mrysh/qwen3.5-9b-prune_unstructured_50

qwen3.5-9b - prune_unstructured_50

Pruned version of Qwen/Qwen3.5-9B made for a pruning benchmark (4 models x 8 pruning variants). The benchmark compares the variants with each other and with quantization on BFCL, hiring-bias and MedQA, run with vLLM.

• method: unstructured, level 0.5
• library: torch 2.14.0a0+b2c75dd062.nv26.9.68203377
• size on disk: 17.928 GB

Unstructured sparsity: 50.1% of the weights of the decoder linear layers are zero (layer-wise magnitude pruning, no calibration data). The file is dense (zeros are stored as zeros), so size and speed are the same as the base model; only the quality changes.

No distillation or fine-tuning was done after pruning, so expect a quality drop against the base model; the benchmark measures that drop. In the benchmark it is evaluated as a text model (vLLM, thinking off). Same license as the base model; see the base model card.