Ctrl+K
- 26_01_16_qwen3-vl-2b_sft_teacher_is_correct_reasoning
- 26_01_18_distill_2b_4b_all_reasoning_80steps
- 26_01_18_distill_2b_4b_teacher_is_correct_revision_120steps
- 26_02_04_distill_2b_4b_reasoning_long_length_all_80steps
- 26_02_15_distill_2b_4b_student_on_policy_all_40steps
- 26_02_15_distill_2b_4b_student_on_policy_all_80steps
- 26_02_15_distill_2b_4b_student_on_policy_forward_80steps
- 26_02_15_distill_2b_4b_student_on_policy_only_distill_2epoch
- 26_02_15_distill_2b_4b_student_on_policy_only_distill_80steps
- 26_02_22_off_policy_distill_2b_8b_correct_2epoch
- 26_02_23_off_policy_distill_2b_4b_correct_2epoch
- 26_03_05_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_01_1_2epoch
- 26_03_11_distill_2b_8b_off_policy_per_4d_attn_uniform_masking_01_1_2epoch
- 26_03_11_distill_2b_8b_off_policy_per_4d_attn_uniform_masking_target_attn_relative_04_2epoch
- 26_03_16_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_consistency_03_05_1_05_2_3epoch
- 26_03_16_distill_2b_8b_off_policy_thinking_4096_2epoch
- 26_03_18_distill_2b_8b_off_policy_per_4d_attn_not_uniform_masking_target_attn_kl_03_05_05_2_thinking_4096_2epoch_a100
- 26_04_06_self_distill_2b_ours_4096_2epoch
- 26_04_14_compodistill_4b_8b_off_policy_4096_2epoch
- 26_04_14_llava_kd_4b_8b_off_policy_4096_2epoch
- 26_04_14_self_distill_4b_ours_4096_2epoch
- 26_04_15_alignti_4b_8b_off_policy_4096_2epoch
- 26_04_15_distill_2b_8b_ours_low_attention_2epoch
- 26_04_15_llava_kd_4b_8b_on_policy_4096_2epoch
- 26_04_17_alignti_4b_8b_on_policy_4096_2epoch
- 26_04_17_compodistill_4b_8b_off_policy_4096_2epoch
- 26_04_17_compodistill_4b_8b_on_policy_4096_2epoch
- 26_04_19_self_distill_8b_fp16_nan_fix
- 26_04_24_compodistill_2b_8b_off_policy_4096_2epoch
- 26_04_29_distill_qwen35_2b_9b_naive_thinking_8192_2epoch
- llm_distill_2b_9b_s1k_on_policy_1epoch
- llm_distill_2b_9b_stratos17k_off_policy_ours_2epoch
- llm_distill_2b_9b_stratos17k_onpolicy_1epoch
- llm_distill_2b_9b_stratos17k_onpolicy_with_ours_2epoch
- qwen3_vl_2b__grpo_thinking__fix_reward__DEBUG_ep2_rb384_gb128
- qwen3_vl_2b__grpo_with_kd__thinking__ep2_rb384_gb128_kdcoef1.0_tau0.3_0.5
- qwen3_vl_2b__grpo_with_vold__thinking__ep2_rb384_gb128_kdcoef1.0e-3
- qwen3_vl_2b__thinking__grpo__papo__fix_reward__REC_ENV_EXP_kl_prcp_0.01__ep2_rb384_gb128
- qwen3_vl_2b__thinking__grpo__papo_with_kd__fix_reward__kl_prcp_0.01__kdcoef1.0_tau0.3_0.5__ep2_rb384_gb128
- qwen3_vl_2b_grpo_masked_kl_stable
- qwen3_vl_2b_grpo_masked_kl_stable_202
- qwen3_vl_2b_grpo_ref_attn_mask
- 12.7 kB