|
model:
|
|
arch: llava-1.5
|
|
version: 'v1.5'
|
|
|
|
|
|
cache_dir: None
|
|
vit_model: "openai/clip-vit-large-patch14"
|
|
freeze_vit: True
|
|
|
|
|
|
freeze_backbone: False
|
|
tune_mm_mlp_adapter: False
|
|
freeze_mm_mlp_adapter: False
|
|
|
|
|
|
mm_vision_select_layer: -2
|
|
model_max_length: 2048
|
|
|
|
|
|
image_token_len: 576
|
|
mm_use_im_start_end: True
|
|
|
|
|
|
bf16: False
|
|
fp16: True
|
|
|
|
|
|
preprocess:
|
|
vis_processor:
|
|
train:
|
|
name: "clip_image_train_336"
|
|
proc_type: "openai/clip-vit-large-patch14-336"
|
|
eval:
|
|
name: "clip_image_eval_336"
|
|
proc_type: "openai/clip-vit-large-patch14-336"
|
|
text_processor:
|
|
train:
|
|
name: "blip_caption"
|
|
eval:
|
|
name: "blip_caption"
|
|
|