Swift-1.5-Qwen3.8-27b-Terse-Coder

Merged weights of ukisai/Swift-1.5-Qwen3.8-27b with the Shockem/Qwen3.8-27b-Terse-Coder-LoRA adapter (round 8, rank-16 DPO) baked in. Swift 1.5 already uses 58.5% fewer thinking tokens than stock Qwen3.8-27B while scoring 0.35% higher on GPQA-Diamond [1]; Terse-Coder adds a further conciseness pass targeted at coding traces. Single checkpoint, no LoRA plumbing.

How it was made

• Merged in fp32: W + B @ A * (lora_alpha / r), with alpha 32 and r 16 (scale 2.0).
• Stored as bf16 using stochastic (unbiased) rounding with a fixed seed (0), so the merge is reproducible.
• MTP head and vision weights untouched — the adapter doesn't modify them, and Swift 1.5 keeps the base model's MTP head, so speculative decoding remains available [1].
• All non-weight files (config, tokenizer, processor, index) copied from the Swift 1.5 base; the chat template is Shockem/froggeric-terse-coder, the one the adapter was evaluated with. Serving without it changes agentic behavior.

Why stochastic rounding

The adapter's weight deltas are deliberately tiny (‖Δ‖/‖W‖ ≈ 4e-4–1e-3), below bf16's per-element resolution. The adapter card measures delta survival of only 31–61% under plain bf16 rounding vs 94–99.9% in fp16. Stochastic rounding is unbiased — each element is rounded up or down with probability weighted so its expected value equals the true merged value — so the delta is preserved on average while keeping the checkpoint at the base's bf16 dtype and size.

Expected behavior and caveats

• This pairing is unmeasured by the adapter's author. The adapter card lists stock Qwen, Signal, and heretic-ara as recommended bases; no Swift base appears in its results tables. The effect should compound as it does on other bases, but the magnitude on Swift 1.5 is unknown, and no independent benchmarks have been run on this artifact.
• The adapter's author recommends runtime LoRA as the full-strength deployment form and measured a capability tax after merging on the stock Qwen base (70% → 60–62% on their held-out-40 after fp32-merge → fp16 → NVFP4 re-quant). This merge stores bf16 with no re-quant, so the tax should be smaller, but it is not zero.
• Swift 1.5 is a deeper fine-tune than Swift 1.0 — built from Swift 1.0 with scaled-up RL and OPD post-training [1] — so two stacked fine-tunes plus a behavioral LoRA is the most layered artifact in this series. If you see odd behavior, runtime LoRA on plain Swift 1.5 is the cleaner diagnostic.
• Do not load the Terse-Coder LoRA on top of this model. Double application over-shortens reasoning (63% pass with no_code failures in the adapter's testing).
• The adapter targets coding tasks with thinking enabled. It is a behavioral edit, not a knowledge edit — if a task needs long derivation, raise reasoning_effort as usual.

How to use

Transformers

from transformers import AutoModelForImageTextToText, AutoProcessor
import torch

model_id = "vwdubb/Swift-1.5-Qwen3.8-27b-Terse-Coder"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

vLLM

vllm serve vwdubb/Swift-1.5-Qwen3.8-27b-Terse-Coder \
  --dtype bfloat16 \
  --tensor-parallel-size 1 \
  --max-model-len 262144 \
  --reasoning-parser qwen3 \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_coder \
  --port 8000

MTP speculative decoding (optional, head is included and untouched) [1]:

--speculative-config '{"method":"mtp","num_speculative_tokens":3}'

License

This model is a derivative of Swift 1.5 Qwen3.8-27B (Swift Open License v1.0) and the Terse-Coder LoRA (Apache 2.0). The merged weights are released under the Swift Open License v1.0: free for personal, research, educational, evaluation, and commercial use by individuals and organizations with gross annual revenue up to US$1,000,000; above that threshold, commercial use requires a separate Swift Enterprise License from UkisAI. Nothing in the Swift Open License limits your rights in Qwen3.8-27B itself under Apache 2.0 [1].

Acknowledgements

• UkisAI — Swift 1.5 Qwen3.8-27B, the base model.
• Shockem — the Terse-Coder LoRA and chat template.
• Qwen Team (Alibaba Cloud) — Qwen3.8-27B.