search

facebookresearch / av_hubert

A self-supervised learning framework for audio-visual speech
Other
848 stars 136 forks source link

decode issue #43

Open EvAlex01 opened 2 years ago

EvAlex01 commented 2 years ago

hi when i run the decode code " python -B infer_s2s.py --config-dir ./conf/av-finetune --config-name base_noise_pt_noise_ft_433h.yaml dataset.gen_subset=test common_eval.path=/home/dhjj/checkpoints/base_noise_pt_noise_ft_433h.pt common_eval.results_path=/home/dhjj/results/test override.modalities=['audio','video'] common.user_dir=pwd override.data=/home/dhjj/lrs3/433h_data/ override.label_dir=/home/dhjj/lrs3/433h_data/ " i got a issue information as follow " omegaconf.errors.ConfigKeyError: Key 'criterion' not in 'InferConfig' full_key: criterion reference_type=Optional[Dict[Union[str, Enum], Any]] object_type=InferConfig "

but by using the default config code " python -B infer_s2s.py --config-dir ./conf --config-name s2s_decode.yaml dataset.gen_subset=test common_eval.path=/home/dhjj/checkpoints/base_noise_pt_noise_ft_433h.pt common_eval.results_path=/home/dhjj/results/test override.modalities=['audio','video'] common.user_dir=pwd override.data=/home/dhjj/lrs3/433h_data/ override.label_dir=/home/dhjj/lrs3/433h_data/ " i can get the decode result...

chevalierNoir commented 2 years ago

That's expected. The decoding config file is conf/s2s_decode.yaml. The fine-tuning config file should not be used for decoding.