File size: 1,805 Bytes
f7de4ba |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 |
dataset:
align_stage_components:
- download/videollava/valley_llavaimage.json
- download/videollava
dataset_id: videollava
dataset_root_dir: data
finetune_stage_components:
- download/videollava/videochatgpt_llavaimage_tune.json
- download/videollava
type: videollava
dataset_class: FinetuneVideoDataset
hf_token: .hf_token
model:
align_epochs: 1
align_global_batch_size: 256
align_learning_rate: 0.001
align_lr_scheduler_type: linear-warmup+cosine-decay
align_max_grad_norm: 1.0
align_max_steps: null
align_per_device_batch_size: 16
align_train_strategy: fsdp-full-shard
align_warmup_ratio: 0.03
align_weight_decay: 0.0
arch_specifier: 3davg+linear
enable_gradient_checkpointing: true
enable_mixed_precision_training: true
feature_fusion: cross_attention_avg_lq
finetune_epochs: 1
finetune_global_batch_size: 128
finetune_learning_rate: 2.0e-05
finetune_lr_scheduler_type: linear-warmup+cosine-decay
finetune_max_grad_norm: 1.0
finetune_max_steps: null
finetune_per_device_batch_size: 8
finetune_train_strategy: fsdp-full-shard
finetune_warmup_ratio: 0.03
finetune_weight_decay: 0.1
image_resize_strategy: resize-naive
llm_backbone_id: llama2-7b-pure
llm_max_length: 2048
model_id: prism-siglip-controlled+7b
num_frames:
- 16
- 16
- 32
- 16
projector_token_length: 64
reduce_in_full_precision: false
type: merv-full
video_backbone_ids:
- languagebind-video-noclass
- dinov2-video-all-tokens
- vivit-google-b-all-no-cls-16frames
- siglip-vit-b16-224px-all-no-cls
visual_feature_length: 1024
pretrained_checkpoint: runs/final-4e-stage1-lr1e-4/checkpoints/latest-checkpoint.pt
run_id: final-4e-stage2-(lr1e-4)
run_root_dir: runs
seed: 7
slurm_id: '751580'
stage: second_finetune
trackers:
- jsonl
- wandb
|