-
### 🐛 Describe the bug
Hi team,
we're now investigating the export to onnx feature and we found that some update logic in the original pytorch model is not working in the converted onnx model. The …
-
do u have tested the text feature from char-CNN-RNN and skip-thought? and which is better for generating images?
-
- [x] Install [warp-transducer](https://github.com/HawkAaron/warp-transducer) in Dockerfile - see https://github.com/ryanleary/mlperf-rnnt-ref/issues/1#issuecomment-559186425.
- [x] Correct `forward…
samgd updated
4 years ago
-
Hi VDelv,
thank you for sharing this implementation publicly.
I wanted to ask about the `Mix` model of the multi frame approach (CNN + LSTM + 1d CONV).
the authors mentioned LSTM layer, yet the imp…
-
Currently both the torch_distributed backend and bigdl backend could only support `output = model(data)` type of applications. For RNN models, we need `output, hidden = model(data, hidden)` in the tra…
-
https://www.etched.com/announcing-etched
-
While our [draft charter](https://www.w3.org/2023/03/proposed-webmachinelearning-charter.html) says that the group:
> priority on building blocks required by well-known model architectures such as re…
-
I have met some problems when implements the RNN-LSTM model, the error is very huge when the part of known trips is short, for instance, only the arrival time of the first stop is known, the predicti…
-
## ❔Question
我初始化模型:
```
model = AutoModel('seq2seq', custom_model_params={
"rnn_type": "lstm",
"bi_direction": False,
"rnn_size": 64,
"dense_siz…
-
Right now the RNN CTC model in `rnn_ctc.py` has the class name `Model` and inherits from `model.Model`. I find this to be confusing and I propose that we rename the `rnn_ctc.Model` class and provide t…