Image-Text-to-Text
Transformers
Safetensors
qwen3_5
clairic
cloudflare
systemone
multimodal
structured-output
information-extraction
custom-code
conversational
Instructions to use Cloudflare/clairic with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Cloudflare/clairic with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="Cloudflare/clairic") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] pipe(text=messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("Cloudflare/clairic") model = AutoModelForMultimodalLM.from_pretrained("Cloudflare/clairic", device_map="auto") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] inputs = processor.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=256) print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Cloudflare/clairic with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Cloudflare/clairic" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Cloudflare/clairic", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker
docker model run hf.co/Cloudflare/clairic
- SGLang
How to use Cloudflare/clairic with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "Cloudflare/clairic" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Cloudflare/clairic", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "Cloudflare/clairic" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Cloudflare/clairic", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }' - Docker Model Runner
How to use Cloudflare/clairic with Docker Model Runner:
docker model run hf.co/Cloudflare/clairic
File size: 3,306 Bytes
f65336d 524fd9d f65336d 524fd9d f65336d | 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 | """Read typed fields from backbone hidden states."""
from __future__ import annotations
import inspect
import json
from pathlib import Path
from typing import Any
import torch
from safetensors.torch import load_file
from . import model as cm
def load_head(path: Path, *, match_bfloat16: bool = False) -> cm.ClairicHead:
configuration = json.loads((path / "clair_head_config.json").read_text())
accepted = inspect.signature(cm.ClairicHead).parameters
head = cm.ClairicHead(
**{name: value for name, value in configuration.items() if name in accepted}
)
state = load_file(path / "clair_head.safetensors")
if match_bfloat16:
# Serving constructs the head in BF16 before widening it to FP32.
# The length bias has an explicit FP32 initialization.
state = {
name: value.bfloat16().float()
if value.is_floating_point() and name != "length_bias"
else value
for name, value in state.items()
}
head.load_state_dict(state, strict=True)
return head.float().eval()
def read_slots(
states: torch.Tensor, output_weight: torch.Tensor, chunk: int = 256
) -> tuple[torch.Tensor, torch.Tensor]:
ids, probabilities = [], []
for rows in states.split(chunk):
logits = (rows @ output_weight.t()).float()
best = logits.max(dim=-1)
ids.append(best.indices)
probabilities.append(best.values - torch.logsumexp(logits, dim=-1))
return torch.cat(ids), torch.cat(probabilities)
@torch.inference_mode()
def read_record(
head: cm.ClairicHead,
states: torch.Tensor,
item: cm.EncodedRecord,
output_weight: torch.Tensor,
) -> dict[str, Any]:
start, length, kinds, queries, items, slot_fields, slots = (
cm.parse_prediction_layout(
cm.prediction_layout(item),
len(item.input_ids),
with_slots=True,
)
)
views = cm.prediction_views(kinds, queries, items, states.device)
output = head.forward_states(states, start, start + length, views)
if slots:
width = max(map(len, slots))
positions = torch.tensor(
[row + [row[-1]] * (width - len(row)) for row in slots],
dtype=torch.long,
device=states.device,
)
if head.answer_slot_routing:
field_queries = views["query_positions"][slot_fields]
slot_kinds = torch.tensor(
[
cm.ROUTE_KINDS.index(
"tags" if kinds[row] == "text_list" else kinds[row]
)
for row in slot_fields
],
dtype=torch.long,
device=states.device,
)
lengths = torch.tensor(
[len(row) for row in slots], dtype=torch.long, device=states.device
)
slot_states = head.answer_slot_states(
states, field_queries, slot_kinds, positions, lengths, start + length
).flatten(0, 1)
else:
slot_states = states[positions.flatten()]
ids, logprobs = read_slots(slot_states, output_weight)
output["slot_ids"] = ids.view(positions.shape)
output["slot_log_probs"] = logprobs.view(positions.shape)
return output
|