2-3B Parameter LLMs — Page 47

alif-3b-fp16

manaba_gemma_2_2b

qwen3_1.7b_easy_rl_reinforce_alpha_0.5

qwen3_1.7b_easy_rl_gspo

qwen3_1.7b_easy_rl_final_gamma_1

meta-llama_Llama-3.2-3B-Instruct-GRPO-vanilla_G_4-checkpoint-292

meta-llama_Llama-3.2-3B-Instruct-GRPO-vanilla_G_4-checkpoint-393

meta-llama_Llama-3.2-3B-Instruct-GRPO-vanilla_G_4-checkpoint-186

qwen25-3b-qwq-aug-teacher-1e5

qwen3_1.7b_sft_final_easy_reinforce_ours_adv_fixed_gamma_0.9

qwen3_1.7b_easy_rl_ours_adv_fixed_gamma_1_98_gem_ms_seq_is

16b_SFT

qwen3_1.7b_easy_rl_ours_adv_fixed_geo_ms_only

qwen3_1.7b_easy_rl_ours_adv_fixed_geo_ms_seq_is_epoch3

qwen3_1.7b_easy_rl_ours_adv_fixed_geo_ms_seq_is

qwen3_1.7b_new_standard_A_sft_overfit_lr_5e_6__global_step_96

qwen3_1.7b_new_standard_B_sft_overfit_lr_5e_6__global_step_396

qwen3_1.7b_new_standard_C_sft_overfit_lr_5e_5__global_step_1184

qwen3_1.7b_new_standard_A_sft_overfit_lr_5e_6__global_step_384

qwen3_1.7b_new_standard_C_sft_overfit_lr_5e_6__global_step_1184

qwen3_1.7b_new_standard_C_sft_overfit_lr_5e_6__global_step_592

qwen3_1.7b_sudoku_multi_act_new

4b_RL_DAPO

qwen3_1.7b_sudoku_multi_action_sft_final

agentic-futoshiki-qwen2.5-3B_SAS_SFT

qwen3_1.7b_sudoku_one_action_easy_11_20_epoch2

qwen3_1.7b_sudoku_one_action_easy_11_20_epoch3

qwen3-1.7b-base-adam-3e-6-bs128-kl0.0-global_step_200

qwen3_1.7b_sudoku_multi_action_easy_21_30_epoch2

qwen3_1.7b_sudoku_multi_action_easy_11_20

qwen3_1.7b_new_sudoku_one_action_A_sft_lr_5e_6__step_2248

qwen3_1.7b_new_sudoku_one_action_B_sft_lr_5e_6__step_3324

Affine-at01-12-31-01

Affine-at02-12-31-02

final-d2-1.7b

Llama3B-Cot

nvidia_qwq_aug_1e5

qwen3-1.7b-base-svd-muon-adam-1e-6-bs128-kl0.0-global_step_200

mixed_set1_correct_12k_ep10

qwen3-1.7b-base-adam-1e-6-bs128-kl0.0-global_step_40

qwen3-1.7b-base-svd-muon-adam-2e-6-adamlr-2e-6-bs128-kl0.0-global_step_200

qwen3-1.7b-base-adam-1e-6-bs128-kl0.0-global_step_60