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-flash-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/clairic-flash-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-flash-4bit") config = load_config("mlx-community/clairic-flash-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-flash-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-flash-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-flash-4bit" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent
How to use mlx-community/clairic-flash-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-flash-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-flash-4bit
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use mlx-community/clairic-flash-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-flash-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-flash-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"
File size: 5,655 Bytes
ae97c5d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 | from __future__ import annotations
from typing import TYPE_CHECKING, Any
from . import batch_completion as shared
from . import schema
if TYPE_CHECKING:
from collections.abc import Callable
from transformers import PreTrainedTokenizerBase
def eligible(
record: dict[str, Any],
tasks: list[tuple[str, list[int], Any]],
kwargs: dict[str, Any],
) -> bool:
media = kwargs.get("image_data")
object_lists = any(
(
spec.get("type") == "list"
and isinstance(spec.get("items"), dict)
and (spec["items"].get("type") == "object")
for spec in record.get("fields", {}).values()
)
)
guarded = (
not record.get("_null_recovery")
and (not object_lists)
and (
record.get("_meta", {}).get("construction") == schema.EVIDENCE_CONSTRUCTION
)
and (not record.get("_media_payload"))
and (not record.get("_original_images"))
and all((payload is None for (_, _, payload) in tasks))
and isinstance(media, list)
and (len(media) == len(tasks))
and all((payload is None for payload in media))
and (
not set(kwargs)
- shared._BATCHED_KEYS
- shared._SHARED_KEYS
- {"parse_result", "normalize"}
)
)
return guarded and bool(
shared._ordinary_lists(record, tasks, schema.list_item_name)
)
async def generate(
engine: Any,
record: dict[str, Any],
tasks: list[tuple[str, list[int], Any]],
eos: int,
tokenizer: PreTrainedTokenizerBase,
*,
initial_frontier: int = 0,
next_frontier: int = 0,
parse_result: Callable[..., dict[str, Any]] | None = None,
normalize: Callable[
[dict[str, Any], dict[str, dict[str, Any]]], dict[str, dict[str, Any]]
]
| None = None,
**kwargs: Any,
) -> list[dict[str, Any]]:
enabled = initial_frontier != 0
if enabled and (
type(initial_frontier) is not int
or type(next_frontier) is not int
or initial_frontier < 1
or (next_frontier <= initial_frontier)
):
raise ValueError(
"Ordinary list scheduling requires positive increasing integer endpoints"
)
groups = (
shared._ordinary_lists(record, tasks, schema.list_item_name)
if enabled and eligible(record, tasks, kwargs)
else {}
)
if not groups:
return await shared.generate(
engine,
record,
tasks,
eos,
tokenizer,
parse_result=parse_result,
normalize=normalize,
**kwargs,
)
size = len(tasks)
rids = kwargs.get("rid")
if (
not isinstance(rids, list)
or len(rids) != size
or len(set(rids)) != size
or (not all((isinstance(rid, str) and rid for rid in rids)))
or (kwargs.get("input_ids") != [ids for (_, ids, _) in tasks])
):
raise ValueError(
"Ordinary list scheduling requires original distinct task bindings"
)
shared._select_kwargs(kwargs, list(range(size)), size)
results: list[Any] = [None] * size
unexecuted: dict[str, int] = {}
async def submit(indices: list[int]) -> None:
if not indices:
return
selected = shared._select_kwargs(kwargs, indices, size)
received = await shared._generate(
engine,
record,
[tasks[index] for index in indices],
eos,
tokenizer,
**selected,
)
if len(received) != len(indices):
raise RuntimeError("Ordinary list scheduling received wrong batch size")
for index, result, rid in zip(indices, received, selected["rid"], strict=True):
if result.get("meta_info", {}).get("id") != rid:
raise RuntimeError(
"Ordinary list scheduling received wrong task identity"
)
results[index] = result
tails = {index for group in groups.values() for index in group[initial_frontier:]}
await submit([index for index in range(size) if index not in tails])
active = dict(groups)
(start, end, following) = (0, initial_frontier, next_frontier)
while active:
continued: list[int] = []
for parent, group in list(active.items()):
stopped = False
for index in group[start:end]:
try:
if shared._completed_null(results[index], eos, tokenizer):
stopped = True
break
except (KeyError, TypeError, ValueError, OverflowError):
continue
if stopped:
for index in group[end:]:
(name, ids, _) = tasks[index]
unexecuted[name] = len(ids)
results[index] = {
"_clair_unused_row_proposal": True,
"_ordinary_list_unexecuted": True,
"meta_info": {"completion_tokens": 0},
}
del active[parent]
elif end >= len(group):
del active[parent]
else:
continued.extend(group[end:following])
if continued:
await submit(sorted(continued))
(start, end, following) = (end, following, following * 2)
if any((result is None for result in results)):
raise RuntimeError("Ordinary list scheduling left an unaccounted result")
record["_unexecuted_list_items"] = unexecuted
return results
|