4-5B Parameter LLMs — Page 96

gemma3-4b-it-lora-loglm

affine-pop-5CDUswY2ZK2nXnkaWhBAWD47CQE3KvMm6AyKhJ1Txm5R5tdi

Affine-5Ey2gdmMeDJ1Z3XGzDKfpYq18jEZ83gqx7pz78pLsGrY6KL5

Affine-5HHUVVn7Ws3bepfj9ZhbE5ffHg1DYxiLwf7c4DPLKSWnTrZj

gemma-3-4b-pretrain-ml-merged

gemma-3-4b-pt-with-it-tokenizer

config_bct_gemma-3-4b-it_ground_truth_lora_merged

seed0_sample30000_mmmlu_google-gemma-3-4b-pt_multi_1.0-1.0_1.0

seed0_mmmlu_google-gemma-3-4b-it_multi_0.1_calm_1e-06

seed0_mmmlu_google-gemma-3-4b-pt_multi_0.1_calm_1e-06

gemma-3-4b-mtaste-16bit

Mimir-Phi-3.5

seed0_sample5000_mmmlu_google-gemma-3-4b-pt_en-es_1.0-1.0_1.0

csrsef-instruct-20260323T195339Z-it01-pubmedqa

8b_rft

Qwen3-4B-Instruct-2507-sft

qwen3-4b-agentbench-merged02

c2

c5

qwen3-4b-agentbench-merged-B

c10

c15

c16

c17

c19

c20

R1_1_4b

R1_2_4b

AT-qwen3-4b-ultrachat-hhrlhf-15360-rm-ppo-clean-p0_05-step-50

F_R1_1_4b

F_R1_1_4b_T2

qwen3-4b-dpo-qwen-cot-_2-3_05_DPO

P2-split2_prob_strlen_cutoff_0p5_filtered_Qwen3-4B-Base_0330

fixed_rl_v3_tmax_combined_agent

rt-sam.backdoor_81_lr1e-5_rho0.01

rt-sam.backdoor_81_lr1e-5_rho0.05

rt-sam.backdoor_81_lr3e-5_rho0.01

rt-sam.backdoor_81_lr3e-5_rho0.05

rt-sam.backdoor_81_lr3e-5_rho0.1

rt-sam.backdoor_9_lr1e-5_rho0.01

rt-sam.backdoor_9_lr3e-5_rho0.05

rt-sam.backdoor_9_lr3e-5_rho0.1