--- license: other license_name: swift-open-license-1.0 license_link: https://huggingface.co/ukisai/Swift-1.5-Qwen3.8-27b/blob/main/LICENSE library_name: transformers pipeline_tag: image-text-to-text base_model: ukisai/Swift-1.5-Qwen3.8-27b base_model_relation: finetune tags: - qwen3_8 - swift - swift-1.5 - uncensored - abliterated - bf16 - efficient-thinking - reasoning - multimodal - tool-calling - conversational --- > **Related models:** [all models](https://huggingface.co/d0xin) > > **[Swift-1.5-Qwen3.8-27B-Uncensored-BF16](https://huggingface.co/d0xin/Swift-1.5-Qwen3.8-27B-Uncensored-BF16)** · [Swift-1.5-Qwen3.8-27B-Uncensored-FP8](https://huggingface.co/d0xin/Swift-1.5-Qwen3.8-27B-Uncensored-FP8) · [Swift-1.5-Qwen3.8-27B-Uncensored-FP8-NInfer](https://huggingface.co/d0xin/Swift-1.5-Qwen3.8-27B-Uncensored-FP8-NInfer) · [Swift-1.5-Qwen3.8-Flash-Next-NVFP4-FP8PLE](https://huggingface.co/d0xin/Swift-1.5-Qwen3.8-Flash-Next-NVFP4-FP8PLE) · [Swift-1.5-Qwen3.8-Flash-Next-Rank2-Abliteration-Patch](https://huggingface.co/d0xin/Swift-1.5-Qwen3.8-Flash-Next-Rank2-Abliteration-Patch) · [Swift-Qwen3.8-27B-FP8](https://huggingface.co/d0xin/Swift-Qwen3.8-27B-FP8) · [Swift-Qwen3.8-27B-Uncensored-BF16](https://huggingface.co/d0xin/Swift-Qwen3.8-27B-Uncensored-BF16) · [Swift-Qwen3.8-27B-Uncensored-FP8](https://huggingface.co/d0xin/Swift-Qwen3.8-27B-Uncensored-FP8) · [Swift-Qwen3.8-27B-Uncensored-NVFP4-LocalHessian-ActivationHeadroom-NInfer](https://huggingface.co/d0xin/Swift-Qwen3.8-27B-Uncensored-NVFP4-LocalHessian-ActivationHeadroom-NInfer) # Swift-1.5-Qwen3.8-27B-Uncensored-BF16 > 🔓 Independent uncensored BF16 derivative of `ukisai/Swift-1.5-Qwen3.8-27b`. > > Produced using rank-1 directional residual-stream ablation. > > Designed to preserve Swift 1.5's reasoning-efficient, coding, agentic, > tool-calling, multimodal and long-context capabilities while reducing refusal behavior. ## What makes this build different? This model is derived directly from [`ukisai/Swift-1.5-Qwen3.8-27b`](https://huggingface.co/ukisai/Swift-1.5-Qwen3.8-27b). Swift 1.5 is the updated UkisAI Swift checkpoint, with expanded post-training focused particularly on coding and long-horizon agentic workloads. This release applies rank-1 directional residual-stream ablation to the Swift 1.5 BF16 checkpoint. Behavioral, capability-preservation and performance evaluations for this specific uncensored derivative are ongoing and may be added to this model card in a later revision. ## Model summary - BF16 checkpoint - approximately 52 GB - 18 safetensors shards - configured context: 262,144 tokens - ablation layer: 38 - rank: 1 - modified residual writers: 131 - vision tower unchanged by the ablation - MTP residual writers included - multimodal architecture retained - tool-calling architecture retained ## Model lineage ```text Qwen/Qwen3.8-27B ↓ ukisai/Swift-Qwen3.8-27b ↓ ukisai/Swift-1.5-Qwen3.8-27b ↓ rank-1 directional residual-stream ablation ↓ d0xin/Swift-1.5-Qwen3.8-27B-Uncensored-BF16 ``` ## Ablation The uncensoring transformation uses rank-1 directional residual-stream ablation. Release metadata: - source: `ukisai/Swift-1.5-Qwen3.8-27b` - precision: BF16 - ablation layer: 38 - rank: 1 - modified residual writers: 131 - vision tower: unchanged - MTP residual writers: included - checkpoint shards: 18 Transformation metadata is included in `ABLITERATION.json`. ## Validation status The release artifact has passed checkpoint-level structural validation. Current verified properties include: - all 18 safetensors shards are present - model index is present - model configuration is present - tokenizer assets are present - chat template is present - multimodal preprocessing configuration is present - MTP weights are retained - ablation metadata is included Behavioral and performance benchmark results are intentionally not reported yet for this derivative. They can be added without changing the released weights. ## Recommended generation settings Swift 1.5 uses the Qwen3.8 reasoning interface. A good starting point based on the upstream Swift 1.5 serving configuration: | Parameter | Value | |---|---:| | `reasoning_effort` | `xhigh` | | `temperature` | `1.0` | | `top_p` | `0.95` | | `top_k` | `20` | | `min_p` | `0.0` | | `presence_penalty` | `0.0` | | `repetition_penalty` | `1.0` | ## SGLang ```bash python -m sglang.launch_server \ --model-path d0xin/Swift-1.5-Qwen3.8-27B-Uncensored-BF16 \ --served-model-name Swift-1.5-Qwen3.8-27B-Uncensored-BF16 \ --trust-remote-code \ --context-length 262144 \ --reasoning-parser qwen3 \ --tool-call-parser qwen3_coder \ --port 30000 ``` Adjust context length and memory parameters for the available hardware. ## Tool calling Use: ```text --tool-call-parser qwen3_coder ``` ## Multimodal support The Qwen3.8 multimodal architecture and vision tower are retained. The directional ablation process does not modify the vision tower. ## Safety and responsible use This is an uncensored / refusal-reduced derivative model. The model has been intentionally modified to reduce refusal behavior. As a result, it may generate content that the upstream model would normally refuse, restrict, or handle more cautiously. Outputs may be inaccurate, offensive, unsafe, unlawful, or otherwise inappropriate for a particular use case. Users are responsible for evaluating model outputs and ensuring that their use complies with applicable laws, regulations, licenses, platform policies and other requirements. Do not rely on model output without appropriate review where errors could cause harm, financial loss, security incidents or other significant consequences. ## Upstream model Derived from: `ukisai/Swift-1.5-Qwen3.8-27b` Swift 1.5 is UkisAI's reasoning-efficient derivative of Qwen3.8-27B and an updated version of the original Swift checkpoint. ## License These weights are distributed under the Swift Open License v1.0. The underlying Qwen3.8 components remain subject to their applicable upstream license. See the included `LICENSE`, `LICENSE-APACHE-2.0`, and `NOTICE` files for the applicable terms and attribution requirements. ## Attribution - Base model: `Qwen/Qwen3.8-27B` - Swift post-training: UkisAI - Swift 1.5: `ukisai/Swift-1.5-Qwen3.8-27b` - Directional ablation and release packaging: `d0xin` ## Citation ```bibtex @misc{swift-1.5-qwen3.8-27b, title = {Swift 1.5 Qwen3.8-27B}, author = {UkisAI}, year = {2026}, url = {https://huggingface.co/ukisai/Swift-1.5-Qwen3.8-27b} } ```