Instructions to use DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic")
model = AutoModelForCausalLM.from_pretrained("DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic

SGLang

How to use DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic 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 "DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic" \
    --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": "DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic" \
        --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": "DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic with Docker Model Runner:
```
docker model run hf.co/DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic
```

Qwen3-Reranker-4B-FP8-Dynamic

File size: 2,239 Bytes

d8c7374

---
base_model: Qwen/Qwen3-Reranker-4B
base_model_relation: quantized
library_name: transformers
tags:
- fp8
- compressed-tensors
- llm-compressor
- vllm
- reranker
---

# Qwen3-Reranker-4B-FP8-Dynamic

[Qwen/Qwen3-Reranker-4B](https://huggingface.co/Qwen/Qwen3-Reranker-4B) quantised to FP8 with [llm-compressor](https://github.com/vllm-project/llm-compressor),
for serving with vLLM.

## What was done

| | |
|---|---|
| Scheme | `FP8_DYNAMIC` — weights static per-channel FP8, activations dynamic per-token FP8 |
| Calibration | none needed; dynamic activation scales are computed at inference time |
| Left in bf16 | `lm_head`, and the token embeddings |
| Weights before | 7.49 GiB |
| Weights after | 4.83 GiB (35% smaller) |

The score is read from the `yes`/`no` logits of the model head, which is left in bf16. Quantising it would put error directly into the number documents are ranked by.

## Quality gate

[`WikipediaRerankingMultilingual`](https://huggingface.co/datasets/mteb/WikipediaRerankingMultilingual) from MTEB — reranking Wikipedia passages in 16 languages, scored by `map_at_1000`.

| language | bf16 | FP8 | Δ |
|---|---|---|---|
| de | 0.9600 | 0.9594 | -0.0006 |
| en | 0.9715 | 0.9726 | +0.0011 |
| it | 0.9710 | 0.9702 | -0.0008 |
| **mean** | **0.9675** | **0.9674** | **-0.0001** |

Languages evaluated: de, en, it. Tolerance: 0.0100 map_at_1000 per language.

**PASSED — no language lost more than 0.0100 map_at_1000.**

## Why

Serving [Qwen/Qwen3-Reranker-4B](https://huggingface.co/Qwen/Qwen3-Reranker-4B) in bf16 leaves little room for KV cache on a small GPU: the weights take
what the cache needs, and the context length has to be cut until it fits. Halving the weights
gives that memory back — the same card serves a longer context without any other change.

## Serving

```bash
vllm serve DCC-BS/Qwen3-Reranker-4B-FP8-Dynamic \
  --max-model-len 32768 \
  --gpu-memory-utilization 0.85
```

The checkpoint is in `compressed-tensors` format, so vLLM detects the quantisation from
`config.json`; no extra flag is required.

FP8 arithmetic is native on Ada and Hopper (compute capability 8.9+). On Ampere it runs through
Marlin: the memory saving still applies, the speed is roughly unchanged.