You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2022/07/28 07:12:02 UTC
[GitHub] [tvm] zhang-yi-chi opened a new pull request, #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
zhang-yi-chi opened a new pull request, #12213:
URL: https://github.com/apache/tvm/pull/12213
Add RNN operator for ONNX frontend. Note that currently onnxruntime have wrong default values for all activations so I removed default value tests for HardSigmoid.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] AndrewZhaoLuo commented on pull request #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
Posted by GitBox <gi...@apache.org>.
AndrewZhaoLuo commented on PR #12213:
URL: https://github.com/apache/tvm/pull/12213#issuecomment-1201450141
I'll take a look today
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] zhang-yi-chi commented on a diff in pull request #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
Posted by GitBox <gi...@apache.org>.
zhang-yi-chi commented on code in PR #12213:
URL: https://github.com/apache/tvm/pull/12213#discussion_r935526678
##########
python/tvm/relay/frontend/onnx.py:
##########
@@ -2756,6 +2757,138 @@ def _activation_needs_beta(cls, activation):
]
return activation.decode("utf-8") in needs_beta
+ @classmethod
+ def bidir_rnn_cell(
+ cls,
+ input_seqs,
+ weight_dicts,
+ acts,
+ ):
+ """
+ Bidirectional RNN cell
+ """
+ seq_len = len(input_seqs)
+ forward_outputs, fw_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[0],
+ act=acts[0],
+ )
+
+ reverse_outputs, rev_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[1],
+ act=acts[1],
+ backwards=True,
+ )
+
+ final_outputs = []
+ for i in range(seq_len):
+ final_outputs.append(
+ _op.stack([forward_outputs[i], reverse_outputs[seq_len - 1 - i]], axis=0)
+ )
+
+ return (
+ _op.stack(final_outputs, axis=0),
+ _op.stack([fw_H_t, rev_H_t], axis=0),
+ )
+
+ @classmethod
+ def _impl_v7(cls, inputs, attr, params):
Review Comment:
Sure. I found that opset 14 has an extra attr `layout` but onnxruntime doesn't support `layout == 1`, so I commented out the test for `layout == 1` for now.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] zhang-yi-chi commented on a diff in pull request #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
Posted by GitBox <gi...@apache.org>.
zhang-yi-chi commented on code in PR #12213:
URL: https://github.com/apache/tvm/pull/12213#discussion_r935526159
##########
tests/python/frontend/onnx/test_forward.py:
##########
@@ -5212,7 +5341,7 @@ def verify_eyelike(indata, dynamic=False):
"test_reduce_sum_keepdims_random",
"test_reduce_sum_negative_axes_keepdims_example",
"test_reduce_sum_negative_axes_keepdims_random",
- "test_rnn_seq_length",
+ "test_rnn_batchwise",
Review Comment:
I found that ONNX doesn't has test_rnn_batchwise node so I simply removed this line in the following commit.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] AndrewZhaoLuo merged pull request #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
Posted by GitBox <gi...@apache.org>.
AndrewZhaoLuo merged PR #12213:
URL: https://github.com/apache/tvm/pull/12213
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] zhang-yi-chi commented on a diff in pull request #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
Posted by GitBox <gi...@apache.org>.
zhang-yi-chi commented on code in PR #12213:
URL: https://github.com/apache/tvm/pull/12213#discussion_r935527341
##########
python/tvm/relay/frontend/onnx.py:
##########
@@ -2756,6 +2757,138 @@ def _activation_needs_beta(cls, activation):
]
return activation.decode("utf-8") in needs_beta
+ @classmethod
+ def bidir_rnn_cell(
+ cls,
+ input_seqs,
+ weight_dicts,
+ acts,
+ ):
+ """
+ Bidirectional RNN cell
+ """
+ seq_len = len(input_seqs)
+ forward_outputs, fw_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[0],
+ act=acts[0],
+ )
+
+ reverse_outputs, rev_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[1],
+ act=acts[1],
+ backwards=True,
+ )
+
+ final_outputs = []
+ for i in range(seq_len):
+ final_outputs.append(
+ _op.stack([forward_outputs[i], reverse_outputs[seq_len - 1 - i]], axis=0)
+ )
+
+ return (
+ _op.stack(final_outputs, axis=0),
+ _op.stack([fw_H_t, rev_H_t], axis=0),
+ )
+
+ @classmethod
+ def _impl_v7(cls, inputs, attr, params):
+ # Unpack inputs, note that if optional and not provided then value will be None.
+ X = inputs[0]
+ Wp = inputs[1]
+ Rp = inputs[2]
+ Bp = inputs[3]
+ # Sequence length currently unused as it can be inferred from shapes.
+ # sequence_lens = inputs['sequence_lens']
+ Hp_0 = inputs[5]
+
+ num_directions = infer_shape(Wp)[0]
+ W_dtype = infer_type(Wp).checked_type.dtype
+
+ if num_directions not in [1, 2]:
+ raise ValueError("num_directions must be either 1 or 2!")
+
+ X_shape = infer_shape(X)
+ hidden_size = infer_shape(Rp)[-1]
+ batch_size = X_shape[1]
+
+ if Hp_0 is None:
+ Hp_0 = _op.zeros((num_directions, batch_size, hidden_size), W_dtype)
+
+ if "activations" in attr:
Review Comment:
I refactored both onnx.py and test_forward.py.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [tvm] AndrewZhaoLuo commented on a diff in pull request #12213: [Relay][Frontend][Onnx] Add RNN operation for ONNX frontend
Posted by GitBox <gi...@apache.org>.
AndrewZhaoLuo commented on code in PR #12213:
URL: https://github.com/apache/tvm/pull/12213#discussion_r934979601
##########
python/tvm/relay/frontend/onnx.py:
##########
@@ -2756,6 +2757,138 @@ def _activation_needs_beta(cls, activation):
]
return activation.decode("utf-8") in needs_beta
+ @classmethod
+ def bidir_rnn_cell(
+ cls,
+ input_seqs,
+ weight_dicts,
+ acts,
+ ):
+ """
+ Bidirectional RNN cell
+ """
+ seq_len = len(input_seqs)
+ forward_outputs, fw_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[0],
+ act=acts[0],
+ )
+
+ reverse_outputs, rev_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[1],
+ act=acts[1],
+ backwards=True,
+ )
+
+ final_outputs = []
+ for i in range(seq_len):
+ final_outputs.append(
+ _op.stack([forward_outputs[i], reverse_outputs[seq_len - 1 - i]], axis=0)
+ )
+
+ return (
+ _op.stack(final_outputs, axis=0),
+ _op.stack([fw_H_t, rev_H_t], axis=0),
+ )
+
+ @classmethod
+ def _impl_v7(cls, inputs, attr, params):
Review Comment:
why target v7 when we have an operator v14?
v7 vs v14 don't seem very different, v14 just cuts out the `hidden_size` attribute which we dont use anyway
Can we change this to _impl_v14?
##########
python/tvm/relay/frontend/onnx.py:
##########
@@ -2756,6 +2757,138 @@ def _activation_needs_beta(cls, activation):
]
return activation.decode("utf-8") in needs_beta
+ @classmethod
+ def bidir_rnn_cell(
+ cls,
+ input_seqs,
+ weight_dicts,
+ acts,
+ ):
+ """
+ Bidirectional RNN cell
+ """
+ seq_len = len(input_seqs)
+ forward_outputs, fw_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[0],
+ act=acts[0],
+ )
+
+ reverse_outputs, rev_H_t = rnn_cell(
+ input_seqs,
+ **weight_dicts[1],
+ act=acts[1],
+ backwards=True,
+ )
+
+ final_outputs = []
+ for i in range(seq_len):
+ final_outputs.append(
+ _op.stack([forward_outputs[i], reverse_outputs[seq_len - 1 - i]], axis=0)
+ )
+
+ return (
+ _op.stack(final_outputs, axis=0),
+ _op.stack([fw_H_t, rev_H_t], axis=0),
+ )
+
+ @classmethod
+ def _impl_v7(cls, inputs, attr, params):
+ # Unpack inputs, note that if optional and not provided then value will be None.
+ X = inputs[0]
+ Wp = inputs[1]
+ Rp = inputs[2]
+ Bp = inputs[3]
+ # Sequence length currently unused as it can be inferred from shapes.
+ # sequence_lens = inputs['sequence_lens']
+ Hp_0 = inputs[5]
+
+ num_directions = infer_shape(Wp)[0]
+ W_dtype = infer_type(Wp).checked_type.dtype
+
+ if num_directions not in [1, 2]:
+ raise ValueError("num_directions must be either 1 or 2!")
+
+ X_shape = infer_shape(X)
+ hidden_size = infer_shape(Rp)[-1]
+ batch_size = X_shape[1]
+
+ if Hp_0 is None:
+ Hp_0 = _op.zeros((num_directions, batch_size, hidden_size), W_dtype)
+
+ if "activations" in attr:
Review Comment:
Can you refactor this as it seems identical to what's in LSTM?
##########
tests/python/frontend/onnx/test_forward.py:
##########
@@ -5212,7 +5341,7 @@ def verify_eyelike(indata, dynamic=False):
"test_reduce_sum_keepdims_random",
"test_reduce_sum_negative_axes_keepdims_example",
"test_reduce_sum_negative_axes_keepdims_random",
- "test_rnn_seq_length",
+ "test_rnn_batchwise",
Review Comment:
Why does this fail?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org