Update results and move rnn directory

egs/librispeech/ASR/RESULTS.md

@@ -1139,17 +1139,18 @@ You can find the tensorboard log at: <https://tensorboard.dev/experiment/D7NQc3x
 
 #### 2021-11-09
 
-The best WER, as of 2021-11-09, for the librispeech test dataset is below
-(using HLG decoding + n-gram LM rescoring + attention decoder rescoring):
+The best WER, as of 2022-20-06, for the librispeech test dataset is below
+(using HLG decoding + n-gram LM rescoring + attention decoder rescoring + rnn lm rescoring):
 
 |     | test-clean | test-other |
 |-----|------------|------------|
-| WER | 2.42       | 5.73       |
+| WER | 2.32       | 5.39       |
 
 Scale values used in n-gram LM rescoring and attention rescoring for the best WERs are:
-| ngram_lm_scale | attention_scale |
-|----------------|-----------------|
-| 2.0            | 2.0             |
+
+| ngram_lm_scale | attention_scale | rnn_lm_scale |
+|----------------|-----------------|--------------|
+| 0.3            | 2.1             | 2.2          |
 
 
 To reproduce the above result, use the following commands for training:
@@ -1170,11 +1171,27 @@ export CUDA_VISIBLE_DEVICES="0,1,2,3"
   --start-epoch 0 \
   --num-epochs 90
 # Note: It trains for 90 epochs, but the best WER is at epoch-77.pt
+
+# Train the RNN-LM
+cd icefall
+export CUDA_VISIBLE_DEVICES="0,1,2,3"
+./rnn_lm/train.py \
+  --exp-dir rnn_lm/exp_2048_3_tied \
+  --start-epoch 0 \
+  --world-size 4 \
+  --num-epochs 30 \
+  --use-fp16 1 \
+  --embedding-dim 2048 \
+  --hidden-dim 2048 \
+  --num-layers 3 \
+  --batch-size 500 \
+  --tie-weights true
 ```
 
 and the following command for decoding
 
 ```
+rnn_dir=$(git rev-parse --show-toplevel)/icefall/rnn_lm
 ./conformer_ctc/decode.py \
   --exp-dir conformer_ctc/exp_500_att0.8 \
   --lang-dir data/lang_bpe_500 \
@@ -1184,8 +1201,15 @@ and the following command for decoding
   --num-paths 1000 \
   --epoch 77 \
   --avg 55 \
-  --method attention-decoder \
-  --nbest-scale 0.5
+  --nbest-scale 0.5 \
+  --rnn-lm-exp-dir ${rnn_dir}/exp_2048_3_tied\
+  --rnn-lm-epoch 29 \
+  --rnn-lm-avg 3 \
+  --rnn-lm-embedding-dim 2048 \
+  --rnn-lm-hidden-dim 2048 \
+  --rnn-lm-num-layers 3 \
+  --rnn-lm-tie-weights true \
+  --method rnn-lm
 ```
 
 You can find the pre-trained model by visiting

egs/librispeech/ASR/conformer_ctc/decode.py

@@ -43,6 +43,7 @@ from icefall.decode import (
 )
 from icefall.env import get_env_info
 from icefall.lexicon import Lexicon
+from icefall.rnn_lm.model import RnnLmModel
 from icefall.utils import (
     AttributeDict,
     get_texts,
@@ -208,19 +209,6 @@ def get_parser():
     return parser
 
 
-def get_rnn_lm_model(params: AttributeDict):
-    from rnn_lm.model import RnnLmModel
-
-    rnn_lm_model = RnnLmModel(
-        vocab_size=params.num_classes,
-        embedding_dim=params.rnn_lm_embedding_dim,
-        hidden_dim=params.rnn_lm_hidden_dim,
-        num_layers=params.rnn_lm_num_layers,
-        tie_weights=params.rnn_lm_tie_weights,
-    )
-    return rnn_lm_model
-
-
 def get_params() -> AttributeDict:
     params = AttributeDict(
         {
@@ -768,7 +756,13 @@ def main():
 
     rnn_lm_model = None
     if params.method == "rnn-lm":
-        rnn_lm_model = get_rnn_lm_model(params)
+        rnn_lm_model = RnnLmModel(
+            vocab_size=params.num_classes,
+            embedding_dim=params.rnn_lm_embedding_dim,
+            hidden_dim=params.rnn_lm_hidden_dim,
+            num_layers=params.rnn_lm_num_layers,
+            tie_weights=params.rnn_lm_tie_weights,
+        )
         if params.rnn_lm_avg == 1:
             load_checkpoint(
                 f"{params.rnn_lm_exp_dir}/epoch-{params.rnn_lm_epoch}.pt",

egs/librispeech/ASR/prepare.sh

@@ -41,9 +41,9 @@ dl_dir=$PWD/download
 # It will generate data/lang_bpe_xxx,
 # data/lang_bpe_yyy if the array contains xxx, yyy
 vocab_sizes=(
-  # 5000
-  # 2000
-  # 1000
+  5000
+  2000
+  1000
   500
 )
 
@@ -277,6 +277,8 @@ if [ $stage -le 10 ] && [ $stop_stage -ge 10 ]; then
   for vocab_size in ${vocab_sizes[@]}; do
     lang_dir=data/lang_bpe_${vocab_size}
     ./local/compile_lg.py --lang-dir $lang_dir
+  done
+fi
 
 if [ $stage -le 11 ] && [ $stop_stage -ge 11 ]; then
   log "Stage 11: Generate LM training data"

icefall/rnn_lm/compute_perplexity.py

@@ -33,8 +33,8 @@ import torch
 from rnn_lm.dataset import get_dataloader
 from rnn_lm.model import RnnLmModel
 
-from icefall.checkpoint import average_checkpoints, load_checkpoint
-from icefall.utils import AttributeDict, setup_logger
+from icefall.checkpoint import load_checkpoint
+from icefall.utils import AttributeDict, load_averaged_model, setup_logger
 
 
 def get_parser():
@@ -165,17 +165,12 @@ def main():
 
     if params.avg == 1:
         load_checkpoint(f"{params.exp_dir}/epoch-{params.epoch}.pt", model)
-        model.to(device)
     else:
-        start = params.epoch - params.avg + 1
-        filenames = []
-        for i in range(start, params.epoch + 1):
-            if start >= 0:
-                filenames.append(f"{params.exp_dir}/epoch-{i}.pt")
-        logging.info(f"averaging {filenames}")
-        model.to(device)
-        model.load_state_dict(average_checkpoints(filenames, device=device))
+        model = load_averaged_model(
+            params.exp_dir, model, params.epoch, params.avg, device
+        )
 
+    model.to(device)
     model.eval()
     num_param = sum([p.numel() for p in model.parameters()])
     num_param_requires_grad = sum(

icefall/rnn_lm/test_model.py

@@ -40,7 +40,7 @@ def test_rnn_lm_model():
     )
     lengths = torch.tensor([4, 3, 2])
     nll_loss = model(x, y, lengths)
-
+    print(nll_loss)
     """
     tensor([[1.1180, 1.3059, 1.2426, 1.7773],
             [1.4231, 1.2783, 1.7321, 0.0000],

icefall/rnn_lm/train.py

@@ -120,13 +120,6 @@ def get_parser():
         default=50,
     )
 
-    parser.add_argument(
-        "--use-ddp-launch",
-        type=str2bool,
-        default=False,
-        help="True if using torch.distributed.launch",
-    )
-
     parser.add_argument(
         "--lm-data",
         type=str,
@@ -165,7 +158,7 @@ def get_parser():
     parser.add_argument(
         "--num-layers",
         type=int,
-        default=4,
+        default=3,
         help="Number of RNN layers the model",
     )
 
@@ -206,7 +199,7 @@ def get_params() -> AttributeDict:
             "batch_idx_train": 0,
             "log_interval": 200,
             "reset_interval": 2000,
-            "valid_interval": 10000,
+            "valid_interval": 5000,
             "env_info": get_env_info(),
         }
     )