Closed mrysztow closed 6 years ago
In #7561, we have discussed PaddlePaddle's 10 aspects in 2018. For models of FLUID, there are NLP,Speech and Image support:
I give some brief information on NLP support. We have a plan to first focus on some state-of-art models in neural machine translation task in NLP field.
For NMT task, three models are expected to be highly optimized:
About the time schedule:
Important operators expected to be highly optimized:
Thank you for pointing particular NMT models. What topologies are the most important for image recognition and speech? Are they Resnet-50 and DS2?
@qingqing01 @kuke Can you help to answer it? Thanks very much!
About computer vision, what we are doing now are as follows:
Image Classification , now mainly SE-ResNeXt model. The mainly related operators are:
Object Detection, now mainly MobileNet-SSD model. The mainly related operators are:
OCR Recognition: now mainly CRNN CTC model, The mainly related operators are:
@qingqing01 thank you for the list Does SE-ResNeXt is going to replace classic Resnet50, already implemented for Fluid (https://github.com/dzhwinter/benchmark/blob/master/fluid/resnet50.py) ?
@mrysztow Both two networks are classic. This two are all needed. They have the same basic operators.
@luotao1 is https://github.com/PaddlePaddle/Paddle/blob/develop/benchmark/fluid/machine_translation.py implementation of Conv seq2seq, mentioned earlier in this thread (https://github.com/PaddlePaddle/Paddle/issues/8450#issuecomment-368200533)? Or it is another seq2seq model?
@mrysztow https://github.com/PaddlePaddle/Paddle/blob/develop/benchmark/fluid/machine_translation.py is seq2seq model. And conv seq2seq is in https://github.com/PaddlePaddle/models/pull/686.
Thank you
Tier 1 | Tier 2 | Tier 3 |
---|---|---|
ResNet50 | MobileNet-SSD | Conv seq2seq(PR) |
Transformer | RNN Search(PR) | CRNN CTC |
SE-ResNeXt | ||
DeepASR |
Feeds support :
The mainly related operators are:
@luotao1 I would like to confirm, that the current priority list is, considering current CPU deployments and feeds model, would be the following:
Tier 1 | Tier 2 | Tier 3 | Tier 4 |
---|---|---|---|
ResNet50 | text_classification | MobileNet-SSD | Conv seq2seq(PR) |
CRNN-CTC | transformer | SE-ResNeXt | |
language_model | DeepASR | ||
chinese_ner | RNN Search |
@mrysztow The priority list is OK now.
Do you have a roadmap of introducing models to FLUID? We would like to align our plan of providing MKL-DNN OP kernels with model roadmap.