nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_24_size_16384_batchtopk_reg_coeff_0_6nzzzsgy
Updated • 35
Nikita Koriagin
nikoryagin
AI & ML interests
None yet
Recent Activity
upvoted a paper 9 days ago
GigaChat Audio: Time-aware Large Audio Language Model upvoted a paper 9 days ago
GigaAM Multilingual: Foundation Model for Underrepresented Languages upvoted a paper about 2 months ago
Trust-Region Behavior Blending for On-Policy DistillationOrganizations
None yet
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_24_size_16384_batchtopk_reg_coeff_0_tswiushb
Updated
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_24_size_16384_batchtopk_reg_coeff_0_cknjcoac
Updated
nikoryagin/sae_google_gemma-2-2b_resid_post_layer_12_size_16384_hierarchical_cm118jpq
Updated
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_ud5xdbcj
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_axcugxsz
Feature Extraction • 0.1B • Updated • 5
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_1dx0ogpo
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_1j2y85cp
Feature Extraction • 0.1B • Updated • 2
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_c5lcbg4k
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_6cn8t21v
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_1o0w4hso
Feature Extraction • 0.1B • Updated • 2
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_c4u5t3tu
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_v1wafn01
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_m8am14lu
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_g6qk0ghc
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_evxcg0i4
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_zrnggyfa
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_iuzowwb3
Feature Extraction • 0.1B • Updated • 3
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_pteralt5
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_8borkoou
Feature Extraction • 0.1B • Updated • 4
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_2l1q94x3
Feature Extraction • 0.1B • Updated • 2
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_cctd40ic
Feature Extraction • 0.1B • Updated • 4
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_2dd8h4hi
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_l631t1t5
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_5meh0wsg
Feature Extraction • 0.1B • Updated • 2
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_q9z3i7mo
Feature Extraction • 0.1B • Updated • 2
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_pc7fsd6c
Feature Extraction • 0.1B • Updated • 4
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_i61t5s8v
Feature Extraction • 0.1B • Updated • 1
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_8lqih7t6
Feature Extraction • 0.1B • Updated • 3
nikoryagin/sae_Qwen_Qwen2.5-7B_resid_post_layer_25_size_16384_batchtopk_5rhwhk04_lora_bn1zqg9l
Feature Extraction • 0.1B • Updated • 1