-
Checklist:
- [x] run validation on main
- [x] QAT (generic)
- [x] GNR (IAA, DSA)
- [x] GNR-D (QAT)
- [x] FPGA
- [x] SPR (SGX, QAT, GPU, IAA, DSA)
- [x] Make sure `kube-rbac-proxy…
-
Hello @rybakov ,
Is it possible to get a fully INT8 quantized tflite for STREAM_EXTERNAL_STATE and STREAM_INTERNAL_STATE models?
For example: [ds_tc_resnet](https://github.com/google-research/g…
-
### 🐛 Index out of bounds Error with PerChannel Quantization
Hello,
I have encountered this problem while trying to perform per-channel quantization on weights with ch_axis=1 quantization parame…
-
stty: 标准输入: 对设备不适当的 ioctl 操作
Traceback (most recent call last):
File "/home/chenxin/disk1/github/MQBench_Quantize/main.py", line 21, in
from utils import progress_bar, choose_model, choose_b…
-
LSQQuantizer训练量化几个epoch后，通过strip_qat_ops() trip掉量化图，保存模型。之后resume_qat_graph()也不能继续训练了，这是否正常。
另外，我看mnn导出的模型大小不变。
--modelFile quant-checkpoint-epoch017.onnx --MNNModel quant_model.mnn --framework O…
-
### Description of the bug:
The current implementation of the PT2E creates numerous transpose operation (NCHW -> NHWC) for the weights, which slows down the inference, is there a way to have the …
-
I got some errors below when I use openssl in Nginx with QAT for asynq mode. @mattcaswell , could you give me some suggestions to fix it? Much appreciate.
Nginx: 1.19.3.1
Openssl : 1.1.1n
QA…
-
# Prerequisites
Please answer the following questions for yourself before submitting an issue.
- [N ] I am using the latest TensorFlow Model Garden release and TensorFlow 2.
- [ Y] I am reporti…
-
In the GTC talk, it is mentioned that the fp8 quantized weights are precomputed (cached) only for microbatch case such as pipelining.
However, why not just write out fp8 quantized weights during th…
-
@drmint80 , isiqlandir bizleri zehmet olmasa. Harada hansini istifade etmak yaxsidir? Use case-lerle soylesen ela olardi