_proj": 4,
	"layers.13.attention.o_proj": 4,
	"layers.13.feed_forward.gate_proj": 4,
	"layers.13.feed_forward.up_proj": 4,
	"layers.13.feed_forward.down_proj": 4,
	"layers.14.attention.q_proj": 4,
	"layers.14.attention.k_proj": 4,
	"layers.14.attention.v_proj": 4,
	"layers.14.attention.o_proj": 4,
	"layers.14.feed_forward.gate_proj": 4,
	"layers.14.feed_forward.up_proj": 4,
	"layers.14.feed_forward.down_proj": 4,
	"layers.15.attention.q_proj": 4,
	"layers.15.attention.k_proj": 4,
	"layers.15.attention.v_proj": 4,
	"layers.15.attention.o_proj": 4,
	"layers.15.feed_forward.gate_proj": 4,
	"layers.15.feed_forward.up_proj": 4,
	"layers.15.feed_forward.down_proj": 4,
	"layers.16.attention.q_proj": 4,
	"layers.16.attention.k_proj": 4,
	"layers.16.attention.v_proj": 4,
	"layers.16.attention.o_proj": 4,
	"layers.16.feed_forward.gate_proj": 4,
	"layers.16.feed_forward.up_proj": 4,
	"layers.16.feed_forward.down_proj": 4,
	"layers.17.attention.q_proj": 4,
	"layers.17.attention.k_proj": 4,
	"layers.17.attention.v_proj": 4,
	"layers.17.attention.o_proj": 4,
	"layers.17.feed_forward.gate_proj": 4,
	"layers.17.feed_forward.up_proj": 4,
	"layers.17.feed_forward.down_proj": 4,
	"layers.18.attention.q_proj": 4,
	"layers.18.attention.k_proj": 4,
	"layers.18.attention.v_proj": 4,
	"layers.18.attention.o_proj": 4,
	"layers.18.feed_forward.gate_proj": 4,
	"layers.18.feed_forward.up_proj": 4,
	"layers.18.feed_forward.down_proj": 4,
	"layers.19.attention.q_proj": 4,
	"layers.19.attention.k_proj": 4,
	"layers.19.attention.v_proj": 4,
	"layers.19.attention.o_proj": 4,
	"layers.19.feed_forward.gate_proj": 4,
	"layers.19.feed_forward.up_proj": 4,
	"layers.19.feed_forward.down_proj": 4,
	"layers.20.attention.q_proj": 4,
	"layers.20.attention.k_proj": 4,
	"layers.20.attention.v_proj": 4,
	"layers.20.attention.o_proj": 4,
	"layers.20.feed_forward.gate_proj": 4,
	"layers.20.feed_forward.up_proj": 4,
	"layers.20.feed_forward.down_proj": 4,
	"layers.21.attention.q_proj": 4,
	"layers.21.attention.k_proj": 4,
	"layers.21.attention.v_proj": 4,
	"layers.21.attention.o_proj": 4,
	"layers.21.feed_forward.gate_proj": 4,
	"layers.21.feed_forward.up_proj": 4,
	"layers.21.feed_forward.down_proj": 4
	}
	}
	}