역링크

• review:2023-01_gpt_in_60_lines_of_numpy
• review:2023-12_diloco_distributed_low-communication_training_of_language_models
• review:2023-12_unicron_economizing_self-healing_llm_training_at_scale
• review:2024-01_asynchronous_local-sgd_training_for_language_modeling
• review:2024-01_deepseekmoe_towards_ultimate_expert_specialization_in_mixture-of-experts_language_models
• review:2024-01_mixtral_of_experts
• review:2024-01_tinyllama_an_open-source_small_language_model
• review:2024-07_q-galore_quantized_galore_with_int4_projection_and_layer-adaptive_low-rank_gradients
• review:2024-10_fira_can_we_achieve_full-rank_training_of_llms_under_low-rank_constraint