Image-Text-to-Text
MLX
Safetensors
qwen3_5
clairic
cloudflare
systemone
structured-output
information-extraction
multimodal
custom-code
conversational
4-bit precision
Instructions to use mlx-community/clairic-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/clairic-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("mlx-community/clairic-4bit") config = load_config("mlx-community/clairic-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
- Pi
How to use mlx-community/clairic-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 "mlx-community/clairic-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": "mlx-community/clairic-4bit" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent
How to use mlx-community/clairic-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 "mlx-community/clairic-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 mlx-community/clairic-4bit
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use mlx-community/clairic-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 "mlx-community/clairic-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 "mlx-community/clairic-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"
Download clairic_runtime/processor.py from mlx-community/clairic-4bit: direct link, hf CLI and curl.
- Browser
- Download file 3.51 kB
-
https://huggingface.co/mlx-community/clairic-4bit/resolve/main/clairic_runtime/processor.py
- Command line
-
hf download hf://mlx-community/clairic-4bit/clairic_runtime/processor.py
-
curl -L -o processor.py https://huggingface.co/mlx-community/clairic-4bit/resolve/main/clairic_runtime/processor.py
3.51 kB
| """A torch-free stand-in for the Transformers `Qwen3VLProcessor` used by the Clairic request pipeline. | |
| Not part of the Cloudflare release. The upstream pipeline calls four things on its processor: the chat | |
| template, `replace_image_token`, `get_text_with_replacements`, and `processor(text=..., images=...)`. This | |
| class provides exactly those on top of the tokenizer and mlx-vlm's NumPy image processor, so Transformers' | |
| torchvision-based image and video processors are not needed. | |
| """ | |
| from __future__ import annotations | |
| import json | |
| from pathlib import Path | |
| from typing import Any | |
| import numpy as np | |
| MIN_PIXELS = 65_536 | |
| class ClairicProcessor: | |
| image_token = "<|image_pad|>" | |
| def __init__(self, path: Path, tokenizer: Any, max_pixels: int) -> None: | |
| from mlx_vlm.models.qwen3_vl.processing_qwen3_vl import Qwen3VLImageProcessor | |
| config = json.loads((path / "preprocessor_config.json").read_text()) | |
| self.tokenizer = tokenizer | |
| self.image_token_id = tokenizer.convert_tokens_to_ids(self.image_token) | |
| self.image_processor = Qwen3VLImageProcessor( | |
| patch_size=config["patch_size"], temporal_patch_size=config["temporal_patch_size"], | |
| merge_size=config["merge_size"], min_pixels=config["size"].get("shortest_edge", MIN_PIXELS), | |
| max_pixels=max_pixels, image_mean=config["image_mean"], image_std=config["image_std"]) | |
| def apply_chat_template(self, messages: list[dict[str, Any]], **kwargs: Any) -> Any: | |
| return self.tokenizer.apply_chat_template(messages, **kwargs) | |
| def replace_image_token(self, image_inputs: dict[str, Any], image_idx: int) -> str: | |
| merge = self.image_processor.merge_size ** 2 | |
| return self.image_token * (int(np.prod(np.asarray(image_inputs["image_grid_thw"])[image_idx])) // merge) | |
| def get_text_with_replacements(self, text: list[str], images_replacements: list[str] = (), | |
| videos_replacements: list[str] = (), audio_replacements: list[str] = ()) -> tuple[list[str], list]: | |
| """Replace each image placeholder, in order across the batch, with its expanded run of placeholders.""" | |
| if videos_replacements or audio_replacements: | |
| raise ValueError("Clairic accepts images only") | |
| replacements = iter(images_replacements) | |
| expanded = [] | |
| for sample in text: | |
| pieces = sample.split(self.image_token) | |
| if len(images_replacements) == 0: | |
| expanded.append(sample) | |
| continue | |
| expanded.append("".join(piece + (next(replacements) if index < len(pieces) - 1 else "") | |
| for index, piece in enumerate(pieces))) | |
| return expanded, [] | |
| def __call__(self, text: list[str], images: list[Any], return_tensors: str | None = None) -> dict[str, Any]: | |
| processed = self.image_processor(list(images)) | |
| media = {"pixel_values": np.asarray(processed["pixel_values"], dtype=np.float32), | |
| "image_grid_thw": np.asarray(processed["image_grid_thw"], dtype=np.int64)} | |
| expanded, _ = self.get_text_with_replacements( | |
| list(text), [self.replace_image_token(media, index) for index in range(len(images))]) | |
| ids = np.asarray(self.tokenizer(expanded, add_special_tokens=False)["input_ids"], dtype=np.int64) | |
| return {"input_ids": ids, "attention_mask": np.ones_like(ids), **media, | |
| "mm_token_type_ids": (ids == self.image_token_id).astype(np.int64)} | |