Downloads · 30 days
59
23% of all-time downloads
Swift-Qwen3.8-27B-EXL3
erlidev/Swift-Qwen3.8-27B-EXL3
Swift-Qwen3.8-27B-EXL3 is a image-text-to-text model from erlidev. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for exllamav3. The card lists the license as other.
EXL3 quantizations of ukisai/Swift-Qwen3.8-27b, with calibration and quantized vision towers.
All-time downloads
259
Public
Repo size
81.5 GB
Likes
13
Public
Repo makeup
Click a slice to open those files.
.md2.8 KB · 64%
At a glance
- Task
- Image-Text-to-Text
- Library
- exllamav3
- License
- other
- Access
- Public
- Created
- Sep 16, 2026
- Updated
- Sep 16, 2026
- SHA
- f801f265
Try a prompt
Base models
Model card
From the Hugging Face model README
Swift-Qwen3.8-27B-EXL3
EXL3 quantizations of ukisai/Swift-Qwen3.8-27b, with calibration and quantized vision towers.
Each bitrate is on its own branch. Pick a revision:
hf download erlidev/Swift-Qwen3.8-27B-EXL3 --revision SC_4.00bpw_H5_V6 --local-dir ./Swift-exl3-4.0bpw
| branch | body bpw | lm_head | vision | size | predicted KLD¹ |
|---|---|---|---|---|---|
SC_3.00bpw_H4_V4 | 3.00 | 4 | 4 | 13.04 GB | 0.0385 |
SC_3.50bpw_H4_V6 | 3.50 | 4 | 6 | 14.67 GB | 0.0200 |
SC_4.00bpw_H5_V6 | 4.00 | 5 | 6 | 16.35 GB | 0.0100 |
SC_4.50bpw_H5_V6 | 4.50 | 5 | 6 | 17.87 GB | 0.0052 |
SC_5.00bpw_H6_V6 | 5.00 | 6 | 6 | 19.55 GB | 0.0026 |
MTP layers are quantized at 4 bpw throughout. ¹ Values are from sc_optimize calculated from the base model, not the Swift finetune. Should be relatively close, however.
Requirements
ExLlamaV3 ≥ 1.4.4 (earlier versions cannot load a quantized vision tower). Built with 1.5.0
Method
These are self-calibrated (SC_) quants, following the method turboderp used for
turboderp/Qwen3.8-27B-exl3:
- A calibration trace self-sampled by the original Qwen3.8 base model (250 × 2048 token rows, thinking on,
reasoning_effort: medium). - Per-tensor sensitivity measurements reused from turboderp's
noise_attrib.jsonfor the base model. sc_optimize.py --alpha 2.0per-tensor bit recipe for each target bitrate, rather than a uniform bitrate (seems to be all the hype recently for some reason). Tied groups:k_proj+v_proj,gate_proj+up_proj.convert.pywith that recipe and the trace as calibration data; vision tower quantized with an identity Hessian (no vision calibration).
License
Swift weights are distributed under the Swift Open License v1.0, and that license covers these quants as derivative works.
Credits
- Task
- Image-Text-to-Text
- Library
- exllamav3
- License
- other
- Created
- Sep 16, 2026
- Updated
- Sep 16, 2026