From a360da1ec75d24f74b9527a72fba5a2eb6aebc10 Mon Sep 17 00:00:00 2001 From: sennnnn Date: Mon, 11 Mar 2024 20:35:01 +0800 Subject: [PATCH] update videogpt. Former-commit-id: 2e211563c684a8b6ef62dfc0c4f1897224dfb58d --- scripts/videogpt/train_videogpt.sh | 10 +++++----- scripts/videogpt/train_videogpt_dsz2.sh | 12 ++++++------ scripts/videogpt/train_videogpt_dsz3.sh | 10 +++++----- 3 files changed, 16 insertions(+), 16 deletions(-) diff --git a/scripts/videogpt/train_videogpt.sh b/scripts/videogpt/train_videogpt.sh index 8780f9f86..a180a5269 100644 --- a/scripts/videogpt/train_videogpt.sh +++ b/scripts/videogpt/train_videogpt.sh @@ -11,18 +11,18 @@ accelerate launch \ --learning_rate 1e-5 \ --weight_decay 0. \ --num_train_epochs 2 \ - --embedding_dim 512 \ --lr_scheduler_type cosine \ --max_grad_norm 1.0 \ --save_strategy steps \ --save_total_limit 5 \ --logging_steps 5 \ - --save_steps 5 \ + --save_steps 1000 \ --n_codes 2048 \ - --n_hiddens 240 \ + --n_hiddens 512 \ + --embedding_dim 4 \ --n_res_layers 4 \ - --downsample "4,4,4" \ - --resolution 336 \ + --downsample "4,8,8" \ + --resolution 128 \ --sequence_length 16 \ --bf16 False \ --fp16 True \ diff --git a/scripts/videogpt/train_videogpt_dsz2.sh b/scripts/videogpt/train_videogpt_dsz2.sh index c27d87d01..8d5bec29a 100644 --- a/scripts/videogpt/train_videogpt_dsz2.sh +++ b/scripts/videogpt/train_videogpt_dsz2.sh @@ -1,4 +1,4 @@ -export ACCELERATE_GRADIENT_ACCUMULATION_STEPS=8 +export ACCELERATE_GRADIENT_ACCUMULATION_STEPS=4 accelerate launch \ --config_file scripts/accelerate_configs/deepspeed_zero2_config.yaml \ @@ -11,18 +11,18 @@ accelerate launch \ --learning_rate 1e-5 \ --weight_decay 0. \ --num_train_epochs 2 \ - --embedding_dim 512 \ --lr_scheduler_type cosine \ --max_grad_norm 1.0 \ --save_strategy steps \ --save_total_limit 5 \ --logging_steps 5 \ - --save_steps 5 \ + --save_steps 1000 \ --n_codes 2048 \ - --n_hiddens 240 \ + --n_hiddens 512 \ + --embedding_dim 4 \ --n_res_layers 4 \ - --downsample "4,4,4" \ - --resolution 336 \ + --downsample "4,8,8" \ + --resolution 128 \ --sequence_length 16 \ --bf16 False \ --fp16 True \ diff --git a/scripts/videogpt/train_videogpt_dsz3.sh b/scripts/videogpt/train_videogpt_dsz3.sh index dd2416463..c49811ce1 100644 --- a/scripts/videogpt/train_videogpt_dsz3.sh +++ b/scripts/videogpt/train_videogpt_dsz3.sh @@ -11,18 +11,18 @@ accelerate launch \ --learning_rate 1e-5 \ --weight_decay 0. \ --num_train_epochs 2 \ - --embedding_dim 512 \ --lr_scheduler_type cosine \ --max_grad_norm 1.0 \ --save_strategy steps \ --save_total_limit 5 \ --logging_steps 5 \ - --save_steps 5 \ + --save_steps 1000 \ --n_codes 2048 \ - --n_hiddens 240 \ + --n_hiddens 512 \ + --embedding_dim 4 \ --n_res_layers 4 \ - --downsample "4,4,4" \ - --resolution 336 \ + --downsample "4,8,8" \ + --resolution 128 \ --sequence_length 16 \ --bf16 False \ --fp16 True \