snowman0919
/

qwen38-executor-27b-v1

Model card Files Files and versions

Qwen3.8-27B Executor FP16

Merged FP16 model trained with an 8-bit LoRA adapter.

Base revision: 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
Train revision: 17763ac247b5b21f825e14b670723cf087a7990f
Eval revision: d348a8f0d951337f427f37cd027fffa3de26631f
Computer-use revision: b50aeccec6d24a56ed4f8fbb9f5b2a16846b46a9 (CC-BY-4.0 attribution)
Context: 32768
Long text trajectories: response-label-preserving windows with 1024 overlap; no silent target truncation
Text/tool steps: 1500
Vision steps: 500
Checkpoints: snowman0919/qwen38-executor-27b-checkpoints-v1 every 20 steps
Eval primary score: 0.039897 over 64 rows

Downloads last month: 12

Safetensors

Model size

28B params

Tensor type

BF16

·

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for snowman0919/qwen38-executor-27b-v1

Base model

Qwen/Qwen3.8-27B

Adapter

(44)

this model

Datasets used to train snowman0919/qwen38-executor-27b-v1