training / flax /conversion_scripts /run_convert_distilled_train_state_to_hf.sh
nullonesix's picture
Saving train state of step 1
a1be16b verified
raw
history blame
290 Bytes
#!/usr/bin/env bash
TCMALLOC_LARGE_ALLOC_REPORT_THRESHOLD=10000000000 python convert_train_state_to_hf.py \
--model_name_or_path "distil-whisper/large-32-2" \
--output_dir "./" \
--resume_from_checkpoint "checkpoint-15000" \
--cache_dir "/home/sanchitgandhi/.cache" \
--use_scan