Organization · Verified on Hugging Face

ByteShape

byteshape

Accelerating AI by Learning Optimal Datatypes

Models in Library1

Datasets in Library0

Models on Hugging Face15

Followers545

Compare the published evidence

This profile connects the 1 model and dataset record shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.

Published tasks: Image and text to text.

Reported licenses: apache-2.0. Open a record to review its specific license and conditions.

Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.

Models

Model · Image and text to text

Qwen3.8-27B-GGUF

ByteShape

This is a GGUF-quantized version of Qwen3.8-27B produced with ByteShape's ShapeLearn, which learns the optimal datatype per tensor to maintain high quality even at very low bitlengths. This release replaces our earlier ShapeLearn-Lite quantizations of Qwen3.8-27B. Every model here has been through our full multi-benchmark evaluation suite, covering math, coding, general knowledge, instruction following and agentic tool use in both thinking and instruct modes, scored against the BF16 baseline. To learn more about ShapeLearn and to see detailed benchmarks across GPUs, please visit our blog. If you have questions or want to share feedback, reach us on Reddit. Pick a model from the table below…

Open weights apache-2.0 transformers

View model