license: apache-2.0
library_name: transformers
pipeline_tag: image-text-to-text
base_model: Qwen/Qwen3.8-27B
base_model_relation: finetune
tags:

• qwen3.8
• qwen3.5
• multimodal
• image-text-to-text
• reasoning
• tool-calling
• long-context
• 1m-context
• uncensored
• abliterated
• obliteratus
• supertune
• vllm
• bf16
• full-precision
  language:
• en
• ko

SuperQwen3.8-27b-abliterated is a directly loadable, full-BF16 weight release built from
Qwen/Qwen3.8-27B@1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0.
It applies a measured rank-1 refusal-direction edit while preserving the official
vision tower and MTP weights exactly. No LoRA or inference-time adapter is required.

Release highlights

Why this release

• Original-weight quality: BF16 transformer, vision, output head, embeddings, and MTP storage.
• Less refusal without wrecking capabilities: capability floor 7/8, tool call PASS, vision PASS.
• Reasoning that stops: all nine deterministic tasks pass at default, low, medium, and xhigh.
• Multimodal preserved: this remains an image-text-to-text checkpoint, not a text-only conversion.
• Reproducible: the exact parent revision, modified tensor list, and evidence hashes ship with the model.

Bounded reasoning

The upstream template defaulted unspecified reasoning to xhigh. This release defaults
to medium and adds a stop condition to xhigh: once an answer is established, the
model must stop instead of repeating or restarting its deliberation. The checkpoint was
tested across 36 deterministic effort/task combinations; all 36 terminated correctly.

Explicit controls remain available through chat_template_kwargs:

extra_body={"chat_template_kwargs": {"enable_thinking": True, "reasoning_effort": "xhigh"}}

Behavior and capability

Precision and integrity

• Full BF16 checkpoint; this repository is not quantized.
• 100 declared tensors changed and zero unexpected tensors changed.
• Vision (333 tensors) and MTP (15 tensors) remain byte/value exact.
• Parent revision is pinned to 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0.

Decode performance

Measured on one DGX Spark with fixed-length generation and the
sparkDash-style post-first-token contract:

Verified long context

The official native limit is 262,144 tokens. With the included YaRN recipe, a
1,000,045-token prompt completed end to end and retrieved its hidden needle.
Long-context acceptance is not a claim of perfect recall on every task.

Serving

QWEN38_SPECULATIVE_TOKENS=0   bash repro/scripts/serve_superqwen38_replica.sh   /model SuperQwen3.8-27b-abliterated 8888

For the measured 1M profile, also set QWEN38_MAX_MODEL_LEN=1048576,
QWEN38_MAX_BATCHED_TOKENS=2048, and
QWEN38_EXECUTE_MODEL_TIMEOUT_SECONDS=1800, plus the YaRN HF_OVERRIDES_JSON
shown in repro/.

Uncensored behavior

“Abliterated” means that the measured refusal direction was reduced. It does not mean
that every response is correct, harmless, or suitable for every deployment. Operators
remain responsible for access controls and downstream safeguards.

Limitations

• Abliteration changes refusal behavior and may surface content the parent declined.
• The capability, tool, vision, and overthinking suites are finite regression gates.
• Speed is hardware- and runtime-specific.
• YaRN 1M context trades additional latency for reach and is separate from the native window.

Evidence identities

License

Apache-2.0, following the upstream Qwen3.8 release.