Closed sunsetyerin closed 1 year ago
hi may I know which version you are using and the command that you ran? m6anet inference requires you to pass the directories separated by space, so if you are using the m6anet-run_inference API, it will be something like
m6anet-run_inference --input_dir dir1 dir2 dir3
Note that you just need to pass directory test_1, test_2, and so on instead of test1_1/data.json and test_2/data.json
Hi @chrishendra93 I tried with directory names containing .json files but it cannot recognize multiple input directories. I used docker container "docker://docker.io/yuukiiwa/m6anet:1.0"
hi @sunsetyerin , you are using an outdated version of m6anet that does not have the pooling capability. One way to fix this is to install the latest version from pip / GitHub.
It says in the documentation that m6anet supports pooling replicates. But m6anet keeps giving me this error when I fed in multiple dataprep directories for each sample.
"usage: m6anet-run_inference --input_dir INPUT_DIR --out_dir OUT_DIR [--model_config MODEL_CONFIG] [--model_state_dict MODEL_STATE_DICT] [--batch_size BATCH_SIZE] [--n_processes N_PROCESSES] [--num_iterations NUM_ITERATIONS] [--device DEVICE] m6anet-run_inference: error: unrecognized arguments: results/m6anet/m6anet_dataprep/test_1/data.json results/m6anet/m6anet_dataprep/test_2/data.json results/m6anet/m6anet_dataprep/test_3/data.json"
Can you please elaborate on how I can feed in multiple dataprep results for m6anet inference?