Instructions to use redashes/Qwen3.8-27B-BF16-SSMFIX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use redashes/Qwen3.8-27B-BF16-SSMFIX with Transformers:

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

pipe = pipeline("image-text-to-text", model="redashes/Qwen3.8-27B-BF16-SSMFIX")
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("redashes/Qwen3.8-27B-BF16-SSMFIX")
model = AutoModelForMultimodalLM.from_pretrained("redashes/Qwen3.8-27B-BF16-SSMFIX", 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 redashes/Qwen3.8-27B-BF16-SSMFIX with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "redashes/Qwen3.8-27B-BF16-SSMFIX"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "redashes/Qwen3.8-27B-BF16-SSMFIX",
		"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/redashes/Qwen3.8-27B-BF16-SSMFIX

SGLang

How to use redashes/Qwen3.8-27B-BF16-SSMFIX 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 "redashes/Qwen3.8-27B-BF16-SSMFIX" \
    --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": "redashes/Qwen3.8-27B-BF16-SSMFIX",
		"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 "redashes/Qwen3.8-27B-BF16-SSMFIX" \
        --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": "redashes/Qwen3.8-27B-BF16-SSMFIX",
		"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 redashes/Qwen3.8-27B-BF16-SSMFIX with Docker Model Runner:
```
docker model run hf.co/redashes/Qwen3.8-27B-BF16-SSMFIX
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

📖 中文版说明 — 中文模型卡

⚠️ Experimental release — read Section 0 and the Disclaimer before use.

Qwen3.8-27B-BF16-SSMFIX (v2 · luffy per-layer α)

A conv1d-repaired Qwen3.8-27B: fixes the SSM scale-drift that silently degrades long-context generation.

Publisher's statement: I release this model not as a recommendation for use in daily life or work, but as a practical verification of a community hypothesis, and as a foundation platform for those interested in researching this field. All test data reflects verification within my personal capability; having more people validate it in more real-world scenarios will allow the truth of this theory to be tested faster and more authentically.

This model applies per-layer α-scaling to the anomalous linear_attn.conv1d.weight tensors in Qwen3.8-27B, following the methodology first disclosed by LuffyTheFox (Sig-ScaleSync) and independently re-implemented by FGDumitru (qwen-ssm-repair) — this release is the quantitative, community cross-validated proof that the fix works.

0. About This Release — an Independent Verification of the Community "Sig-ScaleSync" Investigation

This repository does not claim to be an official or definitive fix. It is a verification experiment around the community investigation first published by LuffyTheFox (Hugging Face: LuffyTheFox), who named his method Sig-ScaleSync (later folded into his broader "Genesis" pipeline). We replicated his core hypothesis independently — measuring conv1d weight-scale drift on the official Qwen3.8-27B weights, applying minimal per-layer α rescaling, and (unlike the original author) subjecting the repaired weights to a full controlled benchmark battery against the official baseline.

LuffyTheFox's original materials:

Main model card, Genesis project (Qwen3.6-35B-A3B series): https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V8-GGUF
Direct analysis of this exact model, Qwen3.8-27B — discussion #38 "Why Qwen3.8-27B overthinks? Here the reason.": https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V8-GGUF/discussions/38

His core thesis, in his own words ("Genesis" concept):

"During training, ALL models don't just learn knowledge – they also accumulate random noise in their tensors. This noise builds up and creates something I call the Noise Gate — a fundamental barrier that stops LLM models from learning further and makes them unstable, verbose, and prone to hallucinations."

"LLM models often have: … Scale mismatches: one layer's weights are 10× larger than its peers for no good reason …"

"On first stage I scan ssm_conv1d tensors in model, they handle long context memory. I repair balance between heads in them."

"My approach fixes all of that without retraining — pure numerical surgery on the raw bytes of the file."

He concluded with a strong claim about this exact model:

"That is also why I will not make Genesis for 27B. You cannot fix this by patching a few tensors or doing SVD to fix noise gate. The SSM input pathway is damaged across too many layers."

What this experiment adds

His diagnosis confirms our independent measurement. The 8 layers we flagged (52/53/56/57/58/60/61/62) are identical to his α-based list, and our applied scale factors (0.481–0.653) match his α range (0.48–0.65).
We tested, rather than asserted. We ran a full controlled battery (GSM8K, CMMLU, TruthfulQA, IFEval, MT-Bench) against the official baseline on identical hardware/stack. Results are in the Evaluation section below.
Verdict vs. his "cannot fix" claim: partial refutation. A small tensor patch did move generative metrics substantially (TruthfulQA-gen +6~8pp) — but it also hurt closed-book knowledge (CMMLU −1.8pp) and slightly reduced conversational quality under the official MT-Bench protocol (−0.19 vs official, see Evaluation). So a few-tensor patch is not a free lunch: it trades a little knowledge and a little dialogue finesse for noticeably better generation/hallucination behavior.

This release is the measurable record of that experiment, not a recommendation to prefer it over the official weights. Use accordingly.

Why this model exists

Qwen 3.5/3.8 hybrid models mix full-attention layers with GatedDeltaNet SSM layers. The SSM recurrence is governed by 1D convolutional weights (linear_attn.conv1d.weight). In the official Qwen3.8-27B weights, 8 of the last layers have a significantly inflated conv1d std (vs. the ~0.042 sibling median):

Layer	α applied	post-fix std
52	0.5901	0.0471
53	0.5548	0.0437
56	0.5449	0.0425
57	0.5357	0.0410
58	0.6097	0.0432
60	0.4814	0.0398
61	0.6533	0.0420
62	0.6186	0.0452

These layers are the same 8 flagged by LuffyTheFox (α 0.48–0.65) and overlap FGDumitru's detection (α 0.61–0.70) — independent implementations, convergent diagnosis. Without repair, the drifted scales let the recurrent state saturate/collapse: long-context (75k+) collapse, repetition loops, mid-generation truncation, and "philosophizing" drift where the model abandons the task. Short-context perplexity looks normal → silent degradation.

Community Cross-Validation

Source	Method	Anomalous layers	α range
LuffyTheFox (HF discussions #38, Sig-ScaleSync/Genesis)	Noise-gate theory, per-layer α	Same 8 (52/53/56/57/58/60/61/62)	0.48–0.65
FGDumitru (`qwen-ssm-repair`, MIT)	MAD Z-score + peer-group median scaling	Overlapping tail layers	0.61–0.70
This release (v2)	Per-layer strict α (Luffy method)	Same 8	0.481–0.653

This release adopts the strict per-layer α from LuffyTheFox (not FGDumitru's median-normalization), because our full evaluation shows it preserves instruction-following and knowledge better (see table below). All weights are bit-exact except the 8 repaired tensors; model_type=qwen3_5 VLM integrity confirmed (visual / linear_attn / mtp intact).

Evaluation (vLLM, identical harness)

Metric	official BF16	v2 (per-layer α, this release)
MT-Bench avg	8.79	8.60
IFEval prompt strict	0.5194	0.5194
IFEval inst strict	0.6247	0.6343
GSM8K strict	0.9606	0.9644
CMMLU	0.7179	0.6996
TruthfulQA mc1 / mc2	0.3647 / 0.5418	0.3758 / 0.5513
TQA gen rouge1/2/L, bleu	0.284/0.162/0.280/0.178	0.345/0.246/0.345/0.256

Takeaways:

7 of 10 metrics ≥ or ≈ official; the notable gaps are CMMLU (−1.8pp, knowledge-heavy) and MT-Bench (−0.19, conversational).
TruthfulQA generation up +6~8pp across the board → strong hallucination reduction (the main measurable win of the repair).
MT-Bench (official protocol, per-category avg): v2 loses most on reasoning (−0.75), writing (−0.45), math (−0.30); gains on humanities (+0.30) and extraction (+0.15) — see the detailed MT-Bench section below.
v1 (median norm) is deprecated and removed from this repo; v2 is the only SSMFIX variant shipped here.

MT-Bench — updated protocol results (2026-08-19)

⚠️ Supersedes the numbers published earlier. The previous MT-Bench scores on this card (7.05 / 7.15 / 7.47) came from a run with a broken harness:

max_model_len=8192 → long reasoning-model answers retried at max_tokens=8192 overflowed and returned HTTP 400 → the judge assigned fake 1.0 scores to ~1/5 of turns.

A single generic judge prompt was used for all categories, whereas the official FastChat protocol uses a dual-track judge: math/reasoning/coding are graded against the official GPT-4 reference answers (single-math-v1), all other categories use single-v1.

Thinking mode was ON (Qwen3.8 defaults to it). The official MT-Bench protocol assumes non-thinking chat models, so the old numbers were not comparable to official leaderboards.

All three issues are fixed in this rerun: thinking OFF (enable_thinking=false), official per-category temperatures (math/coding/reasoning/extraction 0.0, stem/humanities 0.1, writing/roleplay 0.7), dual-track judge with GPT-4 references, and official turn1/turn2 aggregation. Judge: deepseek-v4-flash (temperature 0), 160/160 valid, zero failed turns. Use the numbers below; the old ones are void.

Category	official BF16 (t1/t2/avg)	v2 SSMFIX (t1/t2/avg)	Δ (v2 − official)
Overall	8.96 / 8.61 / 8.79	8.94 / 8.26 / 8.60	−0.19
writing	9.10 / 8.30 / 8.70	8.90 / 7.60 / 8.25	−0.45
roleplay	9.00 / 8.60 / 8.80	8.40 / 8.70 / 8.55	−0.25
reasoning	9.80 / 9.20 / 9.50	9.50 / 8.00 / 8.75	−0.75
math	10.00 / 10.00 / 10.00	10.00 / 9.40 / 9.70	−0.30
coding	9.00 / 8.20 / 8.60	9.10 / 8.00 / 8.55	−0.05
extraction	7.90 / 8.70 / 8.30	8.40 / 8.50 / 8.45	+0.15
stem	8.20 / 7.10 / 7.65	8.20 / 6.80 / 7.50	−0.15
humanities	8.70 / 8.80 / 8.75	9.00 / 9.10 / 9.05	+0.30

Other metrics — why they remain valid (2026-08-19)

The five non-MT-Bench metrics (GSM8K, CMMLU, TruthfulQA, IFEval) run on a different chain than MT-Bench and were not hit by the three contamination mechanisms that voided the old MT-Bench numbers. The evidence below is verified against the actual run artifacts and code, not asserted.

Endpoint: raw /v1/completions, not chat. All five metrics go through lm_eval's local-completions backend, which sends bare text-completion requests. MT-Bench alone uses /v1/chat/completions (chat template applied → Qwen3.8's thinking mode ON by default), which was one of the old-MT-Bench contamination sources. The eval chain never applies the chat template; run logs show huggingface tokenizer backend, no apply_chat_template call.
Thinking is never triggered (tokenizer-verified). Qwen3.8 enters thinking mode only when the chat template injects the "Reasoning effort is set to xhigh…" system directive plus dedicated thinking tokens (248068/248069). We tokenized the real eval prompts with the actual Qwen3.8-27B tokenizer: bare completion prompts contain zero thinking tokens; only chat-template rendering does. So the eval runs are structurally thinking OFF — the same state as the corrected MT-Bench rerun, by construction.
Deterministic generation. lm_eval's completions payload defaults to temperature=0 (verified in LocalCompletionsAPI._create_payload). No sampling variance.
max_gen_toks=2048 is uniform across every result referenced on this card. All result-file model_args snapshots (official / v2 for GSM8K / CMMLU / TQA / IFEval) carry max_gen_toks: 2048. The only run that ever used the 256-token default (an early port-8134 batch, source of the old "gsm8k fake drop") was superseded by -mg2048 reruns and is not referenced here.
Task-type immunity. CMMLU and TruthfulQA-mc1/mc2 are loglikelihood tasks (they score prompt probabilities, they do not generate); GSM8K / TQA-gen / IFEval are generative but run on the raw-completion chain above where thinking is structurally off. All three models were evaluated on identical chains, so every comparison on this card is apples-to-apples.

Conclusion: GSM8K / CMMLU / TruthfulQA / IFEval numbers on this card are trustworthy and protocol-consistent with the corrected MT-Bench rerun (thinking OFF, temperature 0, 2048-token budget).

Usage

from transformers import AutoModelForImageTextToText, AutoProcessor
model = AutoModelForImageTextToText.from_pretrained("redashes/Qwen3.8-27B-BF16-SSMFIX", trust_remote_code=True)
processor = AutoProcessor.from_pretrained("redashes/Qwen3.8-27B-BF16-SSMFIX", trust_remote_code=True)

Provenance

Base: official Qwen/Qwen3.8-27B BF16 (untouched except repaired tensors)
Repair script: per-layer α on model.language_model.layers.<N>.linear_attn.conv1d.weight; atomic shard rewrites with .orig backups; 1199 keys verified, 48 conv1d keys verified, 0 remaining anomalous layers (ratio > 1.6)
Method credit: LuffyTheFox (Sig-ScaleSync) / FGDumitru (qwen-ssm-repair)
Produced by: hermes-nova

Disclaimer

Weights are derived from the official Apache-2.0 release; the Apache 2.0 license is inherited.
Only 8 conv1d tensors were rescaled; all other tensors are bit-identical to the official release.
This model is an independent verification experiment of a community hypothesis (LuffyTheFox's Sig-ScaleSync, cross-validated by FGDumitru). Do not treat it as a production recommendation. Prefer the official weights unless you specifically need the generative-quality profile measured here.
The original author's materials are linked in Section 0; any claims about his method are his own words, quoted verbatim.

License

Apache-2.0 (model weights follow the original Qwen license terms).

Downloads last month: 164

Safetensors

Model size

28B params

Tensor type

BF16

Model tree for redashes/Qwen3.8-27B-BF16-SSMFIX

Base model

Qwen/Qwen3.8-27B

Finetuned

(209)

this model

Quantizations

3 models