Instructions to use miceworks/Mark-VL-2.2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use miceworks/Mark-VL-2.2B with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf miceworks/Mark-VL-2.2B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf miceworks/Mark-VL-2.2B:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf miceworks/Mark-VL-2.2B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf miceworks/Mark-VL-2.2B:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf miceworks/Mark-VL-2.2B:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf miceworks/Mark-VL-2.2B:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf miceworks/Mark-VL-2.2B:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf miceworks/Mark-VL-2.2B:Q4_K_M

Use Docker

docker model run hf.co/miceworks/Mark-VL-2.2B:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use miceworks/Mark-VL-2.2B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "miceworks/Mark-VL-2.2B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "miceworks/Mark-VL-2.2B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/miceworks/Mark-VL-2.2B:Q4_K_M

Ollama
How to use miceworks/Mark-VL-2.2B with Ollama:
```
ollama run hf.co/miceworks/Mark-VL-2.2B:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use miceworks/Mark-VL-2.2B with Docker Model Runner:
```
docker model run hf.co/miceworks/Mark-VL-2.2B:Q4_K_M
```

Lemonade

How to use miceworks/Mark-VL-2.2B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull miceworks/Mark-VL-2.2B:Q4_K_M

Run and chat with the model

lemonade run user.Mark-VL-2.2B-Q4_K_M

List all available models

lemonade list

Atomic Chat

miceworks commited on about 12 hours ago

Commit

9718b5d

verified ·

1 Parent(s): 8e82997

fix: add config.json specifying exact 2.22B parameters

Browse files

Files changed (1) hide show

config.json +52 -0

config.json ADDED Viewed

	@@ -0,0 +1,52 @@

+{
+  "architectures": [
+    "MarkVLForCausalLM"
+  ],
+  "model_type": "mark-vl",
+  "model_name": "MarkVL 2.2B Dense Visual Perception Engine",
+  "model_version": "1.0.0",
+  "specialization": "Dense Visual Description, High-Resolution OCR, FACS Facial Decoding, and Random Object Identification",
+  "total_parameters": 2222269696,
+  "active_parameters_per_token": "313.5M FLOPs (Top-2 routed + 1 shared)",
+  "vocab_size": 32000,
+  "hidden_size": 1280,
+  "intermediate_size": 3840,
+  "num_hidden_layers": 14,
+  "num_attention_heads": 16,
+  "num_key_value_heads": 4,
+  "use_mla": true,
+  "q_lora_rank": 640,
+  "kv_lora_rank": 320,
+  "qk_rope_head_dim": 32,
+  "v_head_dim": 96,
+  "n_routed_experts": 24,
+  "n_shared_experts": 1,
+  "num_experts_per_tok": 2,
+  "moe_intermediate_size": 1504,
+  "vision": {
+    "enabled": true,
+    "max_resolution": [2048, 2048],
+    "patch_size": 14,
+    "dim": 768,
+    "layers": 8,
+    "heads": 12,
+    "pixel_shuffle_scale": 2,
+    "token_reduction_ratio": 0.75,
+    "positional_embedding": "2D-RoPE"
+  },
+  "mtp_depth": 1,
+  "max_position_embeddings": 16384,
+  "rms_norm_eps": 1e-05,
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "quantization": "IQ4_NL",
+  "target_platforms": ["LM Studio", "Ollama", "llama.cpp", "PyTorch"],
+  "target_hardware": {
+    "gpu": "NVIDIA GeForce RTX 2060 Super",
+    "vram_capacity_gb": 8.0,
+    "model_vram_gb": 1.85,
+    "free_vram_gb": 6.15,
+    "tokens_per_second": "95 - 125 t/s"
+  }
+}