KORE uses knowledge-oriented control as its pivot to synergistically optimize the balance between knowledge adaptation and retention at different stag
Kailin Jiang
kailinjiang
AI & ML interests
continual learning
Recent Activity
liked a dataset 2 days ago
rl-research/dr-tulu-rl-data upvoted a paper 20 days ago
On the Geometry of On-Policy DistillationOrganizations
KORE uses knowledge-oriented control as its pivot to synergistically optimize the balance between knowledge adaptation and retention at different stag
MMEVOKE introduces the first comprehensive benchmark and systematic evaluation framework designed to investigate multimodal evolving knowledge injecti
models 84
kailinjiang/llava_1.5_13b_covariance_matrices_from_onevision_pre_64_seed_rank233_new222
Updated
kailinjiang/llava_1.5_13b_covariance_matrices_from_onevision_pre_64_seed_rank233_new
Updated
kailinjiang/llava_1.5_7b_covariance_matrices_from_onevision_pre_64_seed_rank233
Updated
kailinjiang/llava_OLoRA_KORE_Aug
Updated • 1
kailinjiang/llava_SEFE_KORE_Aug
Updated • 4
kailinjiang/llava_Replay_KORE_Aug
Updated • 3
kailinjiang/llava_MoELoRA_KORE_Aug
Updated • 3
kailinjiang/llava_7B_replay
7B • Updated • 5
kailinjiang/llava_7B_lora
7B • Updated • 4
kailinjiang/llava_7B_full_ft
7B • Updated • 5