scripts/wt103/finetune_att6.sh

#LEARNABLE
WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 \
=======
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103


WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 \
=======
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 \
=======
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 \
=======
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 \
=======
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103
WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 \
=======
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 \
=======
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-learnable-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 \
=======
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

#ROPE
WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 \
=======
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 \
=======
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 \
=======
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 3 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 \
=======
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 \
=======
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103

WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 \
=======
    --seed 40 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 2 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103


WANDB_MODE=online CUDA_VISIBLE_DEVICES=6 python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 \
=======
    --seed 0 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103


WANDB_MODE=online CUDA_VISIBLE_DEVICES=6  python src/lgmodeling/finetune.py \
    --model-path /mnt/data/vinhbk/weights/wt103/lr0.00025-rope-step60000-warm2000-size64-layer12-embd192-heads3-shared1-routed0-topk0/best_60000 \
<<<<<<< HEAD
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 \
=======
    --seed 20 --tgt_len 256 --mem_len 256 --eval_tgt_len 256 --n_head 4 --lmc-layer-indices 0 1 2 3 4 5 6 7 8 9 10 11 \
>>>>>>> master
    --learning-rate 0.00025 --batch-size 64 --max_step 60000 --warmup_step 2000 --dataset wt103 \
    --wandb-project LMC-Attention --wandb-group "GPT2-Wikitext103-FFN" --wandb-entity "vinh-bui0512-hcmut"\
    --model-save-dir /mnt/data/vinhbk/weights/wt103/gpt2-finetune --data-path /mnt/data/vinhbk/datasets/wt103