Instructions to use WaveCut/Qwen3.8-Flash-Next-REAM-288 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WaveCut/Qwen3.8-Flash-Next-REAM-288 with Transformers:

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

pipe = pipeline("image-text-to-text", model="WaveCut/Qwen3.8-Flash-Next-REAM-288")
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)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("WaveCut/Qwen3.8-Flash-Next-REAM-288")
model = AutoModelForMultimodalLM.from_pretrained("WaveCut/Qwen3.8-Flash-Next-REAM-288", 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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use WaveCut/Qwen3.8-Flash-Next-REAM-288 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WaveCut/Qwen3.8-Flash-Next-REAM-288"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WaveCut/Qwen3.8-Flash-Next-REAM-288",
		"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/WaveCut/Qwen3.8-Flash-Next-REAM-288

SGLang

How to use WaveCut/Qwen3.8-Flash-Next-REAM-288 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 "WaveCut/Qwen3.8-Flash-Next-REAM-288" \
    --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": "WaveCut/Qwen3.8-Flash-Next-REAM-288",
		"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 "WaveCut/Qwen3.8-Flash-Next-REAM-288" \
        --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": "WaveCut/Qwen3.8-Flash-Next-REAM-288",
		"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 WaveCut/Qwen3.8-Flash-Next-REAM-288 with Docker Model Runner:
```
docker model run hf.co/WaveCut/Qwen3.8-Flash-Next-REAM-288
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Qwen3.8-Flash-Next-REAM-288

Qwen3.8-Flash-Next with 288 of 512 experts per MoE layer, merged not pruned (REAM), in native bf16 for transformers. 251GB vs 360GB stock. MTP block included (its experts merged 512->288 as well) — vLLM-style qwen4_exp speculative decoding keeps working.

Method

REAM (arXiv 2604.04356, Samsung SAIL Montréal: Merging Improves Pruning of Experts in LLMs), offline adaptation: instead of dropping the 224 non-kept experts per layer (REAP), each is merged into its most-similar kept expert with saliency-proportional coefficients.

Kept set / slot order: exactly the per-layer REAP-288 manifest published by sh0wie (saliency calibrated on ~686K tokens of agentic-coding traffic) — full credit to that work; this build is a direct A/B against it.
Similarity (offline, no calibration run): cosine of router gate rows.
Merge coefficients: measured saliency from sh0wie's saliency_full_505.npz (its argsort reproduces his pinned manifest bit-for-bit), tempered and shrunk: w_i = (S_i/S_c)^2 * router_cos * count_confidence, with the centroid keeping at least 90% of itself. Raw proportional weights are catastrophic here (KL 1.397) because the real kept/dropped saliency gap is only ~1.6x — the merge dissolves the expert REAP chose to keep. Hungarian permutation alignment of intermediate channels before averaging (fp32 accumulate, bf16 out).
Trajectory flat_tsg_tsD: flat one-shot groups (canonical REAM)
Router: absorbed rows deleted; kept rows byte-identical to stock.
MTP block: same machinery, k-center centroids on its own gate rows.
A six-point ladder (see campaign/reports/truesal_ladder.md) locates the optimum: KL is minimised when kept experts give away ~9% of their mass — both heavier blending and near-pruning are worse.

Evaluation (A/B vs REAP prune, same harness, same hardware)

model	KL vs stock (mean)	rare-name probe (10 names)	HumanEval subset (25 tasks)
stock 512e	0	0.97	90.0% (full 50)
REAP-288 recon (prune)	1.3066	0.91	92.0% (stock on same 25: 92.0%)
REAM-288 flat_tsg_tsD (this)	1.2564	0.94	not re-run for tsD; flat_sim (previous main, same pipeline) scored 88.0% vs stock 88.0% on its slice

KL: teacher-forced top-256 KL vs the stock model over a fixed mixed corpus (en/ru/code/rare-entities/agentic, ~10K positions). Probe: rare entities x 10 seeded sampled generations, pass = name reproduced intact; aggregates recomputed on the name subset common to all rows (a monitoring bug killed the winner's third probe batch mid-run). HumanEval: 25-task slices of a frozen 50-task subset, no-thinking, greedy; the two 288e models landed on different slices (2-task overlap), so each is paired with the stock score on its own slice. Single-run evals; ±1-2 point differences are noise. RU slice of the probe (3 Cyrillic names, 30 gens): stock 27, prune 21, merge 24.

Files

Standard HF sharding. experts-layerNN shards carry the merged expert stacks; backbone tensors are byte-identical to stock (verified by streaming sha256).

Limitations

Merge coefficients use the author's measured saliency, tempered (see Method).
Calibration-free similarity (router-row cosine): no activation statistics.
MTP merged without measured saliency (k-center on gate rows).
Vision tower untouched but untested after merging.

Credits

sh0wie / HamsterResearch — REAP-288 saliency calibration, kept-expert manifests, MLX conversion conventions.
Samsung SAIL Montréal — the REAM method.
Qwen team — the base model. License inherited (Qwen Community License 1.0).

Downloads last month: 60

Safetensors

Model size

126B params

Tensor type

BF16

I64

Model tree for WaveCut/Qwen3.8-Flash-Next-REAM-288

Base model

Qwen/Qwen3.8-Flash-Next

Finetuned

(25)

this model

Quantizations

2 models

Paper for WaveCut/Qwen3.8-Flash-Next-REAM-288

REAM: Merging Improves Pruning of Experts in LLMs

Paper • 2604.04356 • Published Apr 6 • 10