PreDoctor/gemma-4-31B-it-qat-w4a16-ct-PreTensor

Gemma 4 31B W4A16 PreTensor

This repository contains the instruction-tuned Gemma 4 31B checkpoint packaged for PreTensor's local NVIDIA CUDA runtime.

• Language weights: compressed-tensors packed INT4, symmetric group size 32
• Language activations and KV cache: BF16
• Vision input: source BF16 tensors, copied without quantization
• Audio input: not present in the 31B source checkpoint
• Tensor payloads: copied from the source checkpoint without requantization
• Runtime path: native NVIDIA CUDA host; WebGPU payloads are not included

The files are valid sharded safetensors. See pretensor_config.json for the storage contract, source revision, component inventory, and per-shard SHA-256.

Long-context prefill validation

Warm native measurements on NVIDIA RTX PRO 6000 Blackwell Max-Q (CUDA device 0, driver 620.12), after model/graph and exact KV-capacity warm-up. Each result is the median of 3 runs.

Input tokens	Prefill	Native wall time
4,096	3,301 tok/s	1,240.9 ms
16,384	2,426 tok/s	6,752.6 ms

These figures validate this PreTensor implementation on the stated machine; they are not general hardware-independent model benchmarks.

Package contents

• Language tensors: 1,653
• Vision tensors: 356
• Audio tensors: 0
• Tensor payload: 23,265,085,560 bytes
• Safetensors shards: 6

The tokenizer, processor configuration, chat template, license, and notice are distributed with the checkpoint. Source revision: 52f3f65bc7a02d555763bc923bd1d9094898219d.