4-5B Parameter LLMs — Page 109

sft3

qwen3-4b-sft-merged2

qwen3-4b-math-sft10

qwen3-4b-sft-merged

qwen3-4b-rft-math

qwen3_4b_instruct_grpo_asymclip_ckpt_step220

medical_teacher_grpo_qwen3-4b-no-thinking_kl-0.1

SFT_HALF_A

qwen3-4b-dapo17k-GRPO

Qwen3-4B-Instruct-2507-UserSim-Factored-DPO-Sample

teacher_sciknow_grpo_kl-1_16k

gemma4-e2b-sft

Distill-mv1

4b_2

Qwen3-4B-KPM-SFT

SFT_HALF_A_V8

poker-action_only-qwen35-4b

poker-env_only-qwen35-4b

phi-3.5-mini-instruct-folktaletrained-lora

brasilia

Qwen3-4B-Thinking-awm-async-grpo-200

Qwen3-4B-Thinking-2507-hqq-nf4-faked-bf16

relm-4-e2b-it

safe-rlsd-qwen3-4B

unison4-w-r50

qwen3-4b-grpo-dapo17k-invmax-linear

82f3a9d663f23546961bb9295b75bf42

59c11273cca7574eeda7f6961b909237

qwen3-4b-grpo-sam-reflex

T14b_14bwrong_S4b_klreg1_checkpoint100

T14b_14bwrong_S4b_klreg1_checkpoint50

appworld-qwen35-4b-9b-type2-iter1

Qwen3-4B-OpenR1-A50-HA-GRPO

appworld-qwen35-4b-manysource-2k-newprompt-solvability-junhee-epoch8

Qwen3-4B-DeepWriting-RL

appworld-qwen35-4b-luna-no-mcqa-epoch1-iter1

AT-qwen3-4b-ultrachat-hhrlhf-15360-rm-ppo-clean-p0_05-step-20

AT-qwen3-4b-ultrachat-hhrlhf-15360-rm-ppo-clean-p0_05-step-40

bs16-k10-lr5e-7-ema0.01-eopd0.8-qwen3-4b-think-sciknoweval_bio_bottom20_nogap-maxsteps150

Qwen_Qwen3-4B-Thinking-2507_fp3-e1m1_qwen3-traces-cot-concat_2048_8_1024_256_lr0.03

seed0_sample3000_geomlama_google-gemma-3-4b-it_en-hi_DPO_5e-06

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_qwen3-random-tokens