Instructions to use dudeman2512/Qwen3.8-2.4T-A95B-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dudeman2512/Qwen3.8-2.4T-A95B-FP8 with Transformers:

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

pipe = pipeline("text-generation", model="dudeman2512/Qwen3.8-2.4T-A95B-FP8")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("dudeman2512/Qwen3.8-2.4T-A95B-FP8")
model = AutoModelForCausalLM.from_pretrained("dudeman2512/Qwen3.8-2.4T-A95B-FP8", 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 dudeman2512/Qwen3.8-2.4T-A95B-FP8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dudeman2512/Qwen3.8-2.4T-A95B-FP8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dudeman2512/Qwen3.8-2.4T-A95B-FP8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dudeman2512/Qwen3.8-2.4T-A95B-FP8

SGLang

How to use dudeman2512/Qwen3.8-2.4T-A95B-FP8 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 "dudeman2512/Qwen3.8-2.4T-A95B-FP8" \
    --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": "dudeman2512/Qwen3.8-2.4T-A95B-FP8",
		"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 "dudeman2512/Qwen3.8-2.4T-A95B-FP8" \
        --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": "dudeman2512/Qwen3.8-2.4T-A95B-FP8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use dudeman2512/Qwen3.8-2.4T-A95B-FP8 with Docker Model Runner:
```
docker model run hf.co/dudeman2512/Qwen3.8-2.4T-A95B-FP8
```

Qwen3.8-2.4T-A95B-FP8

fp8 quantization of Qwen/Qwen3.8-2.4T-A95B, produced with compressed-tensors by streaming the checkpoint tensor-by-tensor (the model is never fully instantiated).

8-bit float weights (float8_e4m3fn), per output channel, with dynamic per-token activation quantization. Highest fidelity of the set and the largest.

All quantizations of this model

Variant	Format	Size	vs BF16	Mean rel. error	Linears quantized
Qwen3.8-2.4T-A95B-FP8 ← this one	`float-quantized`	2453.05 GB	50%	0.0264	143569
Qwen3.8-2.4T-A95B-NVFP4	`nvfp4-pack-quantized`	1382.45 GB	28%	0.0952	143569
Qwen3.8-2.4T-A95B-int4	`pack-quantized`	1268.00 GB	26%	0.1118	143569

Mean relative error is ||dequant(W) - W|| / ||W||, averaged over a sample of quantized Linear layers, measured against the original BF16 weights. Lower is better.

This variant


Format	`float-quantized`
Weight bits	8
Group size	per-channel (no grouping)
Strategy	`channel`
Linears quantized	143569
Left in BF16	0
Shards	572
On disk	2453.05 GB
Mean relative error	0.0264
Shape/dtype conformance failures	0

Use with vLLM

vllm serve dudeman2512/Qwen3.8-2.4T-A95B-FP8

How this was made

Every produced tensor is checked for shape/dtype conformance against what the server expects, then reconstruction error is measured against the source BF16 weights, before anything is published. The numbers in the table above are those measurements — not estimates.

Downloads last month: 137

Safetensors

Model size

2.4T params

Tensor type

BF16

F8_E4M3

Model tree for dudeman2512/Qwen3.8-2.4T-A95B-FP8

Base model

Qwen/Qwen3.8-2.4T-A95B

Quantized

(33)

this model