2023-02-27 20:23:07,473 INFO [export-for-ncnn.py:246] device: cpu
2023-02-27 20:23:07,477 INFO [export-for-ncnn.py:255] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '62e404dd3f3a811d73e424199b3408e309c06e1a', 'k2-git-date': 'Mon Jan 30 10:26:16 2023', 'lhotse-version': '1.12.0.dev+missing.version.file', 'torch-version': '1.10.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.8', 'icefall-git-branch': 'master', 'icefall-git-sha1': '6d7a559-clean', 'icefall-git-date': 'Thu Feb 16 19:47:54 2023', 'icefall-path': '/star-fj/fangjun/open-source/icefall-2', 'k2-path': '/star-fj/fangjun/open-source/k2/k2/python/k2/__init__.py', 'lhotse-path': '/star-fj/fangjun/open-source/lhotse/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-3-1220120619-7695ff496b-s9n4w', 'IP address': '10.177.6.147'}, 'epoch': 99, 'iter': 0, 'avg': 1, 'exp_dir': PosixPath('icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29/exp'), 'bpe_model': './icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29/data/lang_bpe_500/bpe.model', 'context_size': 2, 'use_averaged_model': False, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'blank_id': 0, 'vocab_size': 500}
2023-02-27 20:23:07,477 INFO [export-for-ncnn.py:257] About to create model
2023-02-27 20:23:08,023 INFO [zipformer2.py:419] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
2023-02-27 20:23:08,037 INFO [checkpoint.py:112] Loading checkpoint from icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29/exp/epoch-99.pt
2023-02-27 20:23:08,655 INFO [export-for-ncnn.py:346] encoder parameters: 68944004
2023-02-27 20:23:08,655 INFO [export-for-ncnn.py:347] decoder parameters: 260096
2023-02-27 20:23:08,655 INFO [export-for-ncnn.py:348] joiner parameters: 716276
2023-02-27 20:23:08,656 INFO [export-for-ncnn.py:349] total parameters: 69920376
2023-02-27 20:23:08,656 INFO [export-for-ncnn.py:351] Using torch.jit.trace()
2023-02-27 20:23:08,656 INFO [export-for-ncnn.py:353] Exporting encoder
2023-02-27 20:23:08,656 INFO [export-for-ncnn.py:174] decode_chunk_len: 32
2023-02-27 20:23:08,656 INFO [export-for-ncnn.py:175] T: 39
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1344: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_len.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1348: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_avg.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1352: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_key.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1356: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_val.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1360: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_val2.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1364: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_conv1.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1368: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_conv2.size(0) == self.num_layers, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1373: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert self.left_context_len == cached_key.shape[1], (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1884: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert self.x_size == x.size(0), (self.x_size, x.size(0))
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2442: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_key.shape[0] == self.left_context_len, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2449: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_key.shape[0] == cached_val.shape[0], (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2469: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_key.shape[0] == left_context_len, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2473: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_val.shape[0] == left_context_len, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2483: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert kv_len == k.shape[0], (kv_len, k.shape)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2570: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert list(attn_output.size()) == [bsz * num_heads, seq_len, head_dim // 2]
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2926: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cache.shape == (x.size(0), x.size(1), self.lorder), (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2652: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert x.shape[0] == self.x_size, (x.shape[0], self.x_size)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2653: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert x.shape[2] == self.embed_dim, (x.shape[2], self.embed_dim)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:2666: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert cached_val.shape[0] == self.left_context_len, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1543: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert src.shape[0] == self.in_x_size, (src.shape[0], self.in_x_size)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1637: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert src.shape[0] == self.in_x_size, (
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1643: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert src.shape[2] == self.in_channels, (src.shape[2], self.in_channels)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1571: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if src.shape[0] != self.in_x_size:
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1763: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert src1.shape[:-1] == src2.shape[:-1], (src1.shape, src2.shape)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1779: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert src1.shape[-1] == self.dim1, (src1.shape[-1], self.dim1)
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/zipformer2.py:1780: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert src2.shape[-1] == self.dim2, (src2.shape[-1], self.dim2)
/star-fj/fangjun/py38/lib/python3.8/site-packages/torch/jit/_trace.py:958: TracerWarning: Encountering a list at the output of the tracer might cause the trace to be incorrect, this is only valid if the container structure does not change based on the module's inputs. Consider using a constant container instead (e.g. for `list`, use a `tuple` instead. for `dict`, use a `NamedTuple` instead). If you absolutely need this and know the side effects, pass strict=False to trace() to allow this behavior.
  module._c._create_method_from_trace(
2023-02-27 20:23:19,640 INFO [export-for-ncnn.py:182] Saved to icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29/exp/encoder_jit_trace-pnnx.pt
2023-02-27 20:23:19,646 INFO [export-for-ncnn.py:357] Exporting decoder
/star-fj/fangjun/open-source/icefall-2/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/decoder.py:102: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  assert embedding_out.size(-1) == self.context_size
2023-02-27 20:23:19,686 INFO [export-for-ncnn.py:204] Saved to icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29/exp/decoder_jit_trace-pnnx.pt
2023-02-27 20:23:19,686 INFO [export-for-ncnn.py:361] Exporting joiner
2023-02-27 20:23:19,735 INFO [export-for-ncnn.py:231] Saved to icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29/exp/joiner_jit_trace-pnnx.pt