Hugging Face's logo

esc-bench
/
conformer-rnnt-ami

NeMo
English
esb
Model card Files Community
sanchit-gandhi HF staff commited on
Commit
d84964e
•
1 Parent(s): b49fec5

Correct scripts

Browse files
Files changed (2) hide show
  1. README.md +5 -4
  2. run_ami.sh +1 -1
README.md CHANGED
@@ -2,17 +2,18 @@
2
  language:
3
  - en
4
  tags:
5
- - esc
6
  datasets:
7
- - ami
 
8
  ---
9
- To reproduce this run, execute:
10
  ```python
11
  #!/usr/bin/env bash
12
  CUDA_VISIBLE_DEVICES=0 python run_speech_recognition_rnnt.py \
13
  --config_path="conf/conformer_transducer_bpe_xlarge.yaml" \
14
  --model_name_or_path="stt_en_conformer_transducer_xlarge" \
15
- --dataset_name="esc-benchmark/esc-datasets" \
16
  --tokenizer_path="tokenizer" \
17
  --vocab_size="1024" \
18
  --max_steps="100000" \
 
2
  language:
3
  - en
4
  tags:
5
+ - esb
6
  datasets:
7
+ - esb/datasets
8
+ - edinburghcstr/ami
9
  ---
10
+ To reproduce this run, first install NVIDIA NeMo according to the [official instructions](https://github.com/NVIDIA/NeMo#installation), then execute:
11
  ```python
12
  #!/usr/bin/env bash
13
  CUDA_VISIBLE_DEVICES=0 python run_speech_recognition_rnnt.py \
14
  --config_path="conf/conformer_transducer_bpe_xlarge.yaml" \
15
  --model_name_or_path="stt_en_conformer_transducer_xlarge" \
16
+ --dataset_name="esb/datasets" \
17
  --tokenizer_path="tokenizer" \
18
  --vocab_size="1024" \
19
  --max_steps="100000" \
run_ami.sh CHANGED
@@ -2,7 +2,7 @@
2
  CUDA_VISIBLE_DEVICES=0 python run_speech_recognition_rnnt.py \
3
  --config_path="conf/conformer_transducer_bpe_xlarge.yaml" \
4
  --model_name_or_path="stt_en_conformer_transducer_xlarge" \
5
- --dataset_name="esc-benchmark/esc-datasets" \
6
  --tokenizer_path="tokenizer" \
7
  --vocab_size="1024" \
8
  --max_steps="100000" \
 
2
  CUDA_VISIBLE_DEVICES=0 python run_speech_recognition_rnnt.py \
3
  --config_path="conf/conformer_transducer_bpe_xlarge.yaml" \
4
  --model_name_or_path="stt_en_conformer_transducer_xlarge" \
5
+ --dataset_name="esb/datasets" \
6
  --tokenizer_path="tokenizer" \
7
  --vocab_size="1024" \
8
  --max_steps="100000" \