7-9B Parameter LLMs — Page 444

ablation-study-run-1

gemma-2-9b-it-flora-v1

gemma-2-9b-it-fedavg-v1

Qwen-2.5-7B-grpo-consistent

qwen3_8b_clipcov_baseline_solver_v1

qwen3_8b_hightemp13_baseline_solver_v2

qwen3_8b_hightemp13_baseline_solver_v3

qwen3_8b_nomath_vdrop75_solver_v5

qwen2.5-7b-instruct-gsm8k-sn-tuned-lr3e-5

Qwen2.5-Coder-LEAK-LEETCODE-7B-Base-10

SFT-rubric-checkpoint-100

dpo3-retest-llama2-7b

sft1-retest-llama2-7b

papertalk-qwen2.5-7b

Qwen2.5-Coder-CONTROL-LEETCODE-7B-Base-7

Qwen2.5-Coder-PROD-MCEVALHARD-7B-Base-1

On-policy-SFT

Qwen2.5-Coder-PROD-MCEVALHARD-7B-Base-2

Qwen2.5-Coder-PROD-MCEVALHARD-7B-Base-4

asfeng_train1_qa_r1_8b_step-3200

swerl_qwen35_9b_long_ctx__42__1778998748_step100

swerl_qwen35_9b_fp32lm_b001_testrun_step140

swerl_qwen35_9b_fp32lm_datamix_step200

swerl_qwen35_9b_fp32lm_dppo_g32_step500

Qwen3-8B-Nemotron-SFT

tmax-qwen35-9b-tmax-sft-full-20260513-65k

swerl_qwen35_9b_fp32lm_dppo_g32_step300

cullm

mimiciv_rare_qwen35_9b_selfimprove

mimiciv_rare_qwen35_9b_sft_long

Qwen-9B-NightShift

spam-classifier-llama3

llama3.1_8b_sft_SPEED-24

llama3.1_8b_sft_SPEED-20-BoS

rubric_gdpo_best

llama31-experiments

web-wmrm-best_wm-warm-start

rubric_sagdpo_best

airgrep-asr-gemma-4-e4b

mimiciv_rare_gemma4_e4b_sft

Qwen2.5-7B-Instruct-es-em-bad-medical-advice-crossentropy

adaptive_reasoning_sagdpo_maxresp_best